Making Work Visible: Ethnographically Grounded Case Studies of Work Practice (Learning in Doing: Social, Cognitive and Computational Perspectives)

This page intentionally left blank

Making Work Visible Ethnographically Grounded Case Studies of Work Practice In the 1970s, Xerox pioneered the involvement of social science researchers in technology design and in developing better ways of working. The Xerox legacy is a hybrid methodology that combines an ethnographic interest in direct observation in settings of interest with an ethnomethodological concern to make the study of interactional work an empirical, investigatory matter. This edited volume is an overview of Xerox’s social science tradition. It uses detailed case studies that show how the client engagement was conducted over time and how the findings were consequential for business impact. Case studies in retail, production, office, and home settings cover four topics: practices around documents, the customer front, learning and knowledge-sharing, and competency transfer. The impetus for this book was a 2003 Xerox initiative to transfer knowledge about how to conduct ethnographically grounded work-practice studies to its consultants so that they may generate the kinds of knowledge generated by the researchers themselves. Margaret H. Szymanski is a Senior Research Scientist at the Palo Alto Research Center. She earned her PhD from the University of California, Santa Barbara, and specialized in the study of language, interaction, and social organization. In her work, Szymanski has examined topics such as communication across knowledge boundaries, social engagement at museums around electronic guidebooks, ethnographic training for corporations, and the organization of remote and copresent multiparty conversational interaction. She has published articles in Language in Society, International Journal of Computer Support for Cooperative Work, Linguistics and Education, and Discourse Processes. Jack Whalen is an independent consultant with expertise leading projects that focus on user or customer experience. He is currently working with Luminous Consulting Group in San Francisco and is a visiting professor in the Department of Industrial Design at Aalto University, Helsinki. Previously, Whalen was a Principal Scientist at Xerox’s Palo Alto Research Center and Associate Professor of Sociology and Department Head at the University of Oregon. He is the author of Beyond the Barricades: The Sixties Generation Grows Up (with Richard Flacks); has published articles in Social Psychology Quarterly, British Journal of Sociology, Social Problems, and other journals; and has written chapters for a number of edited volumes, including Workplace Studies: Recovering Work Practice and Informing Systems Design; Organisation, Interaction and Practice: Studies in Ethnomethodology and Conversation Analysis; and The Social and Interactional Dimensions of HumanComputer Interfaces.

LEARNING IN DOING: SOCIAL, COGNITIVE AND COMPUTATIONAL PERSPECTIVES

series editor emeritus John Seely Brown, Palo Alto Research Center

general editors Roy Pea, Professor of Education and the Learning Sciences and Director, Stanford Center for Innovations in Learning, Stanford University Christian Heath, The Management Centre, King’s College, London Lucy A. Suchman, Centre for Science Studies and Department of Sociology, Lancaster University, UK

The Construction Zone: Working for Cognitive Change in School Denis Newman, Peg Griffin, and Michael Cole Situated Learning: Legitimate Peripheral Participation Jean Lave and Etienne Wenger Street Mathematics and School Mathematics Terezinha Nunes, David William Carraher, and Analucia Dias Schliemann Understanding Practice: Perspectives on Activity and Context Seth Chaiklin and Jean Lave, Editors Distributed Cognitions: Psychological and Educational Considerations Gavriel Salomon, Editor The Computer as Medium Peter Bøgh Anderson, Berit Holmqvist, and Jens F. Jensen, Editors Sociocultural Studies of Mind James V. Wertsch, Pablo del Rio, and Amelia Alvarez, Editors Sociocultural Psychology: Theory and Practice of Doing and Knowing Laura Martin, Katherine Nelson, and Ethel Tobach, Editors Mind and Social Practice: Selected Writings of Sylvia Scribner Ethel Tobach et al., Editors The list of books in the series continues after the Index.

Making Work Visible Ethnographically Grounded Case Studies of Work Practice Edited by MARGARET H. SZYMANSKI Palo Alto Research Center

JACK WHALEN Luminous Consulting Group

cambridge university press Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo, Delhi, Tokyo, Mexico City Cambridge University Press 32 Avenue of the Americas, New York, NY 10013-2473, USA www.cambridge.org Information on this title: www.cambridge.org/9780521176651 © Cambridge University Press 2011 This publication is in copyright. Subject to statutory exception and to the provisions of relevant collective licensing agreements, no reproduction of any part may take place without the written permission of Cambridge University Press. First published 2011 Printed in the United States of America A catalog record for this publication is available from the British Library. Library of Congress Cataloging in Publication Data Making work visible: ethnographically grounded case studies of work practice / [edited by] Margaret H. Szymanski, Jack Whalen. p. cm. – (Learning in doing: social, cognitive, and computational perspectives) ISBN 978-0-521-19072-5 (hardback) – ISBN 978-0-521-17665-1 (paperback) 1. Xerox Corporation – Management. 2. Xerox Corporation – Employees. 3. Organizational learning – Case studies. 4. Knowledge management – Case studies. 5. Human engineering – Case studies. 6. Work – Social aspects – Case studies. I. Szymanski, Margaret H. II. Whalen, Jack, 1949– HD9802.3.U64X4763 2011 2011000262 331.250 6–dc22 ISBN 978-0-521-19072-5 Hardback ISBN 978-0-521-17665-1 Paperback Cambridge University Press has no responsibility for the persistence or accuracy of URLs for external or third-party Internet Web sites referred to in this publication and does not guarantee that any content on such Web sites is, or will remain, accurate or appropriate.

Contents

List of Figures and Excerpts List of Tables List of Contributors Series Foreword Foreword by John Seely Brown Acknowledgments Introduction: Work Practice Analysis at Xerox Margaret H. Szymanski and Jack Whalen

page x xiv xv xix xxi xxvii 1

Part I Work Practice Study in Historical Context 1 Work Practice and Technology: A Retrospective Lucy Suchman 2 Engineering Investigations: What Is Made Visible in Making Work Visible? Wes Sharrock and Graham Button

21

34

Part II Applying Work Practice Methods 3 Uncovering the Unremarkable Peter Tolmie

53

4 Work Practices to Understand the Implications of Nascent Technology Francoise Brun-Cottan and Patricia Wall

74

5 Tokyo to Go: Using Field Studies to Inform the Design of a Mobile Leisure Guide for Japanese Youth Diane J. Schiano and Victoria Bellotti

87

vii

viii

Contents Part III Practices around Documents

6 Exploring Documents and the Future of Work Jennifer Watts-Englert, Mary Ann Sprague, Patricia Wall, Catherine McCorkindale, Lisa Purvis, and Gabriele McLaughlin 7 New Ways of Working: The Implications of Work Practice Transitions Mary Ann Sprague, Nathaniel Martin, and Johannes A. Koomen 8 Behind the Scenes: The Business Side of Medical Records Nathaniel Martin and Patricia Wall 9 Seeing the Right Color: Technical and Practical Solutions to the Problem of Accurate Colour Reproduction in the Digital Print Industry Tommaso Colombino, David B. Martin, Jacki O’Neill, Mary Ann Sprague, Jennifer Watts-Englert, Jutta Willamowski, Frederic Roulland, and Antonietta Grasso

109

128 147

160

Part IV The Customer Front 10 Integrated Customer Service: Reinventing a Workscape Jack Whalen and Marilyn Whalen

181

11 Interactions at a Reprographics Store Erik Vinkhuyzen

205

12 Ethnographically Informed Technology for Remote Help-giving Jacki O’Neill, Peter Tolmie, Stefania Castellani, Antonietta Grasso, and Frederic Roulland 13 Sign of the Times at the Department Store: Replacing Paper with Electronic Signs Johannes A. Koomen

225

240

Part V: Learning and Knowledge Sharing 14 Communal Knowledge Sharing: The Eureka Story Jack Whalen and Daniel G. Bobrow 15 Designing Document Solutions for Airline Maintenance Advisories Patricia Wall and Johannes A. Koomen

257

285

Contents

ix

16 Transforming Information System Design: Enabling Users to Design Yutaka Yamauchi

299

17 Rethinking How Projects Are Managed: Meeting Communication across the Organizational Hierarchy Erik Vinkhuyzen and Nozomi Ikeya

312

Part VI: Competency Transfer 18 Fujitsu Learned Ethnography from PARC: Establishing the Social Science Center Koji Kishimoto with a Preface by Jack Whalen 19 The Work Practice Center of Excellence Luke Plurkowski, Margaret H. Szymanski, Patricia Wall, and Johannes A. Koomen 20 Transferring Ethnographic Competence: Personal Reflections on the Past and Future of Work Practice Analysis Brigitte Jordan References Index

327 336

344

359 373

Figures and Excerpts

Figures Transcription Conventions page xxviii Historical Overview xxviii 1 Work practice methodology 12 1.1 Audience in the PARC auditorium, June 15, 1999 (photo courtesy of Andrew Clement) 22 1.2 Reconstruction of technologies as social practice 25 1.3 Advertisement for the 8200 copier (© Xerox Corporation); “Typical users” attempt to operate the 8200 copier (photo by the author) 26 1.4 Working with ID/HI (photo courtesy of Jeanette Blomberg) 28 1.5 From Orr 1996 (photo courtesy of Julian Orr) 29 1.6 Interaction Analysis Laboratory (photo courtesy of Brigitte Jordan) 32 3.1 “All sorts of computing devices will disappear into the background of our everyday lives” (Philips, 2001). Reprinted with permission 55 3.2 Disregarding the alarm going off? 57 3.3 Leaving to walk to school, “That was good timing” 65 4.1 Tracking a file 78 4.2 A map of the workspace provides a foundation to capture and illustrate observations about the spatial and organizational elements required to locate and track a file 79 4.3 Closer examination of the work practices associated with fulfilling a bulletin request reveals a complex picture where Dee troubleshoots a problem with the request and informs those affected how to eliminate this problem in the future 82 x

Figures and Excerpts 4.4 Examples of “what if” opportunities were shared in working sessions with study participants to explore potential technology solutions to current work practice issues 5.1 Relative prevalence of various types of “on the go” leisure-time activities over time and by day of week, for all participants in the MPD study (N = 21) 5.2 Relative prevalence of various types of “on the go” leisure-time activities over time on Friday only, for full-time college students only (N = 3), full-time workers only (N = 4), and part-time workers only (N = 3) in the MPD study 6.1 Two central questions in the study 6.2 Example of a diary page asking participants to list the places where they worked 6.3 Diary entry of one participant’s commute 6.4 Diary page showing participant’s work-related networks 6.5 Finished collage 6.6 Participants constantly checked email while they were mobile 6.7 Virtual collaborative team space: an example of a concept template developed as part of the future of work project deliverables 6.8 Diagram showing how the advisory board was distributed across the company 7.1 Management overview of CTC organization 7.2 Customer documentation workflow 7.3 Publishing organization workflow 7.4 The weekly news bulletin creation workflow 8.1 Formation flow across the clinical enterprise 8.2 Patient Encounter Form (Care Slip) 8.3 Front desk activities are a constant flow of interactions with patients and others in the clinic 8.1 Five minutes at the front desk 8.4 Accounts Receivable activities 9.1 Colour Management choices when opening a file in Photoshop®. Adobe product screenshot reprinted with permission from Adobe Systems Incorporated 9.2 Tone Reproduction Curves can be directly manipulated by the user

xi

84

98

100 110 112 113 113 115 117

122 124 130 133 134 142 148 150 152 153 155

165 170

xii

Figures and Excerpts

9.3 Print Mediator provides comparative soft proofing and guided natural language annotation features in Acrobat®. Adobe product screenshot reprinted with permission from Adobe Systems Incorporated 10.1 Natural monitoring and collaboration 10.2 Phased Interactive Learning in action 11.1 “Half this size” 11.2 The Sparrow page with the video clips 11.3 “You know you put ‘em between the two plastic deals” 12.1 Troubleshooter and customer viewpoints 13.1 Smart paper sign. © Ken Buschner 13.2 Sign cart 13.3 Scanning an item 13.4 Cutting, folding, and displaying sign in hang tag 15.1 Overview of the key steps in the maintenance advisory process 15.2 Workflow in transition from paper to digital 15.3 Aircraft in a heavy maintenance facility, where aircraft go through several weeks of maintenance 15.4 Representation of current and proposed practices for the Read and Sign-off portion of the workflow 16.1 Players in the project 16.2 Design of the sticker 19.1 Creative and Technical Communication’s offerings 20.1 Joint headphone use

174 189 196 212 214 218 236 241 245 247 249 287 292 293

295 300 310 337 350

Excerpts 10.1 10.2 11.1 11.2 11.3 11.4 11.5

INK customer service phone call INK customer service order Half-size pickup Half-size order placement Sometime this afternoon How soon can you get it? Coated (Moore, 2008, p. 395). © Cambridge University Press. Reprinted with the permission of Cambridge University Press 11.6 Good Grief 11.7 Do-it-yourself

181 198 209 210 215 215

217 219 220

Figures and Excerpts

xiii

12.1 17.1 17.2 17.3

232 316 317 317

Troubleshooter’s remote help-giving Project Manager reports on action items Senior Manager scrutinizes project’s progress Senior Manager questions payment negotiations

Tables

5.1 Number of participants desiring various types of additional information to support a leisure activity in the SAS study (N = 294) 7.1 Ethnographic findings and their impact 10.1 Seven-step peer coaching model 16.1 Examples of juxtaposition 17.1 Excerpt from the List of Action Items 18.1 Difference in stances between PARC ethnographers and Fujitsu SEs 19.1 Levels of work practice competence

xiv

page 95 143 194 307 315 330 339

Contributors

Victoria Bellotti, Ph.D., User Experience Research, Palo Alto Research Center, CA, USA. Daniel G. Bobrow, Ph.D., Artificial Intelligence, Palo Alto Research Center, CA, USA. John Seely Brown, Ph.D., Deloitte Center for the Edge, San Francisco, CA, USA. Francoise Brun-Cottan, Ph.D., Anthropology, Conversation Analysis, Video Ethnography. Consultant. Los Angeles, CA, USA. Graham Button, Ph.D., Sheffield Hallam University, Sheffield, UK. Stefania Castellani, Computer Science, Xerox Research Centre Europe, Grenoble, France. Tommaso Colombino, Ph.D., Xerox Research Centre Europe, Grenoble, France. Antonietta Grasso, Xerox Research Centre Europe, Grenoble, France. Nozomi Ikeya, Ph.D., Ethnomethodology, Palo Alto Research Center, CA, USA. Brigitte Jordan, Ph.D., Consulting Corporate Anthropologist, Palo Alto Research Center, CA, USA. Koji Kishimoto, Systems Engineering, Fujitsu Ltd., Tokyo, Japan. Johannes A. Koomen, Ph.D., Xerox Research Center Webster, NY, USA. David B. Martin, Ph.D., Ethnographer, Xerox Research Centre Europe, Grenoble, France.

xv

xvi

Contributors

Nathaniel Martin, Ph.D., User Centered Design, Xerox Research Center Webster, NY, USA. Catherine McCorkindale, Xerox Research Center Webster, NY, USA. Gabriele McLaughlin, D.Sc., Senior Research Fellow, Institute of Knowledge and Innovation, The George Washington University, Washington, DC, USA. Jacki O’Neill, Ph.D., Ethnography and Design, Xerox Research Centre Europe, Grenoble, France. Luke Plurkowski, Sociology and Interaction Analysis, Palo Alto Research Center, CA, USA. Lisa Purvis, Ph.D., Intelligent Reasoning, Xerox Research Center Webster, NY, USA. Frederic Roulland, Computer Science, Xerox Research Centre Europe, Grenoble, France. Diane J. Schiano, Ph.D., User Experience Research, Diane J. Schiano Consulting, Portola Valley, CA, USA. Wes Sharrock, Ph.D., University of Manchester, Manchester, UK. Mary Ann Sprague, Ethnography, Xerox Research Center Webster, NY, USA. Lucy Suchman, Ph.D., Anthropology, Lancaster University, Lancaster, UK. Margaret H. Szymanski, Ph.D., Language, Interaction and Social Organization, Palo Alto Research Center, CA, USA. Peter Tolmie, Ph.D., Ethnography, University of Nottingham, Nottingham, UK. Erik Vinkhuyzen, Ph.D., Social Psychology, Palo Alto Research Center, CA, USA. Patricia Wall, Psychology, Interaction Design, Ethnography, Xerox Research Center Webster, NY, USA. Jennifer Watts-Englert, Ph.D., Cognitive Engineering, Xerox Research Center Webster, NY, USA. Jack Whalen, Ph.D., Ethnography, Knowledge Management, Industrial Design, Aalto University, Helsinki, Finland; Sustainable Fisheries Partnership, Depok Jakarta, Indonesia.

Contributors

xvii

Marilyn Whalen, Ph.D., Sociology, San Francisco, USA. Jutta Willamowski, Ph.D., Computer Science, Xerox Research Centre Europe, Grenoble, France. Yutaka Yamauchi, Ph.D., Organization and Management Research, Kyoto University, Kyoto, Japan.

Series Foreword

This series for Cambridge University Press is widely known as an international forum for studies of situated learning and cognition. Innovative contributions are being made by anthropology; by cognitive, developmental, and cultural psychology; by computer science; by education; and by social theory. These contributions are providing the basis for new ways of understanding the social, historical, and contextual nature of learning, thinking, and practice that emerges from human activity. The empirical settings of these research inquiries range from the classroom to the workplace, to the high-technology office, and to learning in the streets and in other communities of practice. The situated nature of learning and remembering through activity is a central fact. It may appear obvious that human minds develop in social situations and extend their sphere of activity and communicative competencies. But cognitive theories of knowledge representation and learning alone have not provided sufficient insight into these relationships. This series was born of the conviction that new exciting interdisciplinary syntheses are underway as scholars and practitioners from diverse fields seek to develop theory and empirical investigations adequate for characterizing the complex relations of social and mental life, and for understanding successful learning wherever it occurs. The series invites contributions that advance our understanding of these seminal issues. Roy Pea Christian Heath Lucy Suchman

xix

Foreword

I was delighted to learn that this book was being written covering the pioneering efforts of the social scientists at Xerox, especially given that many of these efforts were undertaken long before ethnographic research had become de rigueur in today’s corporate and design worlds. For me, personally, this journey has been extraordinary. After all, I started out as a hard-core computer scientist and an AI junkie with a strong leaning toward cognitive modeling. To many, I have now become a softie, questioning nearly all of the ontological and epistemological assumptions I had embraced in graduate school. This transformation emerged from my rich interactions with many of the authors in this book – especially Lucy Suchman, Gitti Jordan, Jack Whalen, and Marilyn Whalen – but it also emerged from the frequent encounters PARC had with the challenges that were arising in the Xerox world as copiers became more complex to operate and maintain and as the document became the currency of the modern age. These challenges were often more usefully unpacked by the methodologies and sensibilities discussed in this book than by the more formal tools and sensibilities of traditional computer science, engineering, and operations research. These challenges, when properly framed, were tied to the context of work and were more easily met by leveraging the “invisible” resources in the context both to get the job done and to facilitate collective learning. Take a particular task out of its context and it becomes more difficult, often relying on algorithmic solutions. But look at a task in context and one begins to see all kinds of possible resources to deploy in order to make the task easier or even to make it seem as if there is no task at all. Let me recall a few instances of this which ended up radically changing my own perspective and starting me down the path of the social life of information and learning. These instances are more than just personal stories. They were also part of a broader array of examples that helped propel Xerox down a transformational path much like my own. xxi

xxii

Foreword

The first encounter, one that accelerated my appreciation of the importance of context, happened just after I arrived at PARC. It was when I was asked to meet with a master troubleshooter, who all the technicians in the company looked up to, in order to discuss new ways to expedite troubleshooting by applying AI techniques to build intelligent job performance aids for the technicians. I knew the meeting was not going to go well given the deep suspicion the folks in the trenches had of the suits (which they deemed me to be since I had been sent there from corporate headquarters). Before a minute had passed, the master threw out a challenge – well Mr. PhD., suppose this copier sitting here had an intermittent image quality fault, how would you go about troubleshooting it? Here is what the official procedure in the manual says to do – what would you do? The official procedure was relatively simple: using this carefully constructed image quality test sheet run 1000 copies, sort through the output, find a few bad ones and compare them against a diagnostic sheet. Hmmm, 1000 copies even on this superfast machine would take 20 minutes. I knew this was something of a trick question and I knew he was looking for an answer that was better than the blind, mechanistic official procedure. He could tell I was hesitant to offer a solution so he jumped in with a disgusted look on his face. Well, he said, unless you are looking for a coffee break, surely you wouldn’t follow that procedure. Here is what I do – I walk over to the trashcan sitting here by the copier, tip it upside down, and sort through its contents looking at all copies that have been thrown away. The trashcan is a filter between good copies and bad ones – people keep the good copies and throw the bad ones away. So just go to the trashcan to find the bad copies and then from scanning all bad ones interpret what connects them all. Brilliant, I thought to myself. And yes, I had seen the trashcan sitting there but had failed to see it as a resource for the problem at hand. Tapping resources “ready-at-hand” is easy but only when they really are ready-at-hand and not something invisible or out of bounds to your organizational routines or to your cognitive modeling lens. This experience was soon to be built upon when Julian Orr began his now seminal research on how Xerox’s technicians actually went about their daily work of fixing machines in the field and developed his clever proposal for providing a sociotechnical support system for them. I still remember the day that Julian walked in my office to discuss his findings after several months in the field studying these tech reps. This encounter might be best captured by his opening comment to me that nearly everything I had written about troubleshooting was simply wrong. No, these folks do not do in-depth, differential diagnosis using deep, and logically coherent mental models of

Foreword

xxiii

the machine, but rather they weave together fragments of their own experiences and stories that they and others have constructed around similar faults of this class of machines. And when their troubleshooting efforts were failing to converge on the fault they would often call in a tech rep buddy and together they would knit together a story that eventually made sense of all the data from their tests on the machine. Troubleshooting to them was making sense of the faulty machine through a story construction process, thus reframing troubleshooting as an inherently social activity. From this purchase, building sophisticated AI-based job performance aids made little sense. Julian realized that what was called for was a social technology – a twoway radio (like the early Motorola phones with the push-to-talk button) so each tech rep in a region could easily tap the collective expertise of others in his community. Although framed through very minimal comments, this very thin line of communication conveyed great meaning since the technicians all shared the same practices, having learned and worked together on similar machines. Now, several decades later, we see how this foreshadowed one of the roles now played by Twitter in the corporate landscape. Julian Orr’s fieldwork on tech reps influenced PARC’s efforts on another project that decisively shaped my thinking about social technology – Eureka, a system for sharing technical tips authored by technicians that was first developed by researchers at PARC in the mid-1990s and is now used in the field by many thousands of Xerox technicians worldwide. A chapter in this volume by Jack Whalen and Danny Bobrow tells the full story of Eureka, so here I will simply note that it greatly extended the peer-to-peer communication afforded by the two-way radio, allowing technicians to write up their discoveries about machine problems and the solutions they invented – stories presented as practical lessons from their everyday experience in the field, things that the engineers who designed the machines and developed the repair procedures could never have imagined, no matter how smart their planning – and then share them not only with their local work group but across the country and around the globe. Before any tip could be shared, though, it was first vetted by another technician, someone who was regarded by the tech rep community (not management) as an expert on that machine. No one received a financial reward for submitting a tip, either. They did have their name prominently displayed on any tip they wrote, though. As you can see, the system ran on reputational and social capital alone! As Whalen and Bobrow also explain, the service organization’s management was skeptical about sanctioning what today’s social media proponents call “user-generated content.” It represented a radical challenge to the company’s long-standing and strictly controlled reliance on carefully

xxiv

Foreword

documented repair instructions to ensure quality machine service. But PARC’s researchers went directly to the tech reps themselves, enlisting them as codesigners of the system and fervent advocates for its implementation. This enthusiastic support from the front lines, together with convincing evidence of the system’s value that emerged from PARC-sponsored field tests in France, Canada, and the United States, eventually won the day. And just as the tech reps’ use of the radio foreshadowed Twitter, their practice with Eureka was a forerunner of contemporary wikis and weblogs. At about the same time as Eureka, another important experience of mine with the power of social learning was unfolding, at the Institute for Research on Learning, located a few blocks from PARC and established only a few years before with the strong support of the Xerox Foundation. I served on IRL’s Board of Directors and the Institute also had PARC anthropologist Gitti Jordan sharing her time with them, along with sociologists Jack Whalen and Marilyn Whalen, who had been lured by Gitti from the University of Oregon to work with her on an experimental Xerox call center in Dallas, Texas. Gitti, Jack, and Marilyn all describe what happened out there in Texas in two of the chapters that follow. Suffice it to say that their work with call center employees was remarkably successful. It led to the development of a peer-to-peer learning strategy – an “everyone a teacher, everyone a learner” process that the employees helped design – that eventually became the basis for a new learning strategy for all of Xerox, with the strong support of thenCEO Paul Allaire. What was most impressive, however, was the way this learning strategy enabled the employees to make the jump, as they put it themselves, from task to knowledge workers. The work of Orr, Jordan, and the Whalens thus became foundational for rethinking corporate training, and more generally knowledge capture and sharing across communities. At a personal level, this work opened my eyes to just how social the process of learning is. But it also opened my eyes to how pervasive sense making is in our daily and often ho-hum interactions with the world of machines and people and how the cues we use are often invisible, to ourselves and to others, without many of the tools and methods discussed herein. My purpose in recounting parts of my own personal journey from a hard-core geek to a softie, a believer in the power of the social lens, is to call attention to the back stage for much of the work described in this volume. Just as I had to make this journey with the patient help of so many of my colleagues in this book, each of them has had to find ways to help their management come to see the power of the social life of work. Often this was (or is) not an easy task. And it is far from finished. We must continue to find

Foreword

xxv

ways to get our institutions to appreciate the power and significance of the ideas discussed in this book, especially as we enter the rapidly changing knowledge age where we must replace the pursuit of scalable efficiency with the pursuit of scalable, peer-based learning. John Seely Brown

Acknowledgments

We thank Cambridge University Press’s Learning in Doing Series editors for this opportunity; the guidance of Jeanie Lee and Simina Calin has been constant. I gratefully acknowledge the advisory board for their support on this project: Matthias Gutknecht, Brigitte Jordan, and Paul Luff. Many others have provided direction: John Seely Brown, Francoise Brun-Cottan, Graham Button, Luke Plurkowski, Lucy Suchman, Erik Vinkhuyzen, Patricia Wall, and Marilyn Whalen. We acknowledge the love and support of our families: Rohit, Raphael, and Sabina; Marilyn, Michael, Erika, Conrad, and Trevor. We especially thank Lisa Fahey for her invaluable technical assistance in preparing the manuscript for submission; it would not have happened without her! We deeply appreciate the professionalism and generosity of Gitti Jordan for accompanying us on this journey. We are grateful to the Society for Organizational Learning for permission to reprint the chapter by Whalen and Bobrow (Chapter 14), which is a revised and updated version of a paper first published in Reflections, 4(2) (2002), pp. 47–59; to Multilingual Matters, for permission to use the Introduction epigraph originally published by Harvey Sacks (1987, p. 56) in his chapter, “On the preferences for agreement and contiguity in sequences in conversation,” in J. R. E. Lee and G. Button (Eds.), Talk and Social Organisation.

xxvii

Transcription Conventions [ ] ( ) (( )) : ? , . ! (·) (0.2) bye = "less OKAY

overlapping talk unsure hearing transcriber’s and analyst’s comments lengthened pronunciation final rising intonation listing intonation (e.g., more is expected) final falling intonation exclaiming intonation micropause two tenths of a second pause stressed pronunciation latching of speaker’s utterances cut off (e.g., what ti- what time is it?) increased pitch in relation to surrounding talk increased volume in relation to surrounding talk

1990s

1980s

1970s

Historical Overview 1976

J. Rulifson hires 6 ‘ethnoids’, grad students in Office Research Group; E. Wynn stays to write thesis.

1978

J.S. Brown establishes Cognitive and Instructional Sciences (CIS) area at PARC.

1979

L. Suchman interns with Office Research Group at PARC.

1982

L. Suchman and J. Blomberg conduct studies of Xerox copiers’ use.

1983

Institute for Research and Learning (IRL) founded, Palo Alto, CA. EuroPARC founded, Cambridge, England.

1987

Suchman publishes Plans and Situated Actions.

1989

Suchman forms Work Practice & Technology (WPT) area at PARC.

1990

PARC /WPT collaborates with Industrial Design/Human Interface to explore application of ethnography in product design.

1992

Studies of Technology, Organizations and Work (STOW) at EuroPARC formed by G. Button.

1993

EuroPARC becomes Cambridge Lab of Xerox Research Center Europe (XRCE) in Grenoble, France.

PARC anthropologist F. Brun-Cottan moves to Industrial Design/Human Interface (ID/HI) in Rochester. Advanced Systems Development (ASD) formed by B. Bauer to bridge PARC and Xerox Research Center Webster (XRCW)

2000s

ASD/Work Practice & Codevelopment (WPC) area formed by S. Anderson. Several IDHI people join WPC. 1999

Social science at PARC continues with Knowledge Interaction and Practice area formed by M. Whalen.

2000

WPC becomes KnowledgeWorks area in XRCW managed by P. Wall.

2002

KnowledgeWorks begins to collaborate with Xerox Global Services (XGS) in client engagements.

2003

Work Practice ConsultancyToolkit developed (XRCE, XRCW collaboration).

2004

KnowledgeWorks area renamed Work Practice & Technology in XRCW.

2007

Work Practice Center of Excellence forms in Webster, NY (PARC, XRCW collaboration with XGS)

Cambridge facility closes and the work practice competency is transferred to XRCE; Work Practice Technology (WPT) is formed by G. Button.

Introduction Work Practice Analysis at Xerox Margaret H. Szymanski and Jack Whalen

One commonly tends to avoid making “obvious” observations because it is not obvious what thereafter is to be done with them. . . . Rather, we need to see that with some such mundane recurrences we are picking up things which are so overwhelmingly true that if we are to understand that sector of the world, they are something we will have to come to terms with. And, as it happens, they are a tremendous resource. (Sacks, 1987)

Making work visible1 – discovering and describing how people accomplish their tasks, how work actually gets done – reveals what was previously hidden, albeit in plain view. As work practice analysts, our job is to make unbiased observations despite business goals or technology design requirements. If we do our job well, our insights are obvious in retrospect, but by making those insights visible, they become a resource, and we are able to build on them. More than 30 years ago, Xerox pioneered the involvement of social science researchers from anthropology and sociology in the innovation and design of technology and better ways of working. Today, Xerox supports social science groups at three research centers: the Palo Alto Research Center (PARC), located in the heart of California’s Silicon Valley; Xerox Research Center Webster (XRCW, just outside Rochester in upstate New York; and Xerox Research Centre Europe (XRCE), in Grenoble, France. This volume presents examples of the work of this community of work practice researchers with the goal of showing how the research has been carried out and its constructive impact on the ways people work and the technologies that support that work. 1

We thank Lucy Suchman for allowing us to appropriate this title from an ACM article she published about the implications of “Making Work Visible” (1995).

1

2

Making Work Visible

It all began in the late 1970s when John Seely Brown (JSB) brought a sensibility for social scientific research to the Palo Alto Research Center. He credits this sensibility to his early career at the University of California, Irvine where he held a joint appointment as Professor of Computer Science and Cognitive Social Sciences. Here he taught a class with sociologist Harvey Sacks, cofounder of the field of Conversation Analysis (Sacks, 1992); came in contact with sociologist Harold Garfinkel, responsible for initiating the movement towards studies in ethnomethodolgy (Garfinkel, 1967; see also Heritage, 1984); and socialized with anthropologists Brigitte Jordan (Jordan, 1997; Jordan and Henderson, 1995) and Jean Lave (Lave, 1988; Lave and Wenger, 1991). Before coming to PARC, JSB deepened his conviction that social scientific inquiry is powerful while working at BBN Technologies, where he realized that the challenge is not the building of technologies, but the creation of technologies that fit into the workplace – that “organizations can appropriate and people can understand” (personal communication, August 27, 2009). Recalling his job interview with then Xerox Chief Scientist Jack Goldman, JSB explains how he convinced Goldman to hire him. On Goldman’s desk were two telephones: a rather stylish multi-feature device and a very basic phone. So JSB asked, “Jack, why two phones?” And just as JSB had hoped, Goldman gestured at the fancy device and said, “Oh my God, who the hell can use this phone?!? I have it on my desk because everyone has to have one, but when real work gets done, I’ve got to use a regular phone.” With this simple noticing, the importance of designing technology for usability had been brought to Xerox’s attention. Accordingly, JSB came to PARC in 1978 to start a new research area called Cognitive and Instructional Sciences (CIS). At the time, PARC already had its first “ethnoids”: Jeff Rulifson, Manager of the center’s Office Research Group, had hired six University of California, Berkeley anthropology graduate students as summer interns in 1976; one of those students, Eleanor Wynn, stayed on to write her PhD thesis. Once at PARC, JSB successfully persuaded George Pake, Founding Director of the center, and Bert Sutherland, head of the laboratory in which CIS was based, System Sciences, to allow him to hire broadly. He then appointed technologists that gravitated towards or were at least sympathetic to social science, including Richard Burton, Johan de Kleer, Austin Henderson, David Levy, and Brian Smith. The members of CIS expanded to such a number that after only three years, the area became its own laboratory headed by JSB. In 1979, Lucy Suchman joined PARC’s Office Research Group as a research intern to write her PhD thesis, which would later become her

Introduction

3

influential book, Plans & Situated Actions: The problem of human-machine communication (1987). Five years later Suchman became a permanent member of the research staff, and in 1989, she began her own group, Work Practice and Technology, which anchored ethnographic methodology within the organization (Suchman, this volume). Ten years later, sociologists Marilyn Whalen and Jack Whalen came to PARC to lead more applied work practice studies, contributing to successful socio-technical interventions like Eureka2 (Whalen and Bobrow, this volume), experimental workscapes (Whalen and Whalen, this volume), and eventually organizational transformations (Kishimoto and Whalen, this volume). This was part of a trend across all Xerox labs of work practice studies becoming much more closely tied to design (whether of technology or organizational process – or more likely both, as was the case with Eureka and its progeny). And through these years, Brigitte Jordan was an enduring presence in the PARC work practice community, coming to PARC in 1981 while on sabbatical from Michigan State University, joining the research staff full-time in 1988, and continuing as a consultant today. PARC succeeded in spreading work practice across the organization in the late 1980s. In 1986, JSB cofounded the Institute for Research on Learning (IRL) an interdisciplinary research laboratory tightly coupled with PARC (see Jordan, this volume; Whalen and Whalen, this volume). Also in 1986, EuroPARC was created as an allied laboratory in Cambridge, England with a coordinated research program around human-computer interaction (Moran, 1989). Like PARC, EuroPARC was established to be an interdisciplinary lab that joined computer, cognitive and social science. The work practice efforts of this lab were formalized in 1992, when the then-Director Bob Anderson (1994, 1997), an ethnomethodologist, hired fellow ethnomethodologist Graham Button to establish the Studies of Technology, Organizations and Work area (Sharrock and Button, this volume). Button would later move to Xerox Research Centre Europe in Grenoble, France, when EuroPARC closed because of Xerox’s financial problems in 1993. Peter Tolmie led the group from 2005–2006 before Antonietta Grasso became manager of the Work Practice Technology group at XRCE. The spread of work practice studies to Xerox Research Center Webster was viral. In the early 1980s, Austin Henderson (1986) championed a Trillium 2

It was PARC researcher, Olivier Raiman (Barth, 2000), who discovered the informal knowledge sharing among the service technicians that inspired the development of Eureka.

4

Making Work Visible

User Interface Design Tool, and Jeanette Blomberg became the Trillium Core’s observing anthropologist, watching as a Xerox-wide community grew around this tool (Blomberg and Henderson, 1990). During this time, there were joint projects between PARC and the Industrial Design/Human Interface group (IDHI) in Henrietta, New York to explore the application of ethnography and product design; in 1992, PARC anthropologist Francoise Brun-Cottan joined IDHI (see Brun-Cottan and Wall, this volume). A year later, Bob Bauer founded the Advanced Systems Development laboratory3 (ASD) a group that bridged PARC and XRCW with the goal of creating value from the integration of work practice study insights for technology and solutions development. Within ASD, the Work Practice and Co-Development (WPC) Group was led by anthropologist Susan Anderson, who migrated to Rochester after five years at PARC and involvement with IRL. Several people from IDHI were recruited to Anderson’s group, including BrunCottan, Andrea Mosher, and Patricia Wall; they brought a combined competency in ethnography and design including a representational “tool kit” for describing work practice observations and insights (Brun-Cottan and Wall, 1995; Wall and Mosher, 1994). WPC, now named Work Practice and Technology, has been lead by Wall since 2001. As awareness of our work practice competency grew throughout the organization, Xerox business groups began to solicit work practice researchers’ involvement in client engagements. Wall’s work practice group at XRCW was the go-to group for these business units and the relationships her team built with them led to an initiative to teach some of the skills necessary to carry out work practice studies to Xerox’s consultancy division (Plurkowski et al., this volume). Building on the Work Practice Toolkit (Button et al., 2003) and our experience with external clients, training materials and workshops were developed for Xerox consultants and others, including Fujitsu system engineers (see Kishimoto and Whalen, this volume). As a result of our training efforts, consultants and others new to social science research have been able to generate the kinds of knowledge that had previously only been generated by researchers.

Orientations and Influences These Xerox social scientists helped inspire a very broad program of work practice studies undertaken in other corporate laboratories as well as the 3

The lower case was intentional to distinguish ASD from the other “Laboratories” at PARC whose acronym ended with “L” as in CSL, EML, and so forth.

Introduction

5

academy – a significant body of work that has informed the design and development of advanced technologies as well as new ways of working over several decades. Other collections have documented and carefully scrutinized these developments (see for example Engeström and Middleton, 1998; Luff, Hindmarsh, and Heath, 2000; Luff and Heath, 2000; Dourish, 2001; Brown and Duguid, 2000; Cefkin, 2009; Llewellyn and Hindmarsh, 2010); here we will only sketch out some of the intellectual currents and lines of investigation that both influenced and were influenced by work that began at PARC in the late 1970s. It is important to first recognize the distinctive character of that work, both then and now. Plainly, as social scientists these researchers were committed to understanding the fundamentally socio-cultural organization of human reasoning and action, whether at work or in any other endeavor. This represented a significant development in what was once commonly called “human factors engineering,” a field that had been dominated by psychologists and physiologists and where the relevant “factors” were limited to what could be learned about the minds and bodies of individuals rather than the interactions between people and their shared ways of reasoning about and acting on the world. Moreover, these researchers were equally committed to naturalistic observation of that action – to leaving the highly controlled environment of the laboratory so that what humans did and how they did it could be studied in real-world habitats and settings, under ordinary, everyday conditions. This immersion in the everyday world meant that researchers could actively participate in the ordinary activities of subjects’ lives, as this would afford detailed understanding of the natural organization of such activities and of the competencies required of participants to produce them. This strategy was another historic departure from the methods of the psychologists and other laboratory-based researchers who dominated the human factors work being carried out at that time to support the design of new technologies. This kind of direct, first-hand observation of daily life is of course, the very essence of cultural anthropology and its ethnographic methodology, as well as basic to the field of “community studies” in anthropology’s sister discipline, sociology. It turned out that recordings, particularly video records, were especially useful for such naturalistic studies, for they serve as an important control on the limitations and fallibilities of intuition and recollection. If the recorded data was collected in an appropriate manner, it exposed the researcher to a wide range of natural materials and circumstances, and provided some guarantee that the analytic conclusions would not arise as artifacts of intuitive

6

Making Work Visible

idiosyncrasy, selective attention or recollection. And perhaps most important, the availability of a taped record enabled repeated and close examination of the events in question and hence greatly enhanced the range and precision of the observations that could be made (Heritage and Atkinson, 1984). For these reasons, audio-video data, beginning with Suchman’s now famous recording of two renowned PARC computer scientists struggling mightily (and futilely) to make sense of the instructions for a purportedly user-friendly copier, have become absolutely central to work practice studies in Xerox and elsewhere, evidenced by the extensive use of recorded data in this volume. But ethnography of the sort practiced in Xerox has been more than an observational methodology. These studies have been committed from the start to a holistic understanding of work, and thus concerned with far more than simply jobs and tasks; rather, the focus has always been on entire “workscapes” – configurations not only of people and their communal practices (the methodical means they use to organize and accomplish their work) but also the environments where this work gets done and the artifacts and devices that populate these sites and are thus intimately involved in the work’s achievement. The argument has been that these phenomena are all closely related, and need to be analyzed in terms of that interrelatedness, and thus holistically, whenever possible. To be clear, this is not to say that analysis of a specific phenomenon of interest in a workscape, such as the way the workers in a retail setting take up and make use of certain artifacts while interacting with customers over the counter (to use an example drawn from Vinkhuyzen’s chapter in this volume), cannot be undertaken, only that it should not be done in isolation from the other features of that workscape. Further, the analysis of these workscapes is built upon more than holistic naturalism; as the chapters in this collection demonstrate, Xerox social scientists have tended to adopt a very distinct analytic stance, beginning with the principle that any social organization or communal gathering, however mundane or exotic, simple or complex, is a local and thus thoroughly endogenous production. It is not a case, then, of “one great blooming, buzzing confusion” (to borrow from William James), a chaotic mess of action and sound bombarding the ethnographer and obfuscating their vision, which then requires a theory to create – indeed, to conceptually stipulate – any recognizable orderliness; instead, Xerox social scientists understood that they confronted a world whose orderly features and properties has been produced by their fellow humans, by the participants in that world and its diverse activities, and that their task as researchers was to then explain, in fine detail, just how this methodically endogenous organization was accomplished. And while this stance is certainly fundamental to the

Introduction

7

ethnomethodological tradition (as is explained with eminent clarity in Button and Sharrock’s chapter in this volume), it also has deep roots in cultural anthropology’s concern with first describing the “native” understanding of any behavior or belief – how it is meaningful to members, to persons inside the culture – as a prerequisite to developing any scientifically valid account by a nonmember (see especially Headland, Pike, and Harris, 1990). This distinct analytic stance played a crucial role in the design-oriented work Xerox social scientists carried out in collaboration with the computer scientists, engineers, computational linguists, mathematicians, and physicists at PARC and the other centers. From this view, design begins from the presumption that any truly useful technology must support and enhance native human capacities and practices. And in making this argument, these social scientists were in fact building on the intellectual breakthrough that had been achieved in computer design at PARC in the mid-1970s. At that time most of what we now regard as standard, essential features of the personal computer – things that make the computer a device that can be used by ordinary people, not just engineers or “techies,” like the graphic user interface and the mouse – were ingeniously brought together in the development of the Xerox Star, which was based on PARC’s Alto computer ( Johnson, Roberts, Verplank, Smith, Irby, Beard, and Mackey, 1989). The Star – and its research forerunner, the Alto – was a machine explicitly designed to capitalize on native human skills and abilities. The user interface was built around the remarkable visual capacities of humans; that is, the deeply visual ways in which humans perceive, represent, and interact with objects in the world. Additionally, the Alto and Star made use of pictorial representations whose form straightforwardly suggested their meaning (icons), in large part because the images were of familiar office and desktop objects: folders, documents, a trashcan, and so forth. The design of this “graphic user interface” thus took ordinary work practice into account; not only the visual capacities of humans but also the ways many of the objects essential for their work could be visually represented, by employing what came to be called the desktop metaphor (the Star relied on icons even more than did the Alto, in an attempt to further simplify the interface). And the mouse was designed to serve as an extension of the body, of the hand, in order to leverage the human predilection for pointing and thus couple the body with the device in a more natural manner than was possible with a keyboard. As several of the Star’s designers once summarized their intentions, “an important design goal was to make the ‘computer’ as invisible to users as possible” (Johnson et al., 1989, p. 12).

8

Making Work Visible

The design of the Star also took into account the common and highly functional human practice of working in concert with others to accomplish shared goals (certainly a natural way for people to work). It was not conceived primarily as a stand-alone device, but rather as a tool for cooperation and collaboration in offices and other workplaces. The Xerox corporate strategy at the time centered on building devices that would support the “architecture for information” in the “office of the future.” A number of researchers at that time, at PARC and elsewhere, recognized that trafficking in information is an essentially social activity, and that such an “architecture” required computer technology that would allow individuals to collaboratively manage and share their information. If the Star were to effectively support this need, it would require a means of linking many computers and peripherals – like printers and mass storage devices – together, and transferring or sharing data between them at high speeds (the Ethernet communications protocol, also invented at PARC but prior to the development of the Alto, served this purpose quite well). Plainly, then, while there were no social science researchers at PARC at the time of the Star’s development, the work of the center’s engineers and computer scientists unquestionably drew upon a “human centered” philosophy of design. They strongly believed in the notion of “eat your own dog food,” that is, becoming users of everything they were designing. And not just experimental users – people who might try out this or that for a short time, and give some feedback – but rather full time users who had to rely on the system to do their work, and thus were forced to confront all its problems and explore all its possibilities. This incipient human-centered approach of PARC technologists was then significantly advanced through the disciplined, empirically driven study of naturally occurring behavior by the anthropologists, sociologists, and like-minded field researchers who joined the research staff in the following years. For what better way is there to discover human capacities for reasoning and action, and the systematic manner in which people endogenously and concertedly organize their actions, than to closely observe and record their everyday behavior as it takes place in their natural environments? At the same time, research undertaken from this perspective has also made a prominent contribution to design through studies of people using already existing technologies, investigating the place and significance of this technology in the everyday conduct of human affairs – what we might best term “the social life of a technology.” For these investigations, as the research reported in this volume demonstrates, the key problem to address is not so much whether technology is in some fundamental way “social” but rather to show precisely how it is social; accordingly, studies of the social life

Introduction

9

of technology must obviously consider “not only the material objects but the collage of activities involved in making technology into an instrument which is incorporated into a weave of working tasks” (Shapiro, Hughes, Harper, Ackroyd, and Soothill, 1991, p. 3). The impact of this style of research can be seen in its adoption by other corporate laboratories, most conspicuously IBM, Intel, Microsoft, and Yahoo!, all of whom have produced important results in social scienceinspired IT research and development (for the most recent examples see Cefkin, 2009; Bell and Dourish, forthcoming; Harper, forthcoming; Churchill, 2008). Indeed, a number of researchers who worked and trained at PARC or other Xerox research centers went on to staff these and other corporate labs. Interestingly, the work of private consulting firms that specialize in or devote significant resources to ethnographic research – such as the Doblin Group, Ideo, and Ethnographic Insight – can also be traced back to the movement initiated at Xerox PARC (and some of the ethnographers who worked there). And researchers in the academy have been equally influenced by the work practice studies initiated at PARC. Particularly important examples here are the Work, Interaction and Technology Research Centre, led by Christian Heath at King’s College London (for a recent example of their work see Heath, Hindmarsh and Luff, 2010) and the Work, Technology and Organizations group, headed by Steve Barley at Stanford’s School of Engineering (see, e.g., Barley and Kunda, 2001; Barley and Kunda, 2004). Other university-based scholars that have collaborated with PARC and XRCE researchers or spent significant time in these labs include anthropologists Chuck Goodwin and Candy Goodwin of UCLA; sociologist Harold Garfinkel of UCLA, the founder of the ethnomethodological movement; sociologist Paul Drew of the University of York; sociologists Geoff Raymond and Don Zimmerman of University of California, Santa Barbara; sociolinguists Ron Scollon and Suzie Wong Scollon; and of course the authors of several chapters in this collection. Additionally, the line of research first taken up in Xerox has shaped the emergence of critical intellectual currents, particularly what has come to be known as “computer supported cooperative work” (see for example Hughes, Randall and Shapiro, 1992; Grudin, 1994; Ackerman, 2000), where the spirit of Scandinavian social democracy and management-worker consultation (as well as an aversion to the American-led trend of using computers to replace workers and their native intelligence) seems to have been joined with technologists’ continued fascination with groupware. Ideas about “human centered design” and a more socio-cultural understanding of the human dimension in the field of human computer interaction (HCI) have similarly

10

Making Work Visible

been shaped by Xerox’s social science tradition. Finally, sympathetic but parallel – although occasionally intersecting – bodies of work by researchers concerned with “distributed cognition” (see especially Hutchins, 1995) and “situated cognition” (Clancey, 1997), both of which represent a sharp break with the individualistic bias of conventional cognitive science and its dependence on a “mind as computer” model of human reasoning, owe more than a little to the ethnomethodologically-informed analytic stance of Xerox’s work practice studies. All in all, then, what began at Xerox PARC almost four decades ago continues to resonate with contemporary research and developments in various disciplines including organizational analysis, studies of work and organizations, technology, and system design. There have been interesting changes in Xerox’s social science research over these years, however. The most important of these took place in 2001 when, in a cost-cutting move by Xerox, PARC was spun-out to become a wholly owned-subsidiary of the company: PARC, Incorporated. This meant that the customer for PARC’s research was no longer simply Xerox and its focus on the reprographic technology market, which was almost exclusively made up of businesses, of other corporations or small firms. This change naturally greatly expanded PARC research opportunities. While there had been PARC fieldwork projects for clients other than Xerox (such as the project with Fujitsu, recounted in the chapters by Kishimoto and Whalen; Yamauchi; and Vinkhuyzen and Ikeya), those projects were nevertheless oriented to workplaces, either because the client’s customers were other businesses (as was the case for Fujitsu system engineering) or the client was interested in redesigning or innovating in their own workplace (as was the case for Eurekainspired projects with General Motors and Telecom Italia). But the move to PARC, Inc. opened up possibilities for research not only on the workscapes (including their customers) found in other types of businesses and markets, but areas of social life that were not work-related, particularly studies devoted to the design of consumer products (detailed in the chapter by Schiano and Bellotti). Accordingly, the ethnographers and other field researchers at PARC now orient their studies to “client services” quite broadly.

Organization of the Book4 Throughout this volume there are numerous chapters that describe projects that Xerox work practice analysts have undertaken in collaboration with 4

We thank and acknowledge several contributors to the writing of this section: Patricia Wall, Robert S. Bauer, Gabriele McLaughlin, Daniel G. Bobrow, and Graham Button.

Introduction

11

Xerox Business Groups. This is testimony to the business value that work practice analysis has for both Xerox and its clients. Many of these collaborations involved work practice researchers working along side Xerox consultants in a particular client engagement. However, the success of these undertakings for the Business Groups concerned resulted in a demand from the business community that stretched the resources of the research community. In order to rectify this situation Xerox once again pushed the boundaries of research. Having taken what was mainly an academic endeavour, and bringing it to bear on customer problems, the company now sought to systematically transfer the methodology used by its work practice research community to its business community. This enabled Xerox consultants to, themselves, furnish their clients with solutions generated, in part, through the work practice analysis, and to differentiate their offerings from those of their competitors. To this end, in 2000, Xerox Global Systems contracted work practice researchers at XRCE and XRCW to develop the Work Practice Consultancy Tool Kit (Button et al., 2003). The objective of the Tool Kit was not to replace Xerox researchers with consultants, but provide consultants with some degree of competency in work practice analysis, that would support a client engagement with a work practice study that would otherwise be missing. As part of the Tool Kit, case studies were provided to illustrate the kinds of findings work practice studies generate and to help readers make sense of the work. So too, this volume was conceived of as a resource for consultants and students of work practice to better understand what ethnographically grounded analysis of work practice is. But a book entirely of case studies would be remiss in not capturing the historical and methodological influences that contributed to the growth of an organizational competency. Accordingly this book is organized into two parts. The first part, Sections 1 and 2, overview the development of work practice study and illustrate the richness of its methods as applied to a variety of settings. The remaining four sections contain case studies that are thematically organized to give the reader a glimpse into the range of work and the areas in which sustained investigations across time and the research centers that have been carried out over the past twenty years. The case studies contained here do not take a traditional academic research approach to recounting their study and its findings. Instead, authors were encouraged to reflect on their experiences and highlight their decision-making processes throughout the life of the project. It is the researchers’ practical knowledge that we aimed to capture, so that readers

12

Making Work Visible

Figure 1 Work practice methodology

could learn how to better adopt the work practice tools and methods in their own practice. Section 1, Work Practice Study in Historical Context, highlights the research and methodological foundations of work practice study as it was developed at PARC and migrated throughout the organization from the late 1980s to the early 2000s. In Chapter 1, Suchman’s Work Practice and Technology: A retrospective summarizes the research foundations and contributions of her group at Xerox PARC from 1989–1999. Suchman provides a historical view on the research fields that influenced and have been influenced by work practice study including ethnomethodology, human/ computer interaction, and computer-supported cooperative work. In Chapter 2, Sharrock and Button’s Engineering Investigations: What is made visible in making work visible? discusses the foundations of work practice for EuroPARC, and subsequently Xerox Research Centre Europe. They refer to their studies of software engineers to make visible how ethnomethodological studies of work are distinctive from other studies of work, and in so doing, bring out what has been particularly distinctive about Xerox’s work practice studies. Section 2, Applying Work Practice Methods, focuses on the kinds of findings that ethnographic methods and field work techniques deliver in a range of contexts. The five main techniques that comprise the heart of our work practice methodology are shown in Figure 1 above. This figure has been a stable reference used within the organization and with external clients to describe the fundamental techniques of a work practice study.

Introduction

13

Observation and the recording of work activities in their natural settings are the most important data collection tools. Open-ended interviewing helps clarify these observations and can shorten the time needed to reveal deep insights about an organization’s practices. Data representation and data analysis are coordinated to analyze patterns of action; this phase often includes group data sessions with coviewing of recordings (see Jordan, this volume). Once the findings have been generated, it is helpful to take them back to the participants of the study or key stakeholders in order to codesign next studies, new processes, and technology prototypes. Our work practice method is iterative in nature whereby data is collected and analyzed and more targeted data is collected and analyzed until it is deemed that a next iteration will not reveal significant insights. Actual study designs vary, based on the goals of the project and specific client needs; this variability is shown in the three chapters contained in Section 2. In Chapter 3, Tolmie’s Uncovering the Unremarkable brings the power of in-context observation to the forefront. Through a careful unfolding of ethnographic data, Tolmie shows how ethnographers can manage the problem of at the same time 1) acquiring membership, in order to recognize what is to be taken as unremarkable about a certain context, and 2) maintaining anthropological strangeness, in order to notice those taken-for-granted aspects. Tolmie also emphasizes the importance of teaming designers and ethnographers in order to facilitate the transition of field experiences and analyses into insights for design. Brun-Cottan and Wall’s Chapter 4, Work Practices to Understand the Implications of Nascent Technology, take us through the life of a client project in which the methods of work practice analysis, especially visualizations of work practice findings, engendered trust and enabled the client organization to envision new possibilities and their potential impacts. With these visual descriptions and tools, study participants could actively participate in and codesign the implementation of a technology or solution; and computer scientists, engineers, and others technologists could embody new ideas and concepts in working prototypes, making it possible for users to test drive new ideas before committing to their development. In Chapter 5, Tokyo to Go: Using Field Studies to Inform the Design of a Mobile Leisure Guide for Japanese Youth, Schiano and Bellotti illustrate the blending of qualitative and quantitative data to shape field studies and develop requirements for technology design. They show multiple ways to conduct fieldwork to capture participants’ leisure activities across time and space: from in-depth interviews, to online surveys, to street activity sampling, to mobile phone diaries. By blending quantitative with

14

Making Work Visible

qualitative methods, the converging results provide robust information to design a recommender system. Section 3 focuses on the Practices around Documents, a key area for Xerox, marketed at one time as The Document Company. This theme examines the social dimension of document creation as well as the communicative import of document use. In Chapter 6, Exploring Documents and the Future of Work, Watts-Englert, Sprague, Wall, McCorkindale, Purvis, and McLaughlin share how they explored the practices around documents for home-based and mobile workers. Because they wanted their findings to have impact on Xerox’s future business strategy, they formalized the participation of key stakeholders in an advisory board, that not only served their communication needs, but also increased the study’s company-wide visibility. In Chapter 7, New Ways of Working: The Implications of Work Practice Transitions, Sprague, Martin, and Koomen discuss how their unique, longterm relationship with an internal business organization developed over time. Whereas they were first asked to help them think about how shortrun publishing could be managed, the power of their insights led the client to ask for help with their own processes as they transitioned to a new technology. This organization was so committed to work practice that it later commissioned the establishment of a Center of Excellence in order to have the work practice capability in-house (Plurkowski et al., this volume). In Chapter 8, Behind the Scenes: The Business Side of Medical Records, Martin and Wall examine the practices of two different clinical functions: the front desk personnel and their accounts payable department. Their description highlights the disparities between prescribed workflow representations and the complexities of the work as it is achieved. In addition, they show how people compensate for inadequacies in their work tools; by uncovering these workarounds, they could be incorporated into the design of new, more supportive technologies. In Chapter 9, Seeing the Right Colour: Technical and Practical Solutions to the Problem of Accurate Colour Reproduction in the Digital Print Industry, Colombino et al. produce an analysis that renders a technical problem – the translation and reproduction of colour “consonantly” across devices and formats – a socio-technical problem. Their analysis in effect closes the gap between the abstract, mathematical manipulations of color management technology and the practical, visual and tangible manipulations that practitioners perform when managing color. Section 4 focuses on The Customer Front, the practices that occur at the interface between employees and customers. These studies highlight both the enormous opportunity for value creation and the pitfalls that can occur

Introduction

15

when time, money, and effort are invested in studies that retroactively try to make the right business connection. In Chapter 10, Integrated Customer Service: Re-inventing a Workscape, Whalen and Whalen address the challenge of cross-training employees in three customer service functions by designing a learning-friendly workscape and capitalizing on the employees’ natural learning practices. Their learning strategy, Phased Interactive Learning, takes the learning process out of the classroom to the front lines where the service interaction takes place with the customer, effectively reducing learning costs, accelerating the time to competency, and increasing service performance outcomes. Vinkhuyzen keeps our focus on the many interactional issues that can arise at the customer front in Chapter 11, Interactions at a Reprographics Store. This study shows the complex interdependence between the store’s physical environment, its paper order taking form, and employee-customer interactions around activities such as initiating an order, discussing money, and asking for help. Vinkhuyzen also discusses the maturation of the study as it moves from observation, to targeted data collection, to participant observation and finally to the codesign of a training solution. Whereas Vinkhuyzen recounts a multi-year study, in Chapter 12, Ethnography-inspired Technology for Remote Help-giving, O’Neill, Tolmie, Castellani, Grasso, and Roulland demonstrate how custom-sizing the ethnographic research into smaller observational chunks can lead to incremental value creation at the point of business need. Here two ethnographers conducted a three-week ethnography of two call centers. This data produced an understanding of the current situation that revealed a number of problems and opportunities that were then put into practice through several rounds of iterative prototyping and user testing, with design starting with the supposedly simplest interventions and continuing through to the more radical. The need for a business opportunity target and timing match is demonstrated in Koomen’s Chapter 13, Sign of the Times at the Department Store: Replacing paper with electronic signs. In this case, the business target, adoption of an electronic signage system, did not yield a significant business opportunity for work practice transformation in the short term. Nevertheless, it showcases a technology-inspired process model that will most likely be the norm five to ten years from now, as the retail technology infrastructure catches up. Section 5, Learning and Knowledge Sharing, illustrates how work practice studies can lead to insights about the organizational barriers that keep its employees’ knowledge from flowing. Further, this set of studies shows how the design of new work processes built on existing practices and how the

16

Making Work Visible

technology that supports these processes can enable learning across boundaries. In Chapter 14, Whalen and Bobrow’s Communal Knowledge Sharing, the technicians themselves identified unusual calls – those with a problem not contained in the repair manual – as the lynchpin of shared knowledge. They report how the practice of sharing validated tips came from a codesign process with the technicians, taking into account company concerns about safety and validity (and Eureka then saved the company many millions of dollars, year after year). Wall and Koomen look at how information flows from airline headquarters to aircraft maintenance sites to aircraft mechanics in Chapter 15, Designing Document Solutions for Airline Maintenance Advisories. The study focuses on a mixed paper and computer based system for managing airline advisories. Their work led to design of an IT system that relieved supervisors of several hours a month of status checking and streamlined the acknowledgement process for mechanics. In Chapter 16, Transforming Information System Design: Enabling users to design, Yamauchi examines how the expectation that engineers are supposed to design systems from the problem specifications of the users/customers impacts the system design process. He reports how their implementation helped move engineers into a mindset that nurtures users to take responsibility and participate in part of the design of their future system. Users also changed as they took part in design; they explored alternatives with the engineers and collaboratively learned new ways of adapting their processes to the future system. In Chapter 17, Rethinking How Projects are Managed: Meeting communication across the organizational hierarchy, Vinkhuyzen and Ikeya show how in one company the roles of senior managers and project managers, while conceived to be complementary, interactionally result in ineffective project management during meetings. With these findings, the authors facilitated changes in the interactional practice of the project meetings that supported effective advice giving, and documents that better reflected what senior managers needed to know. The Chapters in Section 6, Competency Transfer, address the issue of transferring work practice research competency to those outside of the Xerox research community. The most ambitious “competency transfer” initiative was that undertaken in collaboration with Fujitsu. The company wanted to address its high rate of project failure and turned to Xerox work practice researchers to help it support its customers better through equipping its software engineering teams with work practice analytic skills. In Chapter 18, Fujitsu Learned Ethnography from PARC: Establishing the Social Science Center, Kishimoto-san, the Fujistsu project leader, recounts

Introduction

17

the history of the project, the problems that Fujistsu faced, and the subsequent establishment within Fujistsu of the Social Science Center. Plurkowski, Szymanski, Wall, and Koomen, in Chapter 19, The Work Practice Center of Excellence, describe how the idea of the Work Practice Consultancy Tool Kit (Button et al., 2003) was deployed as a practical instructional aid through the establishment of the Work Practice Center of Excellence. Through the Center, work practice analysis was integrated with Xerox’s existing lean Six Sigma methodology to form a hybrid consulting approach. Importantly, and in the tradition of work practice studies, the Center stressed putting the methodology into practice through mentored fieldwork. In Chapter 20, Transferring Ethnographic Competence: Personal reflections on the past and future of work practice analysis, Jordan also provides a personal account of attempts to transfer work practice ethnographic competency into Xerox Business Groups. Drawing on her experiences at IRL and PARC, she reflects on three levels of competency transfer: basic field methods, deep analytic competence and strategic competence. Providing us with a very fitting ending to this volume, she concludes by exploring the future of ethnography both for practitioners and corporations.

Part I

Work Practice Study in Historical Context

1

Work Practice and Technology A Retrospective Lucy Suchman

Preface During the decade from 1989 to 1999, workplace research at PARC was based in the Work Practice and Technology research area. In anticipation of the group’s disbanding in 2000, its members organized a symposium under the heading “Work Practice and Technology: The next twenty years of research.” Over one hundred participants attended from North America, Western Europe, Scandinavia, Japan, and Australia. The symposium consisted of one day of panels at Xerox PARC, followed by two days of working groups and celebrations at Half Moon Bay, California. The text that follows is my opening address for the symposium, aimed at providing context for the gathering as well as a brief, retrospective summary of some of the work practice research accomplished during the preceding twenty years at PARC. While references to publications that provide further detail on the studies mentioned have been added, the text has otherwise been left unedited to give a sense for the spirit of the occasion. One inspiration for us in organizing this gathering was the prospect of bringing our various, partially overlapping networks of friends and colleagues interested in work and technology together in one place. So in discussing how best to welcome you all we decided that we had better begin by introducing you to each other. Rather than have you turn to your neighbor and shake hands, or go around the room and have each of you tell your story (though that would be a fascinating event in its own right!), we decided that I should adopt the Danish practice at a large celebratory gathering, wherein the host explains to their guests who don’t all know each other how they all came to be there. In this case the relevant kinship ties are intellectual ones, so I’ll try to orient you by offering a sketch of the research landscape that we collectively represent. 21

22

Making Work Visible

Figure 1.1 Audience in the PARC auditorium, June 15, 1999 (photo courtesy of Andrew Clement)

This is obviously an assignment fraught with potential for all sorts of shortcomings, so I’ve decided to approach it in a particular way. Rather than make any attempt at comprehensiveness, I’ll try to name some of the research affiliations represented here and give an example for each of the kinds of concerns and insights that have animated and inspired us in our own work over the past two decades. With that as background I’ll then say a few words about how our work here has tried to weave some of these affiliations together, and by then I hope we’ll have some context for the panels and discussions that will occupy us for the remainder of the day. So, to begin with at least some of our beginnings, I’ll take the broad category of investigations that have gone on under the name of social and cultural studies of science and technology (STS). Coming together (as did so many things) in the 1960s, STS has been a wonderfully rich and eclectic home for researchers – particularly anthropologists and sociologists – interested in the sociality of science and technology, and in the materialities of the social world. Among the many achievements of STS has been a radical reexamination of traditional boundaries, particularly in the European West, between the categories of nature/culture, social/technical, and human/artifact. The project has become one that might be thought of as recovering the relations along one dimension, and restoring the excluded middles along the other – for example, by recognizing the deep mutual constitution of humans and artifacts, without losing their particularities and differences.

Work Practice and Technology

23

For many of us here, a particularly compelling agenda for the study of work and technology is provided by ethnomethodology. Ethnomethodology proposes a radically alternate program for social studies that flies in the face of what Harold Garfinkel has named “the worldwide social science movement and its armies of social analysts.” Rather than seeing the specificities of the lived world as in need of analytic reworking – into generalizations, abstractions, formalizations and the like – that then order and explain the concreteness, ethnomethodological studies take as their topic just how, specifically, the work of ordering (including but not limited to its technologies of analysis) gets done. Ethnomethodological studies now make up a rich corpus that respecifies the grounds for the orderliness of everyday activity, including work practice across a range of shop floors and employing various assemblages of associated artifacts. Having in common with ethnomethodology a commitment to the specificities of embodied practice and lived experience, feminist research has been another powerful force in reworking our traditional investment in formal analysis as the foundation of social theorizing. Beginning with the project of denaturalizing and refiguring gender – from an eternal, biological category to specifically situated, historical and cultural productions – feminist research has worked to find alternate bases for responsibility and accountability as well. Rather than requiring some kind of transcendent, omniscient view from outside, responsible action is taken to work through, and only through, engagement with the contingencies and contradictions of our particular, always shifting positionings. Turning to the interests in technology represented here, I’ll start with the field of human/computer interaction. With primary roots in psychology, HCI has been the traditional home for those whose interests in computing center on the peripheral (for academic computer science) questions of just how the encounter between person and computer artifact is to be managed (and by whom). A persistent fault line within HCI has run between those interested in the prospects for the interface as intelligent interlocutor (affiliated with artificial intelligence), and those for whom the computer remains a dynamic artifact to be designed (affiliated with the traditions of graphical and industrial design). More recent critiques call for a refiguring of the relevant field of research in HCI from the interface narrowly defined, to the situations of use and ongoing activities in which our interactions with computer artifacts are embedded; what Liam Bannon aptly summarizes as a move “from human factors to human actors” (1991). One research community in which such a refiguring has taken place is Computer Supported Cooperative Work (CSCW), initiated in the

24

Making Work Visible

mid-1980s out of practical concerns regarding the advent of networked computing and the implications of multiple, distributed computer users for system design. CSCW has since become a forum for those interested in the design of computer-based technologies with explicit concern for the socially organized practices of their users. As a consequence, CSCW has also become a kind of haven for computer scientists and social scientists interested in the possibilities for hybridization of their respective fields. As Schmidt and Bannon observe, “As a research effort that involves a large number of established disciplines, CSCW is an arena of discordant views, incommensurate perspectives, and incompatible agendas” (1992: 11). Nonetheless, and perhaps even because of this diversity, CSCW has been a place where extended studies of complex, multiactivity, technologyintensive worksites have found an eager audience among computer science colleagues as well as, increasingly, a place to report our hybridized collaborations in work-oriented, practice-based system design. Finally, many of us here have come together through our shared interests in developing an alternative to mainstream system development practice that repositions traditional relations of design and use. In place of a model of expert technologist delivering what’s good for him or her to a novice user, initiatives under the heading of participatory design take a radically symmetrical view of the knowledge related to work and technology design. In particular, knowledge of relevant work practices is taken to be at least as critical to the design of new technologies as is technological expertise. This means that technology design needs to take the form of a process of mutual learning around a joint project, located within the site of prospective technology use. This process, moreover, is inevitably a political and interested one. In recognition of that, the Scandinavian roots of participatory design bring as well as an explicit commitment to values of skill enhancement, quality of working life, and industrial democracy. With that brief (and idiosyncratically selective) sampling of who we collectively are as background, let me now turn to sketch something of what’s gone on here over the past 20 years: how we’ve tried to weave these various resources into our own practice. As you all know, it was in the late 1970s that a small cohort of us here at PARC – anthropologists and computer scientists – began developing a joint research program concerned with the ethnographically based study of everyday work practices and with associated projects of design. I’ve recently taken to talking about our agenda as, among other things, the reconstruction of technologies as social practice (Suchman et al., 1999, see Figure 1.2). I like this phrase in part for the ambiguities of the term reconstruction. Let me say a bit more about that.

Work Practice and Technology

25

Figure 1.2 Reconstruction of technologies as social practice

The first sense of reconstruction references the work of documenting or, in ethnomethodology’s happier phrase respecifying, culturally and historically constituted meanings and practices. In this sense the question is how to conduct ethnographies of work and technology, including both practices of design and artifacts-in-use, that are aimed at recovering the projects, identities, and interests that inform those practices. These reconstructions are about making sense of what’s in place. The second sense points to a more reformist agenda, aimed on the one hand at critical analyses of the professional practices and institutional arrangements of contemporary technology production, and on the other at possibilities for the development of alternative approaches. These reconstructions are about remaking what’s in place into something new. Both comprise substantial projects: projects that we take to be intimately and necessarily interrelated. Within that frame, we’re interested more specifically in three increasingly hybrid approaches. First, within the project of making sense of what’s in place, are critical studies of prevailing discourses and practices in the production of new technologies. For example, our early work looked critically at notions of “interactivity” and “reliability” as those terms have been taken up within the fields of product design and technical service work. Second, again as a contribution to making sense of what we have, we need ethnographic inquiries into specific sites of work. Here we take as illustrative the place of stories, the work of design, and the study of technologies-in-use. And finally, in the interest of remaking what we have into something new, we need interventions into processes of work and technology design aimed at the development of alternatives to prevailing practice. For us that’s taken the form of explorations in what I’ll call here as a shorthand participatory, practice-based design. I’ll say a bit more about each of these in turn.

26

Making Work Visible

Figure 1.3 Advertisement for the 8200 copier (©Xerox Corporation); “Typical users” attempt to operate the 8200 copier (photo by the author)

Arriving at PARC in 1979 as a graduate student in anthropology with a strong interest in human interaction, I was intrigued by the ways in which the notion of interaction was being applied to relations between people and machines (see Suchman, 2007). It seemed to me that the challenge of interactive interface design was actually a more subtle and interesting one than it was assumed to be by my colleagues. Basically their assumption was that computational artifacts just are interactive, in roughly the same way that persons are, albeit with some obvious limitations. However ambitious, the problem in this view was a fairly straightforward task of encoding more and more of the cognitive abilities attributed to humans into machines, in order to overcome the latter’s existing limitations. Around this time a project began at PARC to develop an intelligent, interactive expert system that would provide instructions to users in the operation of a particular photocopier, just put on the market and reported by its intended users to be “too complicated.” With Austin Henderson, I initiated a series of studies aimed first at understanding what made the existing machine difficult to use, and later at seeing just what happened when people engaged in “interactions” with my colleagues’ prototype expert advisor. In order to explore these questions in detail we got a machine ourselves and installed it in our workplace. We then invited others of our coworkers, including some extremely eminent computer scientists, to try using the machine to copy their own papers for colleagues, with the understanding that a video camera would be rolling while they did so. This resulted among other things in what has become something of a cult video that I produced for John Seely Brown for a keynote address at CHI in 1983, titled “When User Hits Machine.” This image, taken from a 3/400 reel-to-reel video recording made in 1982, shows two of our colleagues

Work Practice and Technology

27

using the machine to make two-sided copies of a research paper. The CHI audience would recognize Allen Newell, one of the founding fathers of AI. His PARC colleague is a brilliant computational linguist named Ron Kaplan. From looking closely at actual encounters with the machine, I began to develop the idea that its inscrutability was less a function of lack of technological sophistication on the part of its users, than of their lack of familiarity with this particular machine. I argued that the machine’s illegibility was tied not to any esoteric technical characteristics, but to mundane difficulties of sense-making characteristic of any unfamiliar artifact. My point was that reading a new artifact is an inherently problematic activity. I wanted to argue that however improved the machine interface or instruction set might be, this would never obviate the need for learning on the part of prospective users. This called into question, then, the very viability of marketing the machine as self-evidently easy to use. Arriving at PARC around this time, my colleague Jeanette Blomberg was given an assignment that invited a related look at assumptions about machine “reliability” (Blomberg, 1987). We’ll call on another moment from the video “When User Hits Machine,” from the movie “9 to 5,” (Twentieth Century Fox, 1980) and have Jane Fonda set the scene for Jeanette’s work. In this episode Jane, on her first day at work, has been instructed by hyper-efficient office manager Lilly Tomlin in the operation of a very large Xerox copier. Predictably, the moment Lilly leaves the room and Jane presses the big green button, all hell breaks loose. And, at the worst possible moment, she is discovered being incompetent by her brutish boss, who tells her that “any idiot can operate this machine,” and that if she doesn’t get it together her first day might be her last. Product developers and marketers at Xerox in the early 1980s were puzzled by a lack of correlation between the frequency of machine service calls and “customer perceptions” of machine reliability. The expectation had been that as service calls increased, perceptions of reliability would decline, while a machine with few service calls should be seen as highly reliable. Instead it seemed that their users pronounced machines assessed as good performers by the company “lemons,” while others requiring frequent service calls were well-loved. Many explanations were suggested for the lack of correlation, chief among them being the idea that machines developed reputations that persisted despite the measurable behavior of the machine. Jeanette suggested that a closer look at the actual circumstances of machines-in-use might be needed to understand what was going on. Her studies showed that the “same” events measured by the company could be

28

Making Work Visible

Figure 1.4 Working with ID/HI (photo courtesy of Jeanette Blomberg)

experienced in radically different ways by machine users, depending on just how those events were embedded in a specific worksite and course of activity. So for example a paper jam encountered with a knowledgeable person nearby to consult on how to clear it might be experienced by few and quickly forgotten, while the same paper jam encountered at 5:00 by a harried worker unfamiliar with the machine and with no one else in sight would be experienced as a major breakdown. The boundaries of the machine, in other words, were reconfigured by Jeanette’s work to become less defined by the box, more contingent on its relations with its environment. This expanded view of the machine’s place in its environments of use became the basis for a long and fruitful relationship with a small band of product designers inside Xerox, first in the Industrial Design/Human Interface group, later distributed throughout the product organizations as well as in research, for whom Jeanette became both friend and mentor. Julian Orr initiated another major project, this time in response to corporate concern over the training of service technicians (Orr, 1996). Cognitive scientists on the research staff construed the problem as one of crossing the novice/expert divide, changing the mental models of novices to match those of experts. Julian began a study of expert technicians, trying to characterize expertise in the field. While the corporation assumed the work of technicians to be the rote repair of identical broken machines, Julian found this to be a gross oversimplification. He argued that the work was better characterized as a continuous practice of improvisation within

Work Practice and Technology

29

Figure 1.5 From Orr 1996 (photo courtesy of Julian Orr)

a triangular relationship of technician, customer, and machine. Individual machines are idiosyncratic, he argued, with different histories of use and modifications, and set in different social contexts, all of which affect their performance. Service calls are occasioned by a problem in the relationship between user and machine, and while the machine may need repair in itself, fixing the relationship is essential to the satisfactory completion of the call. During an observed diagnosis, moreover, Julian noted that the technicians were exchanging stories about earlier diagnoses and misbehaving machines from the past. When looked for, stories seemed to be everywhere in technicians’ practice. In fact, Julian discovered that narrative is a primary element of their work. More specifically, diagnosis involves the creation of a coherent account of the problematic machine, pieced together from the available, unintegrated facts. These narratives are repeated to colleagues so that accounts created for diagnosis circulate in technicians’ discourse, distributing the technicians’ experiential knowledge throughout the community. These stories are also a critical part of the social life of the community, as technicians demonstrate and share their mastery of the domain, and in doing so celebrate and create their identities as “masters of the black arts of dealing with machines and of the only somewhat less difficult arts of dealing with customers” (1996: 2). Julian’s work formed the basis for interventions, first with radios and later with “Eureka,” subsequently heralded as Xerox’s major platform for “knowledge-sharing.”

30

Making Work Visible

The black arts of networked software design were the topic of another major ethnographic study by Susan Newman (Newman, 1998). Starting with an actor network approach, Susan’s project was to enrich that perspective with a view from inside the lived work of designing social/technical networks – in this case, the design of networked software known as “middleware.” Working from an actor network perspective, Susan describes herself as “inhabiting the emergent actor network [of an associated information system] with, and from the point of view of, a particular group of software engineers . . . In this way, I hope to gain an understanding of the lived work of constructing and maintaining these networks, of the technosocial imaginary that animates and helps produce them, and of the forms of identity and community that are viable there” (1998: 237). The course of design work in Susan’s observation involves the mutual stabilization of the technology-under-development and the network of interested actors, through a process that is endlessly dynamic as networked relations shift and unfold. She argues, moreover, that the relations involved are multidimensional, accountable, and frequently contradictory, in ways that have not been adequately thematized in social studies of science and technology. Susan suggests that a central method of sociotechnical design work is what she calls “techniques of virtuality,” that is, performative practices including talk, textual-graphical productions, and enactments or animations, through which developers bring into the present and into immediate, lively juxtaposition elements that are otherwise distributed in time and space. Her formulation sheds light on the findings of an earlier project in a very different setting, aimed at developing our theoretical and methodological resources for analyzing work and technologies-in-use. I won’t begin to do justice to the project here – many of us were involved in it, including Chuck and Candy Goodwin (Goodwin and Harness Goodwin, 1996; Suchman, 1997). This was a project known as the Workplace Project, which was about looking at technologies-in-use in their actual work environments. We went out to find a multiactivity, technology-intensive environment, and ended up in the operations room of a local airport – the place where all of the work of ground operations is coordinated. Given the limits on my time this morning, I’ll give you just one of the morals of that story. Basically, our study of the operations room led us to a substantial reconceptualization of what actually constitutes an “information system.” Rather than a very large workstation that encompasses more and more of what you want to know, or even a network of workstations in which information is stored, our analysis showed information systems to be made

Work Practice and Technology

31

up of a multitude of partial, fragmented, information resources, which through the skilled practices of people making use of them could be related, juxtaposed, read against each other and in relation to an unfolding situation, and thereby turned into a working information system. The system is, in other words, this irreducible assemblage of artifacts and the practices of their use. This appreciation for the deeply relational character of information resources has formed the heart of David Levy’s work on the document, where the relevant relations are not only work and technology, but also continuities and discontinuities across time and across disciplinary boundaries. David has worked to clarify the relations between historical and contemporary document practices, and between handcraft – for example, calligraphy and bookbinding – and mechanical practices. At the same time he is known as a translator between the computer science and library communities, addressing issues such as cataloging, preservation, document standards, and reuse as common concerns, but often very differently conceptualized and practiced across the fields. Finally David has done much to debunk the breathless hype of the digital, providing a more nuanced analysis of the actual continuities and transformations of new media and what it means to live and work with and through them (Levy, 2001). Studies like these suggest that rather than discrete moments of production and consumption, technology-intensive work practices involve complex relations of appropriation and reworking across sites of professional design and design-in-use. This brings us to our final agenda, of interventions into practices of professional work and technology design. One model of the problem with existing design practice would take it to be a matter of individual behavior, requiring upgrading of designers to bring them in line with some normative prescription of good design. In contrast, we see the problem in terms of the social arrangements and institutions through which design gets done. Our experience is that even designers who would like to have a closer relation with situations of technologies-in-use find themselves in working arrangements that make it very difficult to do so. That means that rather than an intervention aimed at changing individual behavior, ours is really a project of social change. We were able to take on that project directly, at least in a small way, beginning in the early 1990s, thanks to our developing relationship with many of you in this room working on new forms of system development. Around this time as well Randy Trigg, who had cut his ethnographic teeth on the Workplace Project, rejoined us after a stay in Denmark, bringing both rich personal experience in the practice of participatory design as well

32

Making Work Visible

Figure 1.6 Interaction Analysis Laboratory (photo courtesy of Brigitte Jordan)

as an extensive background in the design of hypermedia. Inspired by Randy’s return, we decided to attempt to practice what we were beginning increasingly to preach. Our own projects in practice-based, work-oriented design had to be organized in particular ways given our positioning here at PARC. Most obviously that meant an orientation to document-related work practices on the one hand, and to technologies aimed at translations across paper and electronic media on the other. First in a law firm, and most recently with a team of civil engineers designing a bridge, we’ve worked across what in some respects are incomparable sites of work and technology. In both of these sites our research has involved intensive engagement with the specifics of the work and its contingencies – in part these projects for us are an occasion for workplace ethnography that extends beyond our particular design agenda. At the same time, our projects are also very much about seeing how studies of work and cooperative design can be woven together, through the mediating practice that we call case-based prototyping. And we’ve identified a consistent design focus across these projects; that is, on what we’ve come to call working document collections and the requirements and possibilities for moving them online (Blomberg et al., 1993; Suchman et al., 2002; Trigg et al., 1999). While Jeanette, Randy, David, and I have been toiling in the fields of participatory system design, Gitti Jordan has been taking practices of

Work Practice and Technology

33

participation in a somewhat different direction (Jordan and Henderson, 1995). As a founding member of WPT, Gitti has worked throughout the years to maintain methods of interaction analysis at the center of our practice. That project grows out of her longstanding commitment to methods that recover the details of how people work, and how they engage more broadly with each other and with the material worlds that they inhabit together. Beginning in the early 1980s, Gitti has organized and championed collaborative Interaction Analysis Labs both at PARC and IRL, as well as at universities like the Johannes Keppler University in Austria and the University of Illinois. One of the core concepts that has developed out of Gitti’s work is the notion of “authoritative knowledge,” particularly in its relations to technology. Recognizing that those who do the work often are allocated the least authority, Gitti has pressed to incorporate participants themselves into the process of work analysis, for example, workers’ IALs in the context of a major organizational assessment and development project inside Xerox, the Work Practice and Design Project. So there are the last twenty years for us. The question that we want to take up today is: Where do we go from here? And by “we” in this context I mean not only those of us who have been part of the Work Practice & Technology group. As most of you know, our futures will no longer be quite so intertwined, as we find various new venues that will allow us to pursue these lines of research. So “we” in this context is all of us, the collective: maintaining connections not only to each other but to those of you with whom we’ve worked over the past twenty years is our primary concern. At the same time, our twenty-year time frame is meant somewhat ironically. We’re not futurologists, and we take seriously the suggestion of Donna Haraway that “the only way to find a larger vision is to be somewhere in particular” (1991: 196). Our discussions, in that spirit, are meant to reference our current projects as a basis for imagining what could come next.

2

Engineering Investigations What Is Made Visible in Making Work Visible? Wes Sharrock and Graham Button

Introduction “Work” has been a long-standing topic for a number of disciplines. One of the “founding fathers” of economics, Adam Smith, developed the concept of the “division of labour,” the idea of optimising the organisation of work in mass production. In philosophy, Karl Marx, in part, defines what it is to be human by our need to work on our environment. In psychology, the subdiscipline of ergonomics was developed in order to better design the fit between the physical and psychological needs of individuals; the organisation of the workplace, equipment, and machinery; and the job itself, by measuring working performance. However, it is within sociology that work has assumed a particularly central role. Not only does it play out across the discipline’s major theories, it has become a subdiscipline in its own right: “the sociology of work.” Despite this wide and diverse interest in work across a range of disciplines, and particularly in sociology, Xerox’s interest in studying people doing work, in the midst of them actually doing it (which is what has become known as work practice studies), has been mainly centred around a relatively small branch of sociology, ethnomethodological studies of work, initially developed by Harold Garfinkel (1967). Early research by the manager of the original Work Practice and Technology group at PARC, Lucy Suchman1 was influenced by ethnomethodology and conversation analysis2 (e.g., Suchman 1993, 2000), and although, during its history,

1

2

34

Suchman was initially hired into PARC from the Anthropology Department at the University of California, Berkeley, to explore the relevance of social and cultural matters for the design of technology. Conversation Analysis is a particular development of ethnomethodology founded by Harvey Sacks (1992).

Engineering Investigations

35

members of her team might not have regarded themselves as “ethnomethodologists,” they certainly were not averse to it (Suchman et al. 2002). With the establishment of EuroPARC in the UK as PARC’s European Laboratory in Cambridge in 1988, ethnomethodological studies of work assumed prime importance in Xerox’s research program into work practices. The involvement of ethnomethodology with Cambridge EuroPARC was motivated by Xerox researchers’ developing interest in understanding customers’ actual work practices. Tom Moran, founding manager of the Cambridge Laboratory, understood this to mean that a better understanding of the design process (in developing software systems for work-related and work-supporting technologies) within Xerox as well as elsewhere was required. However, finding UK researchers who were interested in actually conducting “work practice” studies, as opposed to developing theories about work and organisations, or grinding theoretical axes about the social effects of technology development, was not as easy as might have been supposed within Xerox, given the way that work practice studies had developed at PARC. That development might have led PARC researchers to assume that social science interest in work practice studies was more universal than it actually was. Amongst the relative few with a strong, if any, interest in the actual study of work as it was done, were UK sociologists with a background in ethnomethodology. For reasons solely internal to sociological reasoning, they had developed a motivation to develop a (presuppositionless) study and description of the detailed specifics of work activities in organisational settings, the latter interest being one which dovetailed nicely with EuroPARC’s requirement to access the practical actualities of the design process. Throughout the history of EuroPARC, and subsequently Xerox Research Centre Europe,3 researchers with an ethnomethodological and conversation analytic interest were hired to build up work practice studies. Thus Bob Anderson, Wes Sharrock, Graham Button, Richard Harper, Christian Heath, Jon O’Brian, and Peter Tolmie were sociologists strongly associated with ethnomethodology who worked at EuroPARC and Xerox Research Centre Europe (XRCE) over a 17-year period. In addition, other human scientists with an interest in design, designers, and computer scientists who either worked closely with these ethnomethodologists or became actively 3

Following the success of EuroPARC, the growing importance of the European market place, and the willingness of the European Community to fund research into systems development, it was decided to open a European Research Centre, subsequently known as Xerox Research Centre Europe, and EuroPARC was incorporated into this new Research Centre in 1994.

36

Making Work Visible

involved in developing an ethnomethodological perspective in design were part of the work practice contingent: Paul Luff, Paul Dourish, Dick Bentley, James Pycock, Jackie O’Neill, and Dave Martin. There were others, such as William Newman, Marge Eldridge, and Allan MacLean, who, although not adopting an ethnomethodological framework, nevertheless worked closely with some of the ethnomethodologists. The technology developments of the current Work Practice Technology group at XRCE in Grenoble, under the management of Antonietta Grasso, continue to be heavily influenced by an ethnomethodological orientation through the influence of Jackie O’Neill, Dave Martin, and Tommaso Colombino. Over the years, Xerox’s work practice study groups have often been strongly associated with ethnography and design, and there are numerous papers that have championed the use of an ethnographic method. Some of these have sought to clarify the differences between perspectives that are bundled together under the heading “ethnography,” and have attempted to differentiate an ethnomethodological approach from other “ethnographic approaches” (e.g., Button 2000). However, none of these have specifically addressed what is different about ethnomethodological studies of work from other sociological studies of work. Given the title of this collection, Making Work Visible, it is perhaps appropriate in developing an understanding of Xerox’s work practice studies to also make visible the distinctiveness of these studies with respect to other studies of work, just what it is that these studies make visible. There are two sorts of issues which have been and continue to be debated relative to the input of human and social science research into design to support work (though there are current debates about whether supporting work is as important as it once was (Crabtree, Rodden, Tolmie, and Button 2009).4 There are issues relating to the understanding of work, and issues about how that understanding can feed into better design. In addressing the question of what is made visible in making work visible we will focus on some issues of the first kind here, specifically those which relate to the role of plans in work activities. Xerox work practice studies have been very practically oriented. They have tended to fall into two domains. First, they have been intended to either generally influence how systems designers approach design with regard to building in the practicalities of using a system within an organisational context, or to influence the design of a particular system that is being 4

Note, however, the point to be made below, that ethnomethodological studies of work did not assign the study of work any analytical importance.

Engineering Investigations

37

built. Second, those doing work practice studies have also worked alongside Xerox consultants in building an understanding of the operations and organisation of a particular Xerox customer, with the objective of enabling Xerox to better support that customer than its competitors. There are numerous examples of both domains of practical investigation in this volume. This chapter, however, is concerned with emphasising and making visible the distinctness of Xerox work practice studies, for other ethnographically founded studies might also claim such practical credentials.

Ethnomethodology and the Everyday World In many ways the interest in work (and in work in organisations) was incidental to the sociological considerations which drove it. In order to understand how ethnomethodological studies of work are distinctive from other studies of work, and thus what has been particularly distinctive about Xerox’s work practice studies, we will first explore this seeming contradiction. This exploration may be of less relevance for those who are interested only in the outputs of work practice studies for practical design purposes. However, we suggest that given the range of social science interests now being brought into the general design arena, it may enable designers better to weigh the appropriateness of what they are being offered as insights into how people in organisations work. For such offers as are made by conventional sociology and anthropology differ greatly from what has been offered by “work practice” studies. Ethnomethodology took an (sociologically) unprecedented interest in closely examining the world of everyday life in its own terms, which – perhaps because of the pervasive and familiar presence of everyday life – had classically, within the human sciences, been either excluded or marginalised. Before we briefly examine the arguments for examining the world of everyday life, we should note that ethnomethodology’s concern with it could be explored relative to any of the endless catalogue of activities going on in the everyday world, from walking down the street, through putting up a garden fence or driving a delivery truck, to preparing a specimen for biological testing. From an ethnomethodological position it really did not matter what was studied, since any activity could instantiate the everyday organisation of activities. One could as easily and – for ethnomethodology’s purposes – relevantly study activities outside work (such as, inter alia, leisure, domestic, or political doings), and, reciprocally, one could equally well study work, and study work in organisational settings. In key analytical respects, it does not matter which sector of social life is the occasion for study. The interest

38

Making Work Visible

in work in organisations developed within ethnomethodology is thus developed from the point of view of their character as everyday activities in everyday settings. So, why is it that contrary to classical human and social science interests, ethnomethodology stressed the world of everyday activities and everyday settings? The word “presuppositionless” used previously was not inserted as a goad to those who would want to say something like “the idea’s nonsense! All thought cannot help but involve presuppositions!” Crudely, the point of developing a presuppositionless sociology concerns the fact that sociology is very much a top-down operation, which sees research as mainly a matter of applying a preconceived sociological scheme to the description and analysis of empirical phenomena. First one develops a putatively general sociological scheme, then one asks how that scheme’s terms apply to the things that people do. The “presuppositionless” was inserted to convey a different idea to the predominant one in sociology. Instead of observing social life through, and on the basis, of a weighty apparatus of sociological assumptions and preconceptions, this can be countered with a concern for a much more bottom-up possibility, where the aim is not to see how phenomena look when viewed for sociological purposes but rather of observing them in their own right, without having any apparatus of sociological requirements constantly in mind. After all, this is how the people who are doing the activities that the sociologists aims to describe view the things they are doing and have no need for sociological assumptions and presuppositions to do and understand the things they do. The top-down approach (which governs a lot of sociology’s methodological thinking too) can seem back-to-front, as though sociologists generally think of social life as though it were conducted and organised to serve the interests of sociologists. Sociologists commonly attribute the well-known, even notorious, failures of their enterprise to progress as a research discipline to the difficulty of studying their chosen phenomenon, the organisation of social life. It is as though people – despite speaking and having a language which makes it easy to study them by asking them questions – have even so organised their affairs in such a way as to make it difficult for sociologists to study them. Against this, there is the possibility that people organise social life without sociologists in mind, and that the way in which activities are put together has more to do with the things that matter to those doing the activities than to the requirements of sociological theory and methodology. Sociologists revel in theoretical argument, and there is a huge one here about whether we – the sociologists – or they – the people we study – have

Engineering Investigations

39

the better understanding of social reality. However, we would suggest that before getting into such an argument, it might be desirable that sociologists ensure that they have actually got a good grasp on what it is that “they” understand. This kind of question is often treated as asking about the importance of “the actor’s point of view” in sociology, often discussed as though it were no particular problem to capture it, and the only real questions were about where to place it in a general overall scheme. Can sociology as a whole be achieved through seeing things “through the actor’s eyes,” and would not that threaten the very idea of sociology, reduce it to a mere repetition of actors’ beliefs and opinions? Ethnomethodology takes the view, however, that sociology, in general, has not been very successful at “capturing the actor’s point of view.” This task has been fairly low priority for many sociologists, and has not involved any systematic effort to figure out how to capture and portray that point of view. Further, insofar as the actor’s point of view is taken into consideration, the “social actor” tends to be treated rather as an amateur theorist of social life, whose understandings of society would be possible rivals to those of the professional sociologist. For many sociologists, it is meaningful to ask “Do the members of society understand society correctly?” and to determine the answer to that by comparing how members of the society think about society, what sorts of beliefs and opinions they have about it. For ethnomethodology, however, those efforts are generally misdirected just because sociologists neglect the fact that “social actors” theorise about society only relatively rarely, and that those actors are mostly engaged in practical conduct in doing things, meaning that their understanding of “society” is overwhelmingly involved in knowing how to carry out the affairs that make it up. So, rather crudely, (a) the actor’s point of view is a matter of “knowing how” rather than “believing that” and (b) “capturing the actor’s point of view” is a matter of investigating how actors, in the midst of things, do not normally find themselves at (rarely more than a momentary) loss; they know what to do, and know how to respond to the pre-eminent question “what next?” Thus, instead of asking questions as to whether actors (misguidedly, in the view of most sociologists) believe that society is equal, just, and fair; that there are opportunities for social mobility; and so forth, ethnomethodology wants to ask about the “know how” involved in activities, which is, really, the same as asking how activities are competently carried out. To indicate the distance that these questions have from the more traditional sociological concerns, ethnomethodology replaces the

40

Making Work Visible

expression “social actor” with “member,” and, as just indicated, “beliefs” and “ideologies” with “competences.” The gap between the sociologists’ theoretical interest in society, and the members’ practical involvement in it is emphasised by the assumption that the members of society do not live their lives or arrange their affairs for the purposes of sociology, and that therefore there is no inclination to search selectively for “sociologically interesting” things about members’ everyday activities, displaced by an urge to establish, first, how they arrange their affairs, independently of all (sociological) preconceptions about how everyday activities “have to be” because some comprehensive sociological scheme supposes that this or that is what people are “really up to.” As far as they are concerned, the questions are: What are they doing? and How do they organise doing that amongst themselves? Those sociologists who resist the idea of giving the actors’ point of view pre-eminent emphasis often do so because they imagine it must unavoidably be highly individualistic (since it is individuals who have “points of view,” is it not?). To counter this, though, note that those engaged in everyday affairs are well aware that they live in a social world, that the things they individually do must, for all sorts of purposes, take account of what others do. Rather than thinking of sociology as the study of “subjectivity,” ethnomethodology inherits a conception of it as the study of “intersubjectivity.” Rather than as the study of “social action,” it thinks of it as studying “social interaction.” The question “What is the actor’s point of view?” is thus answered by establishing, neither more nor less than “What do members have in mind when they are doing X (or Y or Z)?” This question is answered investigatively on the assumption that the members are organising joint activities in “real time,” by themselves, recursively figuring out practical answers to two preeminent practical questions: “Why that now?” and “What next?” Members’ doings are thus, as is now fashionably said within some quarters in the design community, “situated.” However, this only means that anything anyone does is done in some here-and-now, hence members’ conduct is a response to that here-and-now situation. What the response can be will depend on what the situation is understood to be, hence “Why that now?” or, more elaborately, “What is going on here that implicates ‘me’ and calls for a response?”. Understanding the situation involves figuring out “What next?” recursively over the course of an activity. This is all that is meant by the idea that socially organised activities are understood “from within,” a matter of analysing how people organise a course of action whilst they are carrying it out, as part of carrying it out.

Engineering Investigations

41

Ethnomethodology’s investigations, themselves, feature the question: “How can they tell?” which is a means of focusing queries about what one needs to know/be able to do, in order to participate competently in some perhaps quite run-of-the-mill activity such as a commonplace situation of, for instance, complaining. How can those doing something tell what the effective and/or appropriate thing to do is? Ethnomethodology’s idea is that any and all sociological topics can be converted into inquiries focused on such questions, identifying the kinds of people who encounter that topic as a practical activity in the course of their affairs. All kinds of sociological questions are potentially open to this strategy, and it can be applied to work, workplaces, and organisations as readily as to anything else.

Ethnomethodological Studies of Work Seemingly contrary to what we have said previously about “work” having no special standing amongst ethnomethodology’s topics, Harold Garfinkel initiated a “studies of work” program (1986). However, the point of this was not to single out the study of work as opposed to the study of, for instance, hobbyist golfing, but in order to make again, and very forcefully, ethnomethodology’s point about the exclusion of “everyday life” from sociology’s interests. This was very noticeable with respect to the established sociological specialism of “sociology of work,” which, ironically, seemed to be interested in everything that went on in workplaces except whatever work was getting done there. Garfinkel’s observations were paralleled by those made by Anselm Strauss and colleagues coming from the different “interactionist” tradition: . . . remarkably little writing in the sociology of work begins with the work itself (except descriptively, not analytically) but focuses on the division of labour, on work roles, role relationships, careers, and the like. A concerted analytic examination of work itself ought to provide a needed corrective to more traditional approaches, which, however effective, still leave important issues untouched or unresolved. (Strauss et al. 1985a: xv)

Garfinkel’s idea for ethnomethodological studies of work was that it would reinstate an interest in the organisation of work activities as a suitably sociological topic. After all, work is very much an everyday affair, and members’ understandings of work derive from their experience of doing work. Garfinkel asked a very simple question which was, to paraphrase his argument, “What is it that people are doing, often together, when they understand themselves and can be understood by others to be doing

42

Making Work Visible

work?” It is this “what” that is missing from sociological examinations of work that are interested in issues about, for example, the social role of the person engaged in work, as opposed to what those people are doing when engaged in that work. Mike Lynch restates this point in these terms: Garfinkel introduced his proposal to study the “missing what” of organised complexes of activity by crediting Harvey Sacks with an insight to the effect that virtually all the studies in the social and administrative sciences literatures “miss” the interactional “what” of the occupations studied: Studies of bureaucratic case workers “miss” how such officials constitute the specifications of a “case” over the course of a series of interactions with a stream of clients; studies in medical sociology “miss” how diagnostic categories are constituted during clinical encounters; and studies of the military “misadjust how stable ranks and lines of communication are articulated in and as interactional work. (Lynch 1997: 270)

In the remainder of this chapter we will use an example, taken from our own studies of engineers,5 to show something of the nature and consequence of taking Garfinkel’s advice seriously.

Engineers and Planning One of the central ideas of “work practice,” and of Computer Supported Cooperative Work (CSCW) in general, was that system designers were often solely reliant for understanding what work they were designing for on procedural prescriptions – charts and plans – that the organisations themselves provided. From these, the designer could work out the order and procedure of work processes and could use these as models for the development of the software system. This presumed something that could soon be disproved by even a brief examination of work on the shop floor or in the corporate office – which is that the procedures followed and practices deployed by those doing the work do not automatically, and in many cases at all, match those in the organisation’s manuals (Bowers et al., 1995). It was held that designers could learn from studies of what was actually done to do the work since this would commonly deviate from the official procedures, and if, software systems’ designs were to assist, not obstruct and disrupt, work, they had better know something about what work really, practically, consisted in. However, this sound observation was caught up with the disagreements amongst human and social science approaches (including 5

The engineers we describe were software engineers developing the software for photocopiers. We worked with them on four development projects over a three-year period.

Engineering Investigations

43

orthodox cognitive science, distributed cognition, anthropology, symbolic interactionist sociology, and ethnomethodology) that were all united in thinking that a better understanding of how work gets done could contribute to improved design. One of the disagreements between the different human and social science disciplines that were addressing design was over the relationship between the formal procedures, often described as “rules,” or “plans,” and conducted with respect to what to make of rules or plans in the construction of general models of conduct. Some understood orthodox-cognitive models as entirely planful and rule governed, treating rules as causally determinant of sequences of action in much the same way a program runs a computer (cf Suchman’s 1987 arguments with respect to plans). Against this, it was argued that human action is better understood as much more a matter of “improvisation,” with all action being “situated,” meaning that what people did in any given situation was not in any way preplanned, but decided upon right then and there. Taken to the extreme, the latter view (see Schmidt 1999 for a presentation and criticism of such views) became one that people do not follow rules and plans at all, and that it only seems as if they do because people appeal to rules or plans, but only in an after-the-fact manner, appealing to rules or plans in retrospective justification of what they have done quite independently of any plan.6 There is no room to follow out the intricacies of controversy over these matters. Indeed, looking for understanding in following these intricacies is not the point; the point is to understand how the formalised prescriptions are actually used in practice. No matter how social and human scientists want to theorise (in the top-down manner we referred to previously), people in their actual work will commonly use formalised procedures in some way to accomplish their work, and the issue of the relationship between the procedures and the work will be settled by those using them in the actual doing of their work. The question is “In what ways does what they are doing have to do with being planned?” It is in this missing “what” of the work that an answer will be found. Thus, the place to look for an answer is in the practices of those who, as a matter of plainly observable fact, spend their days working with rules or plans. What counts as “following the rules or plans” amongst them? In doing that we will try and illustrate how ethnomethodology attempts to make visible the “what” of the work that is missed in top-down descriptions. To do this we briefly sketch how plans figure in the work of (some) engineers, because we have found that engineering work is a domain in which planning is a pervasive activity. 6

For more extensive discussion of the general issues, see Schmidt (1999) and Dant and Francis (1998).

44

Making Work Visible

To begin, we need to note that planning is itself an activity, and the formation of plans can be just as problematic and troubled as the activities for which they plan. The project plans we discuss were extensively composed from planners’ estimations, for example, of how long it should take to figure out, design, and install a prototype paper path in a networked printer; the machine process, and what a viable unit cost for the production of the printer would have to be (which in the cases we are familiar with was a matter of the cost at which the product would be marketable, price competitive with comparable machines, both of their own and other manufacturers). The plans were not prescriptions for how the engineers were to go about solving their project’s engineering problems but, instead, generated some of those problems – the specification of a given level of performance from, say, a network printer (in pages per minute) or in frequency of technician call outs (per million pages) could present engineers with serious difficulties; requiring that the speed of paper being transported along the paper path be faster, but only marginally so, than usual, could represent a significant raising of the engineering stakes – a small adjustment in speed can ramify through the entire design. Even if the engineering requirements were not steep, the timetabling of them could create problems: yes, that could be done, but in so short a time? The design and development steps and their timetabling were less prescriptions than challenges: it was not the planners’ but rather the engineers’ job to figure out how to meet (if at all) the requirements the plan sets out. The plans’ mode was more that of “by date X, the following tasks will have been completed” than it was “complete this task by doing first a, then b, etc.” This is why we say that the plan frequently maps achievements – those required from the engineers – and why its constituent tasks and dates are often more like challenges to the engineers than prescriptions for determinate courses of action. No one need be under any illusion that what the plan asks for can easily be attained, nor exclude the possibility that the difficulty of reaching those attainments may turn out to be much greater than anyone, including the planners, had imagined. The engineers treated the plan’s demands as challenges: even if they were deeply skeptical about the possibility of a project actually working out (it was not unusual in their experience for projects to “fail,” though they nonetheless made strenuous efforts to find ways of making viable the projects they were attached to). One project to design and develop an add-on to a printer was planned to deliver only a small number of add-ons so as to make a small number of machines stuck in inventory more marketable. This immediately resulted in a problem with unit costs. The only way the units could be manufactured at the

Engineering Investigations

45

plan-specified cost was through a much larger production run. The team of engineers promptly made efforts to identify additional marketing possibilities for the add-on. There are many different kinds of plans. For example, there are “right first time” plans, plans designed on the premise that each step is taken in the appropriate way at the appropriate time. Such plans are often disparaged since it is a professionally known feature of planning for software development that things rarely go right the first time, so plans with possibilities of iteration are often recommended. Nonetheless the company’s deliberate policy was to design plans on a “right first time basis,” though hardly naïve about whether things do go right first time. All the projects under way at any point in time would not have “gone right first time” and would be struggling with problems of design, scheduling, costing, or all three. The “right first time” structure of plans was a policy to keep project teams at full stretch. It would be, at the least, extremely demanding for a project team to pass milestones on schedule or the first time around, for its technical objectives prove no more problematical than expected, and for unit costs to be within easy reach. The plans were implemented through a procedure of “management by problem solving”, which involved identifying, grading, and enumerating the problems the project was encountering, with one category, that of “major,” for those problems that were possibly “project killers,” that could result in the project’s failure or abandonment, and that, at the least, would have deleterious consequences for schedule, quality, or cost. It was unexceptional that projects had, at certain stages, problems in the “major” category. That things did not proceed according to plan was a “normal, natural trouble” (to use a phrase of Garfinkel’s) of engineering. Project planners, project managers and engineers were all certain that any project would have problems with Q(uality), C(cost), and D(elivery), but they commonly could not anticipate what these problems would turn out to be, nor how disruptive they would prove. The project’s plans provided a point of reference for daily work of the project’s engineers. Much of that work was done in meetings, many of which were called for problem-solving purposes. It was routine in such meetings to invoke the plan for purposes of determining where the project’s progression was relative to the plan. “How are we doing?” was a routine question, to be answered by comparing the planned schedule against actual progress. Such comparison could easily be deceptive, such that, for example, the cumulating record of problem types could be used to figure out whether the project was as securely on schedule as it appeared. It might look as though the tasks currently being disposed of were those that the project was scheduled to

46

Making Work Visible

have completed by the current state, but the number and/or severity of unsolved problems was a working measure of whether progress had only kept to schedule because difficult and time consuming problems were deferred and could yet “come back to bite” them. The matching of planned to actual progress had diagnostic and steering uses – reasoning around enabled (initial) determinations of whether things had gone off track and what specially was holding things up. Working out where the project’s troubles lay was a basis for working out how things might be got back on track, where and when to concentrate effort in what further activities. One project had what everyone recognized was a ridiculously tight schedule, so much so that the planned transfer of the designed product into manufacture would carry the project well beyond its supposedly final and fatal deadline. In light of this, the project team decided to start setting up arrangements with the company’s manufacturing plants at a much earlier state than the project plan required so as to buy time for later. The engineers were inveterate planners. As mentioned, much of their work was done in meetings, and those meetings would themselves be planned at their outset, whilst the objective of the meeting would, itself, be to form a plan for solving an engineering problem. Such routine planning was “good practice” in terms of the corporate methodology, but this was not incompatible with the engineers’ own assessment of how their work was optimally organized. The planning of the meeting itself was employed as a means of generating productive meetings: an outcome for the meeting would be agreed, essential topics for discussion to realize the outcome listed, and then these used to constrain subsequent talk, to constrain it so that it remains thematically focused, so that progress can be assessed relative to the time remaining (how long the meeting was to last was planned, and running out of time brought a dead stop). Planning was not unrealistically conceived as a means of eliminating problems, but rather as minimizing known risks and containing the consequences of problematic contingencies, with no guarantees that the plan could be carried through. Amongst the recurrent risks of “self-inflicted” disruption to projects were (a) overlooking key details; (b) losing sight of the plan; (c) unilateral responses and (d) failing to integrate. The engineers understood that design and development prioritized taking care of details. Getting individual parts of the machine or its software to run effectively and getting those parts to run together in a machine functioning to specification depends upon detailed interdependencies within and between different parts of the system. The engineers saw no necessary proportionality between details and the

Engineering Investigations

47

problems they can cause if neglected. The relation between the neglect and consequences was understood to depend, inter alia, on timing: otherwise trivial mistakes showing up long after they were made when much of the project’s design and development incorporating them is “set in concrete” can be much more difficult and expensive to put right than if corrected nearly as soon as they are made. Relative to this, planning was an attempt to ensure that details were not overlooked, that tasks were not taken on without first cataloguing the details essential to its successful completion, without making sure all the details are dealt with, or without being able to check, on completion, that all details have been covered. Localized task planning was a routine means of keeping the design being actually developed in line with the product being planned. There was always the risk that engineers working on individual tasks would “work out of their heads” and decide what needed doing and how to do it on the basis of their not-necessarilydependable recollection of project specifications. Working out plans for individual tasks on the basis of the project specifications was also a way of keeping the engineers’ problem solving mutually aligned. It was not so much that engineers would unilaterally decide what to do in their individual work as that they would do this and subsequently fail to notify impacted parties. Design departures from the formal specifications could – almost assuredly would have – effects on areas of design that were the responsibility of other parties, meaning that those others would continue to make their design decisions on the basis of what had now become misconceptions. Thus, changes from the specifications have to be planned before they are introduced to evaluate their impact, to make sure they are on record, and to notify affected parties. As mentioned, project work was (then) organized around “management by problem solving” with the accumulated problem’s list, with the task of working on these problems assigned to specific engineers, working them as individuals or engineers. Working on these problems was characteristically a planned activity. Much of the problem solving took the form of investigations to find out what was causing the identified problem, and it was a company standard of good practice to investigate using “root cause analysis” in line with the idea, just explained, that early identification and solution of problems was prized. It was a risk that problems on the identified problems list might be only symptoms of the problem, rather than the problem itself. A fix might eliminate only one of the symptoms, not the problem. Hence, root cause analysis was meant to provide a systematic procedure for distinguishing between a symptom and its underlying cause through collection and analysis of data with the investigation being planned so that the problem

48

Making Work Visible

was clearly defined. It was then worked out what kinds of data would capture the causal relations potentially involved in the problem, with the data then analyzed to identify any candidate cause(s), with corrective measures being implemented and further data collected to monitor the implementation. The use of root cause analysis reflects some aspects of “engineering thinking” as it was on show in engineering investigations in the projects we studied. There was a routine, self-imposed requirement for worked-out treatment of problems to ensure that the problem was extensively thoughtthrough before any action was taken; that problem solving involved both clearly defined problems and clearly defined means so that (a) these could be reviewed by other engineers to see whether all relevant means to investigate the problem had been included (b) to assess whether the investigation was plausibly an effective one before any work was done on the investigation (c) so that during and after the investigation there was something to check back on to ensure that the task was, relative to its requirements, thoroughly completed, and so that there was a clear basis for telling whether the investigation had succeeded or not. In general, then, for those engineers working on the projects we were involved in, planning was very much a routine matter. The question, which directed our investigations, was “How do project plans enter into their work and become used as part of interactionally organizing the projects as they unfold?” In summary, though not a definitive form, we offer the following observations: *

*

*

It is clear that the work is commonly organized with plans in (or near at) hand, and that on innumerable occasions part of the work is checking on what is in the plan, and in figuring out “where we are” with respect to the plan That much of working with the plan is a matter of arranging things to keep on track with the plan (or trying to), figuring how to recuperate deviations from the plan, figuring out what troubles failing to keep to the plan will cause and what to do in face of the fact the plan is clearly not working That individual designers or groups of them use the plan to understand where their current work fits in with that of others and into their assigned schedules, and can use it to work out what they need to do, how they need to organize their work to ensure their outputs interface with those of others or to determine what preparation is necessary “now” in order to ensure that later stages of the work can be successfully delivered

Engineering Investigations *

*

*

*

49

Plans are designed for use by possible users, and the project plans we have discussed were developed on the assumption they would be used by experienced (photocopier) engineers, who can draw upon their standard or specialized engineering skills – the plan assigns appropriate specialists to the project – to organize and execute the tasks the plan assigns them (for example, by invoking “root cause analysis” and “management by problem solving”). Troubles with the project’s plans are the project’s normal natural troubles – design and development engineering work is (in greater and lesser ways) troubled and nobody supposes otherwise, especially where, as is very commonly the case, projects are meant to raise quality standards, deploy unfamiliar technologies and methods, embed their project participation in an internationally distributed project working with previously unknown collaborators, condense schedules – and often to do more than one of these at once. However much planning goes into a project, experienced engineers are certain there will be problems – they just cannot tell what they will be or what will cause them – in new contexts, very familiar and routine engineering tasks can turn out to have severe and surprising difficulties. The definiteness of the plan at any one time resides in the practical engineering decisions that enable them to stay close to the machine specifications, within budget and on time. The plan is mute with respect to its own implementation, and engineers draw off engineering practice in order to do that by invoking matters such as “root cause analysis” and “management by problem solving.”

Conclusion The example we have given, which draws off our studies of four software engineering projects has been provided in order to illustrate our previous argument that the human and social sciences’ general top-down approach fails to capture how people organize work in doing it, and make that “doing” visible and understandable to one another. With respect to the issue of the role of plans in human life, the top-down approach would render humans as virtually puppets of a plan, their actions almost causally propelled by the plan. In attempting to combat that approach, the alternative top-down approach renders human action as essentially without plans, that the idea of a plan is part of a scientific apparatus for rendering human action, as opposed to part of the everyday world of experience.

50

Making Work Visible

Our brief examination of how project plans enter into the work of the software engineers displays that both of these top-down approaches miss the mark. It is quite clear that the project plans did not causally constrain the activities of the engineers, and it is equally quite clear that the engineers used and followed project plans in order to accomplish the ends of the project. Both of these top-down approaches miss the essential work of how the engineers used the project plans as part of their organisation of the project. Though we have been able only to gesture at some of the details of that work, even such a brief examination makes visible that this work: is done in the interactions of the project members and involves working out what to do; that this working out is done within the contingencies of the circumstances and involves invoking those and that in doing so are made relevant. In other words, the work of software engineers involves the application of engineering practice between the project members in order to accomplish the objectives of the project within the unfolding contingencies of the project as those present themselves and their relevance determines. This is “the missing what” of the traditional sociology of work, the interactional work of pulling off the work that ethnomethodology seeks to draw attention to. It is this order of study that has predominantly characterized Xerox’s work practice studies, and it is this order of organisation that is being made visible in the type of studies that are typical of Xerox’s work practice agenda. We reiterate, the majority of those studies have had practical objectives: either, to influence systems designers in terms of generalized design methodology and practice, to directly contribute to the design of a particular system, or to work alongside Xerox consultants in developing a much more detailed understanding of customer operations than competitors might be capable of. There are numerous examples of both types of studies within this volume. Our aim here has been to make visible an aspect of the distinctiveness of Xerox studies of work practice as part of ensuring that this distinctiveness is rigorously maintained. If it is not possible to view the details of the interactional work of pulling the work off in a study then that study may be questionable as being in the tradition of Xerox work practice studies, whatever else it might present or recommend.

Part II

Applying Work Practice Methods

3

Uncovering the Unremarkable Peter Tolmie1

At the beginning of this decade, Xerox Research Centre Europe’s Cambridge laboratory was building on the already successful work practice research undertaken in its Studies of Technology and Organizational Work group through a new research endeavor titled Emerging Office Environments. Whilst a part of this continued the tradition of looking at large-scale organizational phenomena, another part of it was tailored to a brief but concentrated Xerox interest in Small Office–Home Office environments. This interest led not only to studies of small enterprises but also of teleworking and domestic settings. There were a number of outcomes from these studies across many research themes, notably with regard to paper-digital practices. However, one of the most influential ideas to come out of these studies was the concept of “unremarkable computing.” Unremarkable computing was, in part, a response to one of the principal tropes in ubiquitous computing where there is a notion of making computing disappear into the environment or become invisible. Unremarkable computing is a foundational respecification of what that might mean and focuses upon how things might be taken to be unremarkable in practice. However, the notion of uncovering what is unremarkable in practice has a wider application. At the point of pedagogy and analytic rigor it also has considerable purchase for arriving at an understanding of what is “in plain view.” In this chapter I will articulate how the studies of home and telework environments at XRCE were pursued and how these led to the key analytic insights encompassed within the idea of unremarkable computing. Out of 1

This paper owes a debt to my co-authors for their paper entitled “Unremarkable Computing” and my other colleagues working at XRCE’s Cambridge laboratory from 2000 to 2002, the people involved in the MIME Project and the people who allowed me to enter their homes and study their everyday activities with not just good grace but positive enthusiasm for our larger project of understanding the social organisation of everyday domestic life.

53

54

Making Work Visible

this I shall be drawing some of the principal analytic and pedagogic messages this points to for both the conduct of work practices studies and the accomplishment of design.

The Disappearing Computer In 1991 Mark Weiser , one of the most visionary researchers then working at Xerox’s PARC laboratories, wrote the following words: “The most profound technologies are those that disappear. They weave themselves into the fabric of everyday life until they are indistinguishable from it.” (Weiser, 1991)

In 1994 he elaborated this idea further in the following way: “For thirty years most interface design, and most computer design, has been headed down the path of the “dramatic” machine. Its highest ideal is to make a computer so exciting, so wonderful, so interesting, that we never want to be without it. A lesstravelled path I call the “invisible”; its highest ideal is to make a computer so imbedded, so fitting, so natural, that we use it without even thinking about it.” (Weiser, 1994)

This idea of technologies that “disappear” or become “invisible” has proved to be one of the siren calls of ubiquitous computing and more than one well meaning research endeavor has foundered upon a misunderstanding of what it might mean. This vision became the cornerstone of a European Community funded initiative called “The Disappearing Computer.” It brought together research groups who sought to activate this idea through projects focused on a range of different interests including the augmentation of physical artifacts (Luff et al., 2007), design for domestic environments (Lindquist et al., 2007), and the creation of what might be called “smart spaces” (Streitz et al., 2007). The initiative was enormously productive and gave a firm thrust forwards to the nascent ubiquitous computing community. However, for many researchers working in this domain the focus became increasingly one of how to make computers quite literally disappear or be physically removed from sight. The photos in Figure 3.1 included in Philips Design promotional materials of the time (Philips, 2001) are indicative of this particular understanding of what disappearing computing might be about. Researchers at XRCE at the time were not unsympathetic to work progressing in this direction but it was felt that these literal interpretations of what disappearing computing might be about were somehow missing the point. It was against this

Uncovering the Unremarkable

55

Figure 3.1 “All sorts of computing devices will disappear into the background of our everyday lives” (Philips, 2001). Reprinted with permission

conventional view of disappearing computing, which took perceptual awareness of computing to be the issue, that the notion of unremarkable computing was first developed.

Unremarkable Computing “Unremarkable Computing” is the name of a research paper first presented at the ACM’s conference on human factors in computer systems (CHI) in 2002 (Tolmie et al., 2002). As an idea it takes seriously Weiser’s proposition that computing might be made to become “invisible” in practice. The latter part of this phrase is critical. Invisibility is one thing and there are many ways in which people might talk about things as being invisible, but what does it take for something to be understood, spoken of, and oriented to as invisible in the first place? Is it, for instance, all about perceptual availability? If you can’t see, hear, smell, touch, or taste something does it therefore blend into the background and to all intents and purposes “disappear”? Or does this facility of disappearing in use, of blending into the background that Weiser took to be so important, turn on other things aside from matters of physical apprehension?2 What Does It Take to Become Invisible in Use? As the first of two “perspicuous” examples (Garfinkel, 1967) originally tackled in the paper titled “unremarkable computing,” let us consider the 2

There is, indeed, an argument to be had about what matters routinely referred to by many disciplines as “perception” or “apprehension” really turn upon, but we are devoted here to a consideration quite specifically of the notion of “invisibility in use.” For a broader treatment of the issue see Coulter and Parsons 1989.

56

Making Work Visible

following extract from a set of ethnographic field notes captured during a study of a freelance language translator working at home. The translator in question, whom we shall call Lucie for the purposes of reference, lived in a small 3-bedroomed house with her 2 children, a boy aged 12 and a girl aged 10. To facilitate her work, Lucie had converted one corner of her living room into an office. Freelance translation work is typically paid by the word. Lucie therefore routinely started work early in the morning before her children had got up. This way she could get a certain amount done without any kind of interruption. The following notes come from observations of one such early morning session. Lucie had been working at her desk since about six o’clock. When she began her children were asleep upstairs but over the course of the observations their morning routine was initiated through a series of commonplace but significant events. Instance 1: Lucie flicks through some printed sheets on her desk and comments on how the table of contents doesn’t match the text. She returns to the electronic document and continues to translate the next title, saying a segment out loud. It is 7:00 a.m. and an alarm clock goes off upstairs that she shows no reaction to, continuing to key in as before. When she has completed that section of text she switches her monitor off and says “it’s been an hour.” She pushes in the leaf to her desk, stretches, then leans on the ledge under her monitor resting on her elbows, her hands to her cheeks, drinking coffee. Once she has finished her coffee she goes into the hall to call upstairs to the children: « Bonjour mon gros doudou, Bonjour mon lapin . . . »

So what might give an ethnographer licence to observe that something like an alarm clock going off was such a thing that was “[shown] no reaction to?” First of all it is important to note that, for the ethnographer, the alarm going off was something that was clearly remarkable. This is evidenced by special note being made of it in the field notes: it is remarked upon as a feature of potential relevance. However, for Lucie, the self-same thing had been completely unremarkable – so much so that, for a witnessing party such as the ethnographer, not specifically competent in all of the routine features of the setting, they could reasonably say she “showed no reaction to it.” In order to elucidate some of the methodological features of what’s going on here it’s worth reflecting upon some of the perfectly plausible things that didn’t happen. One thing that clearly didn’t happen was that Lucie did not mark out the occurrence as something unusual by saying “whatever is that?” Neither did she orient to it as something regular but still notable by saying something like “heck, there goes that alarm again!” Nor does she remark upon it for its significance in relation to some other matter, perhaps by saying “oh is

Uncovering the Unremarkable

57

Figure 3.2 Disregarding the alarm going off?

it that time already?” In consequence one can see that she displayed an orientation to the alarm going off as being something that was wholly unremarkable: it did not merit any kind of comment or display of interest at all. Indeed, although it was seemingly handled without any kind of reflection, from a methodological point of view one can say that she manifestly did not mark out her interest in the occurrence, and thereby provided for its sense as something that is wholly routine, for who would ever think to comment upon some feature of a routine as though it were something special? This is the point. To show any evident interest is to make the matter intelligible in some other way. By not marking out the alarm in some way makes it clear that there is nothing inherent in it’s going off that obliges her to treat it as something worthy of note or in need of comment. Yet, from a perceptual perspective alarms are understood to be designed specifically to be noticed and attended to. This reveals a certain confusion in conventional reasoning about designing for visibility on the one hand, and invisibility on the other. For the translator, the alarm clock going off was apparently “invisible” in use, yet it was not wholly unattended to. It was still taken to be implicative, but it was taken to be implicative in an ordinary and routine fashion. Just because some alarm going off is treated as unremarkable this does not mean it is without import. Indeed, the alarm going off was full of import

58

Making Work Visible

for Lucie’s children. And the fact that it was available to both Lucie and her children would have made any failure to recognise that import immediately accountable. So, whilst the alarm may have been unremarkable, it was still a resource that was used by members of the family with many other things turning upon it because it was one of the principal devices used to initiate the morning routine. This leads us to another important observation: it is not the case that Lucie had simply not noticed the alarm going off. On the contrary, she did not mark out through some visible display that it was notable because such a display would have made her accountable for her interest in its significance. What she could have done was to display an interest in its failure to go off. If it was ten minutes past seven and she hadn’t heard the alarm yet she might well have remarked upon it. Furthermore, if the children had subsequently failed to materialise she could also have held them accountable for that in terms of having heard the alarm. In this way we can see that things can be “perceptually available yet practically invisible-in-use” (Tolmie et al., 2002). At the same time, things can be perceptually absent yet quite specifically marked out. So what matters is not perceptual character so much as significance. From Invisible to Unremarkable In the previous example we have two somewhat contradictory views upon a phenomenon taking place. For one party, the ethnographer, this thing, an alarm clock going off, is a thing that is in no sense taken to be ordinary. Yet, for the local inhabitant of the setting who was being observed, competent in its ways and means and the significance of the different things that occur within it, it is not worthy of comment or even a pause to suggest it might merit any reflection. It is disregarded, ignored. This is not offered up as a motivational account of some kind. What does disregarding or ignoring look like such that you might say that is what is going on anyway? It looks just like it did here for the ethnographer. It looks like some feature or course of action that you would ordinarily take to be somehow open to remark or interest that is, by some other party, not visibly taken to be that way. This is what gives the licence to observe that it is being disregarded or ignored. How could one reasonably say someone was ignoring someone else if they looked at them, even out of the corner of their eye, or manifestly engaged with some utterance they had made? How could one have said Lucie ignored the alarm going off if she even briefly stopped mid-sentence when it happened?

Uncovering the Unremarkable

59

What is revealed rather nicely by what happened in this particular case is that the recognizability of what is open to remark is differentially organised according to membership. This is why for an ethnographer and an inhabitant of the household a ringing alarm can be two different things. The locally organised character of competence is everything in this respect. We routinely take features of our surroundings to be both remarkable and unremarkable. This is organised around our presumption of what things are ordinarily worthy of our interest. Presumption here is powerful for we cannot be forever going around and having to make a decision about what features of the world we should be paying special attention to. However, when we enter some new setting, what could possibly make it recognisable that something is being taken to be beyond remark? How, in other words, does some new setting instruct us in how to understand its organisation in these terms? One way to tackle this is to think of planes and turbulence and the differential between the first-time flyer and the seasoned traveller. When turbulence happens do first-time flyers routinely panic and cry out “What’s going on? Are we going to crash?” when everyone around is still focused on their newspaper and hasn’t even looked up? So recognition of what is to be taken to be unremarkable is a member’s competence, a vulgar competence as Garfinkel put it (Garfinkel, 1967). In that case what is to be taken to be unremarkable has some measure of cohort dependency, but the competence to recognise that things are being taken to be ordinary and unremarkable is not specialised in any way. And, furthermore, only certain people, for example, ethnographers, children, idiots, foreigners, and so on, have licence to inquire into the matter further. Anyone else is displaying a lack of competence that is itself open to inquiry and the reasoning out of an account by the ordinary inhabitant, for instance, “oh, they must be really scared of flying” (or, more worryingly for airlines handling first-time flyers, “perhaps there really is something wrong after all”). Another point to grasp here is that “remarking” is not about some physical utterance. It is about rendering one’s interest or noticing of some feature visible in some way to others around you. And this manifest noticing is something other people may reason about and attempt to account for. So, to display manifestly special interest in some obviously ordinary activity is not something that can easily slip past without comment – for instance, staring at someone who is buttering a piece of bread. Should someone do this there is an expectation that the interest shown will be accompanied by some kind of account, for example, “What an odd way to hold a knife. Do you always butter it like that?” So we can see through the simple example of an alarm going off that the world is actually replete with devices and engagements with those devices

60

Making Work Visible

that, to all intents and purposes, render them “invisible in use.” It is their ordinariness that makes them invisible in that way, often they are an integral feature of routines, but what accomplishes their invisibility as a matter of method is that they are taken to be unremarkable, with any other orientation to them itself being open to account. In the spirit of many ethnomethodological endeavours, this focus upon methodological, rational, reasonable, accountable, and accomplished features of real human action amounts to a “foundational respecification” (Garfinkel, 1991) of what “invisibility in use” might amount to (see also Sharrock and Button in this volume). And clearly it does not have conventional views of perception at its heart.

Uncovering the Unremarkable Having recounted some of the original insights discussed in the paper titled “Unremarkable Computing” the remainder of this chapter is going to be devoted to how it could be that insights of this order were made open to discovery and deemed pertinent in the trajectory of work undertaken by work practice researchers in XRCE. In particular we shall use another example to elaborate the different orders of concern that can be broached through uncovering the unremarkable and how these might be of broad value to both work practice researchers themselves and the variety of enterprises routinely agglomerated under the heading of “design.” XRCE Studies of SOHO Environments In 2000 and 2001 Xerox Research Centre Europe took as one of its prime foci for research the work conducted in small offices and home offices (SOHO). This interest arose out of the constitution of a research group specifically dedicated to an understanding of “Emerging Office Environments” (EOE) and their import for design. From the moment of its conception, EOE was motivated to take what one might call a skeptical eye to some of the more prevalent hyperbole surrounding how the character of modern work and modern workplaces was changing. At the time there had been (if the media were to be believed) a muchtrumpeted move in many sectors towards home-working. This made homework an interesting target for us to explore. As a result, I conducted a number of studies in people’s homes. The data gathered from these studies was to provide the observational basis for a range of reflections about people’s work practices in the home environment and how these were adapted to the larger round of life in the home. Of course, Xerox was far

Uncovering the Unremarkable

61

from being the only organisation interested in getting better purchase on what was happening in the area of home-working or “teleworking” as it came to be called. In the UK, 2000 saw a more proactive move towards the rollout of broadband. British Telecom (BT) had established a panel of triallists for ADSL at home (ADSL is a type of broadband technology) and were keen to get a sense of how broadband connected up with actual visible courses of action in home settings. Many of the triallists, unsurprisingly, were home businesses. Studies of home-working at XRCE were therefore also coordinated with research efforts at BT. In exchange for access to ADSL triallists to be used as fieldsites for observation, data gathered from these studies were shared with BT. This opened up a much broader potential pool for studies of people working at home. Serendipitously, and of considerable importance for how studies of domestic environments began to take shape at XRCE, at the same time as these developments were taking place within the main thrust of businessdirected research, the lab also received funding under the Disappearing Computer call (see previous) of the Future and Emerging Technologies arm of the European Commission’s IST (Information Societies Technology) Programme (IST 2000 26360). This funding was for a project titled “Multiple Intimate Media Environments” or “MIME” (MIME, 2001). The project commenced at the beginning of January 2001 and was completed at the end of March 2002 and was conducted in partnership with the University of Nottingham and Philips Design (who led the consortium). At XRCE, the principal interest to be drawn from the project related to what might be offered for grounded innovation through various kinds of coordination between ethnography and design (see both Suchman and Sharrock & Button in this volume regarding the importance of this to Xerox research). This was fuelled by the opportunity to engage with industrial designers at Philips, which was seen at the time to be a potential way of expanding the applicability of ethnography to completely new kinds of design enterprise. In this context, some initial reflection was made upon existing studies of domestic environments and then new studies were undertaken. In addition, work was conducted to complement the innovative process and to explore the methodological issues at a deeper level. One of the main outcomes of the project was that it provided the grounds for engaging in three wholly new detailed studies of family life where the objective was simply to follow family members around their homes as they went about their ordinary everyday lives. Thus the focus of domestic studies in XRCE shifted from just work in the home to “domestic life in the round,” with all of its attendant dynamism and diversity.

62

Making Work Visible

All of the studies referred to were what has come to be termed ethnomethodologically informed ethnographies (Hughes et al., 1994b) in the Human-Computer Interaction (HCI) and Computer-Supported Cooperative Work (CSCW) communities (see also both Sharrock & Button and O’Neill et al., this volume). This has long been the prevalent mode for conducting field studies at XRCE. Indeed, it was derived from the seminal understanding of work practice studies articulated by Lucy Suchman at Xerox PARC (Suchman, 1983) that had preceded the establishment of XRCE (see Szymanski & Whalen and Suchman in this volume). A strong part of this approach is the immersion of the fieldworker in the setting over a period of time sufficient to be able to acquire some understanding of what it might take to reason as a member of that setting. Of equal importance to the work of immersion, however, is the application of an analytic perspective growing out of ethnomethodology (Garfinkel, 1967) (Sharrock & Button, this volume). The ethnomethodological aspect of the work encourages one to work through the data and tease out all of the organisational features of embodied action and interaction, the associated reasoning and accountability, and, in short, all of the concrete features that provide for the accomplishment of whatever is being done. This perspective is critical to the story of uncovering an important insight here. The foundational respecification at the heart of unremarkable computing could not have been arrived at without it. In order to elaborate the connection between methodological approach, analytic insight, the perspicuous character of investigations in home environments, the scope these offer for informing design, and how the constitution of research teams in XRCE at that time helped to facilitate the study to design trajectory, I am going to make use of another example we used in the original paper on unremarkable computing. The Knock on the Door As a part of the studies being conducted under the auspices of the MIME project I had been following the everyday routine of one particular family for some time. There were certain features of this family’s routines that had piqued our interest. One of the things that had got us especially excited about this family was a certain arrangement the mother had set up with her next-door neighbour involving knocking on each other’s doors. Both the mother in this family and her next-door neighbour had children at the same school and would be setting out to pick up their children from school at the same time each day. The school was close enough that they nearly always walked.

Uncovering the Unremarkable

63

The first time I saw the knock on the door happen it was about three o’clock in the afternoon. The mother I was watching had been sitting out in her garden reading in the sunshine. She looked at her watch then went into her house and locked her back door and started going around shutting windows. Then she headed up her hallway towards the living room and at that moment there was a knock on the door. When this happened she opened the door a fraction but then, instead of opening the door properly, she went into her living room, continuing to gather bits and pieces together. When she finally did go out of the door the person who had knocked on it, her next-door neighbour, was already walking off up the road. Now, of course, this happening aroused my curiosity, because it’s rather unusual for someone to knock on a door and walk away without waiting for an answer. Typically knocks can achieve a number of things such as being a summons to the people inside, or a way of checking if some room is empty. Walking away without waiting for someone to answer a knock on a door is therefore something that is typically considered to be rather rude. In fact, there is a game of dare played by British children called “knock down ginger” where this is just how the game proceeds with the goal being to get away unseen and uncaught by the irritated householder who has been needlessly brought to their door. At the same time, it’s pretty unusual to answer a knock by only partly opening the door and then walking away. This certainly isn’t the way people normally deal with a summons because it offers no scope for engagement with the person who has summoned you. As the study progressed it became clear this was no chance happening but something systematic because it happened day after day in the same fashion. It transpired that neither of them had ever discussed this arrangement but had just fallen into it as a way of telling one another they were setting off to school now, so that they might walk together rather than separately. In these circumstances then, the knock on the door was just enough to say “I’m about to leave,” and the half opening of the door was just enough to acknowledge that announcement of imminent departure. They had honed it all down to the barest minimum so that, with beautiful economy, they could bring about a particular coordinated routine between their respective households. And within this they understood each other’s accountability such that they didn’t need to explain why they’d walked off after knocking on the door, or only half-opened it. Now it needs to be recognised that this arrangement only applies at around that time of day on a school day. It doesn’t take much to see that,

64

Making Work Visible

were they to do the same at some other time of day or at the weekend they would be quizzed as to why they should do such a thing. Indeed, it was perfectly visible over the course of the study that, for the mother I was watching, a knock on her front door was taken quite expressly to amount to a summons and she always opened her door fully to see who might be there. So one can see how this knock on the door activity is only a thing that is intelligible at a specific time on specific days and that they are mutually oriented to this local and precise intelligibility. Furthermore, and in view of how this paper is directed to explicating a certain kind of ethnographic sensibility and its methodological implications, it should be noted that this orientation was most especially seeable for someone else like me as I watched them because not once did they pause to remark upon the oddness of it all or problematise it in any kind of way. Yet it was wholly remarkable to me because it breached all of the ordinary ways in which I might have expected knocks on doors to be handled. In fact, the only time the parties involved saw fit to comment upon the practice in any way was the time I set about trying to capture it on video. The problem, of course, was that I had always been inside the house with just one of the two parties when this thing happened and I wanted to capture film of it so that I could show my colleagues the phenomenon “in the flesh” so to speak. It is a testimony to the foolhardiness of ethnographers that I thought I could simply drive one and a half hours down the road and park outside their houses on a school day before three o’clock and thereby capture it. Nonetheless, that was what I set out to do. It is not until you are sitting over the road from someone’s house, filming it through a car window, that you start to reflect upon just who might be engaged in the more remarkable behaviour. Clearly, for everyone else passing by I was an object of suspicion. Worse still, as I sat there waiting I started to be plagued by nagging doubts. “Surely,” I reasoned, “when one considers the odds, there just have to be occasions when rather than knocking on one another’s doors they both come out of their houses at the same time.” Furthermore, this was now the last day before the schools’ summer holidays so quite probably my only chance. Of course, as you can see in Figure 3.3, this was indeed one of those occasions where they both exited their houses simultaneously. As the mother I had been observing came down the path she commented to her neighbour “That was good timing.” This was the only time they ever said a word about what they were doing. And this time something quite different had occurred. However, as I recovered from my inevitable disappointment I began to see that what I’d captured was, if anything, even more fortuitous.

Uncovering the Unremarkable

65

Figure 3.3 Leaving to walk to school, “That was good timing”

I realised that the remark “that was good timing” is not a remark upon the routine itself but rather upon the perfection of its realisation. The beauty of it was that, in these circumstances, the very need for the knock on the door had simply faded away and one could see that it was never simply about knocking on one another’s doors at all. Rather, the knock on the door was a resource to bring off what they were really after all along, which is to walk to school together rather than alone. What Does It Tell Us? From Methodological Approach to Analytic Insight In the original paper on unremarkable computing we were at pains to use this example to elaborate the importance of seeing how even the most manifestly out-of-the-ordinary matters when considered from the point of view of what just anyone anywhere might know, can blend into everyday practice to the point of being wholly unremarkable. This highly specialised but completely intelligible phenomenon is a powerful demonstration of what Weiser was driving at with the notion of invisible-in-use. However, here I want to do a slightly different job with this example. What I want to ask is how could one move from the position of observing some particular piece of routine practice to seeing the scope this might have for saying something important about a design principle? I want to start with a reminder of the observation I made when describing the approach of ethnographers at XRCE to their investigations of the real world. It was observed that, in the context of ethnomethodologically informed ethnography, the goal is to “acquire some understanding of what it might take to reason as a member of [the] setting.” Now clearly a central part of being able to reason as a member might reason is to take for granted what

66

Making Work Visible

a member might take for granted. In fact, members make a plethora of assumptions about the world. Most of what members do is never opened up to account or remark in any kind of way. It could, of course, be otherwise, but to imagine a world where one could take nothing for granted but always treat everything as though it was the first encounter and had to be grasped as though it were the first time through is to imagine a world very different to the one we inhabit. On the contrary, one can see that an important characteristic of the social world having the orderly characteristics it does have is that, for the larger part it can be taken for granted. The term used in ethnomethodology for this taken-for-granted orderliness to the world is that the world as encountered is taken for the larger part to be “naturally accountable” (Garfinkel, 1967). The term recognises that we all have some sense of what is a usual or unusual way of proceeding for any course of action, that most of what we do passes by without remark, and that some account might be needed is not even reflected upon at the time. If pushed, however, we would account for these matters as being in the ordinary way of things. In other words they are taken to be “natural.” This naturally accountable order to the world is clearly a vital part of what one must unpack to be able to see how the world according to a member is organised. But unpacking it is not necessarily so straightforward. To do this one must effectively take what is encountered as wholly unremarkable and open it up for remark. In the case of the knock on the door this was not in fact so very difficult because, to the ethnographer, the phenomenon already breached ordinary practice so it was no great stretch to look at it in this light. But for the larger part of what an ethnographer might witness, things are otherwise. Much of what one sees has the appearance of wholly ordinary, unremarkable conduct. And, as one acquires membership, the sense of it being naturally accountable just grows. The problem that confronts us here is that, as so much turns upon being able to unpack the taken-for-granted character of situated action, it is important to find ways in which, as an analyst, one might accomplish that. Garfinkel’s choice of phrase here was making things “anthropologically strange” (Garfinkel, 1967). This obviously reflects the spirit of the enterprise, but for many analysts this is not necessarily enough. Uncovering the unremarkable is one way of going about doing this. It amounts to taking a course of action and quite deliberately bringing to account all the things no one in the setting would ordinarily bring to account within that course of action. In this way one begins to unpack just what is in plain view. So, as an ethnomethodologically informed ethnographer, the first part of one’s enterprise when one enters a setting is to acquire some capacity to reason as

Uncovering the Unremarkable

67

a member of that setting. This amounts to being able to naturally account for the phenomena of that setting in the same kinds of ways as others in the setting might account for them, for instance, seeing that when someone knocks on your door at around 3 o’clock in the afternoon on a school day you need do no more than open the door a crack and continue about your business without looking to see who is there because that is enough to acknowledge that you have heard the knock and why would you need to do more? Indeed, to open the door fully, having got this kind of arrangement going, would be one of the things one might expressly have to account for, for example, “Oh, I was hoping to catch you. I’m not walking in today but wanted to ask you if you could pick Sally up for me tomorrow.” However, the second part of one’s enterprise is to then be able to explicate for others what one’s competence as a member of the setting consists in. To do this, one needs to be able to speak of how one would reason about any particular course of action as a member, and what one would do methodologically to make that reasoning manifest or embodied in one’s actions. One way of approaching this is to ask of oneself, as a member, what would make each feature of what is being engaged in remarkable to those around you? One can also ask of oneself, just why would it be open to remark? These two questions together assist in unpacking: a) the methodological character of one’s actions, that is, what one needs to do to preserve the natural accountability of one’s actions so that they are just seen as what anyone would properly do; and b) the reasoning one must engage in as one undertakes those actions, that is, what you would presume to be an ordinary way of doing those things so that no-one else would think to remark upon them and what you would presume of others if they were doing the same kind of thing. There are some specific aspects of any course of action one can probe that will, in the course of one’s analysis, assist further in teasing out just what is going on here: Embodiment: What features of how people physically comport themselves will preserve the unremarkable character of their actions? How would they need to physically position themselves? How would they need to position and orient themselves in relation to others in the setting? What gestures, ways of looking, postures, etc would make their actions open to account and what does that tell you of how things must be done to avoid that order of accountability? So, with the knock on the door, notice how the neighbour does not wait at the door but walks away. If she stood waiting at the door, doing what a caller does

68

Making Work Visible instead of what someone doing this walking to school together activity would do (where summoning someone to directly talk to them is not what it’s about) what kind of account would she have to provide for just standing there? Talk: What features of what people say and, just as importantly, what features of how they say it will obviate the possibility of remark or the need for account? It is also sometimes important to reflect upon where things might be said that have not been said. So, in the case of the knock on the door the mother I was watching did not call out as she opened the door “Yes, who is it?” and the neighbour does not ask as the mother joins her on the street “Didn’t you hear me knocking? Where were you?” Both of these things might be reasonable in other courses of action related to knocks on doors. In this case, then, we can see an aspect of what is being accomplished and how one must reason about it is that one doesn’t ask who is there because it should be obvious and one doesn’t presume the person whose door you have knocked on will appear immediately. Indeed, it is an ordinary part of the arrangement here that the other person may not appear at all. Which is another reason why just standing there and waiting would be remarkable. Physical Action: What features of what people physically do to or with the objects and people around them are unremarkable in this particular course of affairs? In what ways might they become remarkable? And in what ways might just how they are done become remarkable? In the knock on the door example, a critical part of what happens is that the door is opened just a crack. The door thereby becomes instrumental in providing acknowledgement of the knock. One can see how, were one not to provide this acknowledgement the other party might well, instead of just meandering up the street, allowing the other one to catch up, set off purposefully and not look back. So, to not provide this acknowledgement when one is there becomes open to remark. If one were there and then subsequently caught up with the other person who had struck up their ordinary pace they could well remark “Oh, didn’t you hear me?” or you might yourself say, as you caught up, “Sorry about that. I was out in the back garden and hadn’t noticed the time.” So, the slight opening of the door is methodologically coherent, implicative, and unremarkable in this course of action. Yet, as we discussed previously, it would be an altogether remarkable way of handling an ordinary summons.

Obviously, this is only beginning to scratch the surface of analytic possibility here. One can move on to other matters such as collaboration and how

Uncovering the Unremarkable

69

particular features of how courses of action are collaboratively achieved might make them open to remark or otherwise. Above and beyond all of this one needs to look at what is being accomplished in the course of doing just this particular set of things in just this particular way so that no one would ever think to comment upon them. The backdrop to unremarkability in the case of the knock on the door is that it is all understood to be about coordinating their actions so that they can walk to school together. Were it about asking if you could borrow some sugar, many aspects of what both parties are doing would become instantly subject to remark. What I have begun to elaborate here is how an attention to just what makes specific features of some course of action unremarkable is informative for explicating the methodological character of action and associated reasoning. And, to turn things back full circle, being able to see just what is or isn’t unremarkable about any state of affairs is itself dependent upon one having done the job of acquiring a member’s competence for that setting. In the final part of this paper I shall look at how this approach is also informative for design. From Analytic Insight to Design To tackle the relevance of the analytic approach I have outlined to design and to think afresh about design principles, I want to briefly review once more some elements of the knock on the door example. Some things to observe are: someone knocks on a door; knocks on doors can serve as summons; opening a door can serve as an acknowledgement; knocks on doors and partial openings of doors can serve to coordinate a specific routine (in this case walking to school together). I have chosen to focus here upon the door because it is what happens with and around the door that is of such moment for being able to recognize what is going on and for being able to understand what is being accomplished. Although it might not be interesting in itself, we can see that the door and what happens with it offers a physical interface and activities relating to that interface that are instrumental in some fashion. Thus the door and what happens with it offers up the kind of space a designer might be interested in. The question I want to pose here is: where should that interest be pitched? One can, to be trite, design a means of knocking on doors. That is, one can provide a surface and a means for people to interact with it that can make it distinct from its normal state in some way such that it could have attention drawn to it. Then one can design a means by which something can, as well as having attention drawn to it, actively serve as a summons. That is as well as having

70

Making Work Visible

the capacity to change state when one party acts upon it, that change of state can be broadcast to others in such a way as to make it clear that there is someone there changing the state of the object who wants your attention. Some things to note about that are that you don’t just want attention drawn to the object but to offer the means to reason that the change in state of the object is the consequence of someone wanting to talk to you. One could also recognize certain other features of how summons like this operate. For instance, there is a certain sense of elapsed time that one is prepared to wait before repeating the action. Furthermore, those who are being summoned can be called to account for any failure to respond. Beyond this, one might want to design this thing in such a way as to allow for the possibility that people could provide a response to the summons that is richer than just making themselves fully available. That is, they might do something comparable to opening the door a crack rather than fully. This could serve the purposes of a wider grammar of action such that one can do more than just respond but do things such as acknowledgment. Differential means of engaging with the object might also allow for more refinement in how the initial action could be understood. The knock on the door in the original example is not so much a direct summons as a message, and that is one of the reasons why acknowledgment can be enough. However, it’s hard to make distinct as a person doing the knocking that you are knocking as a message as opposed to knocking as a summons, although just how one knocks can have some impact here. One can just tap on a door or one can hammer on it and a whole range of different degrees of knocking inbetween. Tapping on a door is enough for the kind of message we see in our own example, but I need you to answer the door right now is not best accomplished in that way. Redesign could offer further tuning of such recognisability. Finally, one might want to pitch one’s design at the level where you recognize that all of the features of what happens with the door are bound up with the specific understanding of this being about coordinating a routine. The beauty of a real door is that it is able to do all of those things listed, be used to coordinate a routine, and still serve as a point of entry and exit to a building, a physical barrier to would-be intruders, a place to pin notes or place numbers or names that might help the postman to recognize your building, something that through the very sound of it can tell you that your partner’s just come home from work, and so on. An understanding of what doors might be about in this way, and the place of knocking as a means to articulate a specific course of action within all of that, adds up to a very rich design space indeed. And one can see how, once one sees doors in this light and how

Uncovering the Unremarkable

71

generic assumptions about their use can be subverted to highly local ends which would be remarkable if one didn’t have the membership to recognize their appropriateness, one might begin to reflect more deeply upon too literal an understanding of design principles such as “invisible in use.” Design as a Setting for Ethnographic Work Two further things remain to be said here. The first of these is that, for designers to have access to the scope of possibility outlined above there is a need for someone to have both gone out to acquire membership of some setting and then to have come back with the means to explicate just what reasoning as a member might turn upon. This was built into the model at XRCE. Without ready-access to ethnographers, designers should consider going out and looking at the actual organization of the everyday world for themselves. The other aspect of the model in operation in XRCE was that teams were deliberately constituted of both designers and ethnographers who could work together on a day-to-day basis (see also O’Neill et al. and Colombino et al. in this volume). For the purposes of ongoing informed and iterative design, nothing could be better. It facilitated the transition from analytic insights about the ordering of the social world to insights about the potential orientation of design. A whole series of designs related to bridging the paper-digital divide grew out of the collaboration between ethnographers and designers at XRCE during the period we have discussed in this paper, and it underpinned the design recommendations made in the MIME project. The continual work association also meant that ethnographers could become sensitized to the practical concerns of design, local design interests, and the kinds of design principles to which ethnography might be able to offer further insight. The potential here is exemplified in Button and Dourish’s 1996 paper on “Technomethodology” and a later discussion of what might constitute the “adequacy” of an ethnographic output for design purposes (Diggins and Tolmie, 2003) (see also Sharrock & Button in this volume). There is much talk in the latter parts of this paper of design. One final thing that should be emphasized is that “design” here is not considered to be all of a piece. Design is a gloss for a whole gamut of interests from infrastructures and middleware to industrial design. The principle in play here is that design across all of these interests can be significantly enriched and more finely-tuned to what people do if it is informed by an understanding of how people actually organize their day-to-day affairs. Even where “unremarkability” is not required this can still matter. Some designers, by nature of their enterprise, are actively concerned with trying to render technologies

72

Making Work Visible

both provocative and remarkable (see Gaver et al., 2004 for an example of just such an orientation). Nonetheless, it is much easier to see what will breach local practice and encourage open comment if one understands what practices one is breaching in the first place. Furthermore, there is a whole community of skilled individuals beyond technologists who have an equal role to play in design. Here, though, the design in question is the design of organizational solutions rather than technological ones (see Whalen & Whalen, Vinkhuyzen, and Vinkhuyzen & Ikeya in this volume for an indication of just how important this can be). Solution architects, implementation managers, business process managers, change management consultants, and indeed everyone who might wish to bring about some kind of organizational refinement or change could also stand to gain deeper insight and more effectively tailored solutions through a consideration of the kinds of inquiries into local practice I have been discussing here.

Conclusion This chapter has looked at how a number of factors, including the direction given by Weiser to ubiquitous computing through the notion of “invisibility-in-use,” the evolution of a European initiative around the idea of the “disappearing computer,” a move in XRCE towards studying domestic environments, and the routine embedding of ethnographers and ethnomethodologists in design teams at XRCE, together worked to prompt important new insights regarding how design principles such as “invisibilityin-use” might be best conceived. The chapter has also elaborated the ways in which the notion of “unremarkability” has wider purchase for grasping the orientation ethnomethodologists adopt towards analysing ethnographic data. Beyond this, it has indicated ways in which design may also learn from these approaches. Ultimately uncovering the unremarkable involves serious practical and intellectual endeavour and it has deep ramifications for both the conduct of ethnomethodological analysis and how ethnomethodology might speak to design. In one of his earliest articulations of ethnomethodology and its program, Harold Garfinkel, discussing the practical analyzability of the practical organisational accomplishments of the members of any particular setting, observed that: For members doing sociology, to make that accomplishment a topic of practical sociological inquiry seems unavoidably to require that they treat the rational properties of practical activities as “anthropologically strange.” (Garfinkel 1967, p 9)

Uncovering the Unremarkable

73

It is to this that the idea of uncovering the unremarkable is specifically dedicated and I have begun here to outline ways in which ethnographers working at XRCE in the early years of this century sought to practically manage the problem of acquiring both membership and anthropological strangeness at the same time. These concerns might seem contradictory at first but, as I have elaborated, they are actually tightly interconnected. Without both membership and the capacity to render its taken-for-granted aspects anthropologically strange, it is hard to bring the reasoning and methods that provide for the character of any setting into plain sight.

4

Work Practices to Understand the Implications of Nascent Technology Francoise Brun-Cottan and Patricia Wall

In 1992 social scientists and industrial and interface designers in Rochester, New York were presented with a unique opportunity to conduct a work practice study intended to identify and prioritize functionalities in a nascent technology. The place was the communications department of a multinational industrial mixer manufacturer and the technology was scanning. Scanning? It held great promise, but there was little clarity about which features to develop first and how to build use cases, for example, stories, we could tell customers that would demonstrate why they needed to spend money and migrate from current equipment and update their existing ways of working. In other words, what did scanning bring to the party? Tangentially, but tellingly, this project also provides an example of ways in which the rich descriptions of work ethnographic inquiry differ from the possibly accurate but essentially inadequate understandings represented in standard “work flow.” Fieldwork began at Sterling Manufacturers,1 based in upstate New York. Sterling, with satellite offices elsewhere in New York and field offices in 60 countries, made large, customized construction machinery. A write-up in their company’s “World News” quarterly letter describes the project as it was presented to the world. Since December of 1992 a “partnershipping” agreement has been going on between Sterling’s Communications Department and Xerox’s New Products/Systems Engineering Group. Xerox was in search of a manufacturing company who created, printed, stored, distributed, and retrieved a wide variety of documents using a combination of manual and automated systems. Sterling’s Communication Department was seeking professional assistance to develop and implement more effective ways to perform the document services that they are responsible for. “We are very pleased to 1

74

Actual names of study participants have been changed.

Work Practices to Understand the Implications of Nascent Technology

75

have been considered by Xerox. The major factor for our selection was the sincere desire on the part of all of the communications staff to improve our management of documents,” observed JM, Manager of Communications. “And, improving the efficiency and quality of how we handle our documents is a goal we all can benefit from.”

Understanding how Sterling became a “work practice” fieldsite proves interesting from a historical point of view. After all, Sterling was an existing Xerox customer and had been visited previously by Marketing and Sales personnel.

A Brief Historical Perspective In the mid to late 1980’s a few Industrial and User Interface Designers in Xerox’s Industrial Design and Human Interface (IDHI) center became increasingly interested in how they might engage more directly, and at an earlier stage, with people who would be using the devices and interfaces they were designing. They suspected that early, direct interaction with user/ customers would provide user-centered understandings of ways in which people understood their work, and the role device functionality and design might play in early adoption of innovative technologies. As well, they believed that insights gained from understandings of workplaces in which technologies were to be placed would be of interest to engineers and others concerned with developing those technologies. Another aspect, from the designers’ perspective, was that the technology was becoming too complex to be studied in a lab, out of the context of the work places and activities in which the technology would be used. Meetings and discussion with people from Palo Alto Research Center (PARC) –most especially Anthropologists Jeannette Blomberg and Lucy Suchman regarding ethnography and Austin Henderson regarding humancomputer interaction (HCI) – bolstered the designers’ resolve to incorporate anthropological perspectives and ethnographic techniques in their design process. (See Suchman, Chapter 1, for a retrospective on ethnographically based research as it evolved at PARC, the source of inspiration and support for much of the work in Rochester.) In 1992, Francoise Brun-Cottan, an anthropologist working with Lucy Suchman in the System’s Science Lab at PARC, went to join Pat Wall’s group in the IDHI group in Rochester for a year’s experiment in folding ethnographic methods with design processes. Even at the time it was hoped that symbiosis would prove viral; this was hardly a given. Indeed, at that time at Xerox and other large technology manufacturers, neither Design nor Ethnography were held in the high regard they

76

Making Work Visible

enjoy currently. (See Watts-Perotti et al., Chapter 6, which discusses some approaches for garnering support for ethnography within the company.) Designers – Industrial Design, Human Interface, and so on – were continually being asked to prove their value, to demonstrate their return on investment. Such demonstrations were hard to provide when there was neither time nor money for alternate versions of devices and designs to be built and tested in actual user workplaces. Requests for additional resources, it was pointed out, would have to be taken from some other department’s budget. Moreover, assertions from other departments that customers would not be receptive to, or allow, intensive fieldwork to be conducted on their work sites needed to be overcome. Hardly least came the need to address a perennial question and challenge: Could findings specific to one fieldsite be generalized or applied in a useful way – if at all – across a number of different work environments in which devices had to function in order to be economically viable? Nevertheless, Pat Wall and a small band of designers persisted, and PARC facilitated and the union was made. The anthropologist ended up staying for nearly ten years and the group that evolved, moved organizationally from IDHI (Industrial Design Human Interface) to PARC Research based in Rochester. We became a multidisciplinary team of work practice practitioners composed of anthropologists, designers, and computer scientists who sought to engage partnerships with other organizations and departments throughout Xerox. Sterling provided the initial project and fieldsite for the newly formed five member IDHI team of designers and the new anthropologist. Observing and presenting what were going to be the unique insights that it was claimed this ethnographic approach might provide about existing work practices around records creation and management in the Communications Department was not the only, perhaps not even the central, task at hand. The project was meant to discover information about document and records’ keeping practices that could prove useful to prioritizing functionalities of a new device – code named the Document Machine – which could revolutionize the creation, duplication, manipulation, distribution, and archiving of electronic as well as traditional paper-based documents. It would, in fact, help revolutionize the very concept of “document.”

Methodology WORKSHOPS: We had initial workshops with management and staff at Sterling in which we told them: what we thought we were on about, what

Work Practices to Understand the Implications of Nascent Technology

77

previous studies had unearthed, how we were going to proceed, what we hoped to address, and how we would report back to them. We had midway and follow-up reports in which we described what we had done and areas in which existing and forthcoming technologies might play. We told them they were the experts and we were there trying to discover, understand, analyze, and represent their understandings of their work and their ways of accomplishing it. Yes, we hoped to be able to assist them in assessing the value of state of the art and future technologies for their enterprise, but by participating in the project, they were going to get to inform how some of those technologies would address “their” issues. When we described some of the techniques we would be using and just how embedded we would be, the reception by staff members was civil but less than enthusiastic, and in some cases decidedly unfriendly. Mention of audiotaping and videotaping people doing their work did not appear to much buttress our assurances that we were not going to provide secret personal efficiency evaluations to management. That we would not was indeed the understanding our group had with Sterling. It was only in more informal, that is, manager absent, meetings that we were able to gain some trust on this issue. Some people would try the “catch me if you can avoidance tactic,” others the basic “monosyllabic response to questions” tactic. Yet after a bit, after their seeing and hearing from those who initially had been more open or comfortable about being interviewed or observed, people clearly relaxed and more freely voiced opinions about work activities: difficulties, obstacles, work-arounds, providing golden nuggets for us to pursue. It was a source of some pleasure for us that on the last day of the project the one person who had absolutely refused to be recorded in any way expressed her regret at not having been audiotaped.

Toolkit and Techniques Basically it would not be much of an exaggeration to say we used every technique of our fledgling methodological tool box. (It should be remembered that team members had other work obligations, Sterling was not our single focus.) By the end of fieldwork over a period of weeks we had conducted twenty six hours of audio and video recordings, eleven interviews, ten observations and shadowings, and five group meetings with study participants. We drew office maps to get and preserve a sense of peoples’ spatial relationships. We made lists of who was doing what with which technologies and then drew charts showing how that played out by particular activities. These would help us identify intersections, bottlenecks and common problems across processes.

78

Making Work Visible

Figure 4.1 Tracking a file. Requests to find, retrieve, and deliver envelopes from storage are received by the file room staff. This illustration shows the variety of ways the request can be interpreted and the various information sources the file room staff may have to search to track down the requested envelope

The observation, interviewing, and shadowing techniques employed had a solid history in anthropological praxis (and in other disciplines’ practices as well) and were intended to gather information in such a way that it could then be analyzed, described, and represented. The maps and lists and charts were intended to capture relationships: among spaces, technologies, and peoples’ activities. We tried to construct an overall sense of place and activities and then zoom in tighter on details. Maps, plans, tables, lists, chronologies (in our case, the course of information movement through activities from inception to termination) had, of course, been staples of ethnographic fieldwork in Anthropology since well before “We, the Tikopia.” Members of our team, with expertise in human factors, and industrial, graphics and interface design, brought an experienced orientation to visualization. This provided at least three benefits. We were better able to preserve and visualize spaces and the trajectory of documents. Documents, photographs, maps that were pinned to walls and white spaces provided and

Work Practices to Understand the Implications of Nascent Technology

79

Figure 4.2 A map of the workspace provides a foundation to capture and illustrate observations about the spatial and organizational elements required to locate and track a file

refreshed memories of pathways and activities “at a glance.” We were able to easily take representations; hand drawn or printed images on paper, back to informants for verification or corrections. Informants seemed to enjoy correcting us; they certainly enjoyed being consulted about whether or not we had correct understandings. Such corrections often allow further elaborations as people realize they really are being considered the experts about their work; they were encouraged to provide them. As well, “verifying” allowed them to see what we were actually looking at. That we showed them what we were focusing on, as well as what we were trying to understand, in itself tempered their misgivings. Not infrequently there was some amazement, “You want to know more about that?” when whatever “that” was seemed so eminently, trivially, obvious to them. Silly researchers. But at least there was reassurance in the belief that no manager was going to get embroiled in examining activities at the fine levels of granularity we were exploring. Additionally, we would be able to use the representations to help

80

Making Work Visible

provide some contextual grounding when we worked with people outside the group – engineers, other designers, business groups.2 As well, it should be noted that these early visualizations by the IDHI designers would develop and bloom into sophisticated yet straightforward tools such as the Work Practice Graphics Toolkit (aka “the blue guys”) that became familiar throughout the company. (Examples from the toolkit can be seen in illustrations in Chapters 7, 8, and 15.) Cameras, tape recorders, movie/video recorders, had also long been part of the ethnographic toolkit. As those technologies developed they provided better recordings while becoming smaller and more portable, enabling increased use in more environments. Simultaneously, they facilitated the development of more refined, empirically based analytic methods for gaining understandings of the in-context, coconstruction of meaning by participants in work activities as well as in quotidian life, which constituted basic precepts of the disciplines of Ethnomethodology and Conversation and Interaction Analysis as well as Anthropology. (See Sharrock and Button, Chapter 2 for an in depth examination of these precepts and their application by particularly adept practitioners.) Repeated viewing of audiotapes and videotapes by team members – as individuals and as a group – enabled identifying what we thought might be important key sequences and activities. (These coviewings followed the model started at PARC and at IRL, the Institute for Research and Learning in Palo Alto.) However much we would have liked to make detailed and precise transcriptions, we generally had to content ourselves with logging the tapes, noting topics, and pinpointing activities. Time stamps allowed us to retrieve sequences which could be reviewed by the group or more closely analyzed. As maps of document and information management trajectories were being formulated, it was not uncommon for gaps to appear. Consulting tapes often allowed us to reconstruct how and why something moved from point A to point B. And if they did not then it became clear we were missing something and that participants needed to provide more input. Though we did not get to do Participatory Design sessions3 in this project we achieved something close to iterative design prototyping sessions via a combination of 2

3

We would not then realize how useful these representations would be across time to us, and members of other projects, to refresh memories and allow cross site/cross industry/ cross technology comparisons. Participatory design is an inclusive approach which advocates bringing together technology or services developer stake holders: researchers, engineers, designers, and so forth, with end users in a creative process. (See Schuler and Namioka, 1993.).

Work Practices to Understand the Implications of Nascent Technology

81

confirmations and “what ifs” where the participants provided preferences in ways in which they would like the work facilitated. These, in turn, spurred our own imaginings tied more closely to identifying opportunities for ways the applications, the very functionalities, we were trying to develop could be of service. At the same time that we were engaged in designing fieldwork and analysis techniques and methods, we were simultaneously devising how we could label, catalogue, view, transcribe, share, and analyze what we had with one another. These efforts straddled the divide between the most mundane – if essential – that is, labeling tapes and notes, to the divine, that is, learning how to see. In his book “The Back of the Napkin” (Roan, 2008), Dan Roan narrows the process of “visual seeing” to four steps: Look, See, Imagine, Show. The process we followed included those steps in a slightly different order, and recursively. It would be more accurate to say we Looked, Saw, Showed what we had seen, Imagined what might be different, and then Showed that. The first showing was for us to represent what we had seen in ways which could be shared – among ourselves and with others – and confirmed; the second showing was for us to share projections with others in the development enterprise who had not been in the field. Moreover, it seems as if each showing and imagining further informed the next cycle of looking and seeing, so the process itself became recursive. There is a heuristic used in Conversation Analysis which loosely paraphrased goes something like this: initially upon observing peoples’ actions, assuming they are acting rationally to accomplish something in an orderly way, regularly yields greater understanding of actually what – and how they bring about what – they are doing than does assuming they are acting in some rote – inaccessible to them – manner. Considering merely the broad strokes of their actions without asking and observing how it is exactly that they know what to do, and how to do it, is like mapping an iceberg from only what extends above water. A thirty second piece of video exquisitely demonstrates this point. One of the communications personnel, we’ll call her Dee, was routinely in charge of dealing with requests for supply bulletins from company representatives in the field. Her work is easily characterized in the following way: Sales Bulletins Request/Reports Orders Process 1. Salesman fills request order form Sends to Industrial Manufacturer 2. Dee gets request form (via fax?)

82

Making Work Visible

Figure 4.3 Closer examination of the work practices associated with fulfilling a bulletin request reveals a complex picture where Dee troubleshoots a problem with the request and informs those affected how to eliminate this problem in the future

3. Dee lists requested items – In inventory/log book/s – By sales rep’s territory – By item requested 4. Dee addresses UPS label and shipping notice 5. Dee collects requested materials from mail/print room shelves 6. Dee leaves: labeled shipping notice and collected materials on cart for shipment 7. Dave or Dan ship materials by UPS to Salesman

Such a characterization, familiar to us from many “Sale’s and Marketing” workflow descriptions,4 is not exactly inaccurate, but it is definitely inadequate. Consider the following: As Brun-Cottan and Wall wrote in 1995 “In this tiny moment of activity the seamless but intricate intermixing of artifacts and actions, of paper, pencils, faxes, copiers, of inscribing, gesturing, 4

Detailed descriptions of work flow are not the primary job of Sales and Marketing personnel,. Selling and Marketing are.

Work Practices to Understand the Implications of Nascent Technology

83

annotating, highlighting that are brought to bear on the reps order form is vividly displayed. More importantly, clearly in evidence is the diagnostic and remedial work that Dee performs to fill the request in a way which is both timely and which informs and instructs all the affected parties of what is happening. The range of resources on which she draws is enormous, from a detailed familiarity with sources of product within the company to exploiting ways in which handwriting styles allow for differentiating writers’ identities. That people do these things, and how they do them might be valuable to know if you wanted to build a tool to support their work.”

In terms of looking and seeing, the relevance of that layer of granularity afforded by close examination of video is that it exposes the immense reach of organizational knowledge employed by Dee to accomplish the seemingly routine and trivial but vital process of filling a salesman’s bulletin request form. Moreover, the range of resources she uses to accomplish the tasks at hand can surely inform the range of useful functionalities a device such as the still embryonic “document machine” might fruitfully provide. Coming from a perspective and time rather different from Mr. Roan’s, Samuel Johnson – as he was going about creating one of the first definitive dictionaries of the English language – was heard to complain in the following exchange: BOSWELL: “Then, Sir, what is poetry?” JOHNSON: “Why, Sir, it is much easier to say what it is not. We all KNOW what light is; but it is not easy to TELL what it is” (Boswell, 2006). For us that telling entailed how to triage and present the most salient, striking elements for sharing with our colleagues and organizational others within Xerox. From “Noticings” to “What Ifs” Review of our material created during the project provides an eclectic mix of observations. For example: filing involves climbing on ladders, kneeling on the floor, and repetitive bending and lifting. As documents keep piling up filing them becomes a lower priority, leading to trouble finding files, which is in turn exacerbated by files – and copies of files – being physically located in different locations, which complicates the updating process, which leads to incorrect information being used by field representatives. Or, though important documents are passed around the company the ultimate responsibility for the documents (and for the accuracy and timeliness of their content) remained with the file room, resulting in extensive tracking mechanisms, internalized systems, and enormous time spent searching for “lost” documents. Or, . . . the list goes on, dozens of observations of the actions performed in the face of various difficulties to achieve the orderly process of

84

Making Work Visible

Figure 4.4 Examples of “what if” opportunities were shared in working sessions with study participants to explore potential technology solutions to current work practice issues

business. Many of these processes are familiar to workers in, or observers of, businesses large and small across North America. From the most prevalent of these we culled ones possibly highly relevant to what a nascent technology might be shaped into addressing. For example:5 “Systems that would preserve and enable the essential work practices performed by the Communications Department staff in maintaining the storage, distribution, filing, copying, retrieval, and so on, of information and documents including: monitoring the current state and whereabouts of information, activities, and documents; supporting the internal document processing needs of communications personnel while at the same time servicing the information needs of other departments throughout the company.” (Sounds like something a lot of companies might use.) What if instead of filing documents in envelopes and drawers the documents could be scanned into an electronic filing system, either via a scanner or via a fax?

5

What follows is lifted completely from our “what ifs” of the proj

Work Practices to Understand the Implications of Nascent Technology

85

Or what if you could get people to tell a device to: make a hard copy of a document, and send a copy of it to a database, and also send a copy to an email or fax? And do it via a paper interface? What would you call a device that could do all that? You would call it a DocumentCentre (which was one of the first (the first?) office multifunction devices that supported (provided a bridge) for paper and electronic-based processes, enabling copy, scan, fax, print, archive and store content).

Conclusion It is always possible that people (study subjects, workers, users) may resent being closely observed. Sharing our study objectives, acknowledging their expertise in their work domains, and asking them to confirm or correct our understandings went a long way towards making the Sterling participants more comfortable with our presence. Using their own terms and activities and opinions provided them with a channel to voice their concerns and suggestions. This reconfirmation of our understandings became a standard practice in our studies whenever possible. Using video to record people’s actions while going about doing their work made it possible to repeatedly analyze components, facilitating identification of critical features in their work practices which would inform possible technological avenues for us to pursue. The social, organizational, and analytical knowledge required to perform some of the seemingly simple tasks glossed as “clerical” were revealed and made shareable through the use of video in ways difficult for written descriptions to accomplish. Having discovered the “bottom of the iceberg” – so to speak – in fragments such as Dee’s, meant that we would add being alert for such signposts. Unpacking obscured sense making aspects of work became a fixture of all our future engagements. As previously stated, maps, pictures, and film clips helped situate us when we were not in the field. Pictorial descriptions of how tasks presently were being done and then positing how they might be achieved using new technologies and functionalities helped us convey opportunities to members from diverse communities within our own corporation. In some respects these representations – embedded in the situated context of Sterlings’ workplace – helped us avoid some of the difficulties inherent in formulations which must serve varied perspectives, disciplines, and interests.6 These would include 6

For a critical treatment of this issue, see Brun-Cottan 2009, The Anthropologist as Ontological Choreographer.

86

Making Work Visible

those of our customers as well as those within our own larger enterprise. In the years to come these representations continued to be refined, not surprisingly primarily by the designers in the team, and have become a staple within Xerox. The ethnographic and design practices we employed helped identify and describe which functionalities an innovative but as yet unarticulated technology should address. Looking at one fieldsite we gained insights into issues and obstacles people ran into while trying to accomplish how and where to find and share the information they needed, when they needed it, with some assurance that the information was correct and current. Further studies have confirmed that these problems, however common were, and remain, critical and central to information and document management concerns in enterprises ranging from law offices to school classrooms, from university libraries to medical practices, from distributed work teams to small business owners working from home.

5

Tokyo to Go Using Field Studies to Inform the Design of a Mobile Leisure Guide for Japanese Youth Diane J. Schiano and Victoria Bellotti

Dai Nippon Printing Co, Ltd. (DNP), one of the world’s largest printing companies, is well-acquainted with the growing shift from paper to digital media worldwide. Consumers are demanding ever more rich and customizable digital content, ideally accessible by mobile devices. This trend is especially dramatic in Japan, where access to advanced content on the mobile web has been ubiquitous for years. DNP approached PARC to collaborate in creating innovative, consumer-friendly mobile digital media technologies for Japanese use. They then collaborated in a joint project with PARC fieldworkers and technologists in applying a user-centered approach to innovation design. This project (codenamed “Magitti1”) lasted over two years and comprised several phases, including: opportunity discovery and assessment, technology design and feasibility testing, platform engineering, and technology transfer. PARC design fieldworkers played a primary role in leading the opportunity discovery and assessment efforts. We led brainstorming sessions and workshops with DNP and PARC technologists, and we worked to clarify and prioritize a range of promising candidate design concepts. We then performed extensive field studies in Tokyo to obtain further design insights and requirements, and targeted user feedback on increasingly refined design prototypes. The project concept evolved into a working prototype of a context-aware leisure guide for young adults “on the go” in Tokyo. Further developed and refined by DNP, it is currently available as an iPhone app in Japan called “Machireco” (“town recommendations”). See Bellotti et al. (2008) for a comprehensive technical overview of the project and its relation to previous technologies.

1

The name “Magitti” reflects its evolution from two prior concepts, “Magic Scope” and “Digital Graffiti,” which contributed substantially to the final design concept.

87

88

Making Work Visible

This chapter provides a brief overview of the methods PARC fieldworkers used to inform the Magitti project. We discuss four field studies, focusing primarily on the use of converging methods and some novel techniques for gathering, analyzing, and displaying field data. Our methods were adapted to fit the priorities of the client organization, specific project goals and constraints, and the cultural context of study participants. For example, DNP was primarily interested in focusing on innovative consumer products, and ultimately, a leisure guide. Logistical issues around Japanese concerns for privacy arose in the process of conducting the research. In addition, quantitative data analyses were specified as project requirements, both to obtain a measure of statistical reliability for major conclusions and for modeling purposes. (These and related topics are discussed in more detail later.) The research reported here provides an illustration of ethnography-inspired design fieldwork, which is becoming increasingly common in corporate settings. Thus our work differs somewhat from those described in other chapters in this volume. Most notably, we studied leisure rather than work activities. Moreover, since leisure is a subjective concept – one may dine out for business or pleasure, or take a train to travel for fun or to commute to work, as well as for logistical reasons, we relied more on selfreport and less on observational data than did the ethnographic research discussed in other chapters of this volume (e.g., Whalen and Bobrow, this volume; Whalen and Whalen, this volume; Vinkhuyzen, this volume). Here, we illustrate how datasets of various kinds, obtained under diverse and lessthan-ideal conditions, can – when considered together – nevertheless yield a rich set of findings to inform the technology design process. In so doing, we make visible the practices our participants used in planning, coordinating, and executing leisure outings (which can require substantial work!). We also illustrate the research design and analysis practices we used to maximize what we could learn in the field given project constraints.

Motivating the Design Concept After initial market analysis, we held a series of workshops with DNP and PARC technologists, using brainstorming techniques and exercises to promote collaboration and get the creative juices flowing (e.g., wearing “crazy hats,” creating scenario posters, role-playing, making rough form-factor mockups). These were especially helpful with DNP participants, who were less familiar with this approach but participated heartily. We then posted design ideas on a board using sticky notes, and had participants group them together (affinity clustering) to identify major conceptual

Tokyo to Go

89

themes (e.g., communication, recommendations). Design concepts were further discussed, refined, and then prioritized by voting. Sketch mockups of high-priority concepts were made later for field-study feedback, which was coded and characterized along several emergent dimensions. We eventually focused in on the digital city guide concept, related to but several steps beyond the popular Walker series of Japanese leisure guide magazines (e.g., Tokyo Walker, Osaka Walker), recommending places to go and things to do about town. Over time, Magitti evolved into a context-aware, personalized mobile recommender system that city dwellers could use to get information about leisure venues while “on the go.” Commercial enterprises and endusers could provide content such as summary descriptions, reviews, and ratings. The user experience would be pleasant, with no effortful search required. Users could – but need not – directly enter profiles, preferences, or explicit search queries. Magitti would automatically provide and update recommendations through context-sensing and activity-inferencing mechanisms.

Field Studies We conducted a series of studies in Tokyo to inform the evolving Magitti design. The first two described here (in-depth interviews and online surveys) were conducted in 2004–5. Upon favorable review of the results of these studies, and acceptance of a proposal for further research, we were able to conduct two further studies (“street activity sampling” and “mobile phone diary study,” see following) in 2006–7. Our approach was to focus multiple methods on a set of key topics, starting out with a broad (and more qualitative) exploration of the design space and then narrowing in on very specific design requirements and model parameters. We were especially interested in collecting data on: (1) leisure priorities and practices of Japanese young adults of various ages (to help understand who might benefit most from a mobile guide and why), (2) resources used to support leisure outings (for a sense of information needs and pain points in using current tools), (3) common characteristics of leisure outings to parameterize recommendations (e.g., common modes of transportation, acceptable length of travel) and to help understand contexts of use (e.g., one-handed use on trains), and (4) common leisure activity types, their relative frequencies and temporal distributions. (Since Magitti bases recommendations on predicting activities in context, initializing data for new users whose patterns and preferences were not yet learned was an important system requirement.)

90

Making Work Visible

Methods Four primary methods were used for data collection: (1) in-depth interviews, (2) online surveys, (3) “street activity sampling” (SAS), and (4) “mobile phone diaries” (MPD). We started with fairly open, but extended, interviews with several individual participants to get an initial, qualitative grounding in the design space. While qualitative data was gathered in each study, the later studies were also designed for quantitative analyses, to allow us to address key issues in increasingly specific ways. We used what we learned in the interviews to structure the online surveys, which required a large sample size for statistical tests on general issues around leisure preferences and practices. The SAS study included brief open-ended and frequency-estimation items, this time on more specific aspects of leisure practice in situ (such as how long it took them to get to this leisure venue from their home, and the mode of transportation used), again to allow us to obtain a large sample for statistical tests. The diary (MPD) study responses, in situ and open-ended, were coded for quantitative analysis and modeling. Again, due to the subjective nature of leisure, we relied primarily on selfreport data, although there were observational components to the in-situ SAS – and to some extent, MPD – studies. Since every design study has strengths, weaknesses, and pragmatic constraints, we focused multiple methods on the key topics listed previously. This converging methods approach helps identify data patterns robust enough to hold across various contexts, conditions, and methodologies. Some further insights were derived from supplemental methods (e.g., expert interviews, focus groups on mobile phone use/usability, design critiques with Japanese technology students, mobile phone photo diaries) that are mentioned in the Results section.2 In-depth Interviews Twenty 16–33-year-olds (first round), and twelve 19–25-year-olds (second) were recruited by a Tokyo market research firm to participate in extended, semistructured interviews on their leisure priorities and practices. We had hoped to conduct interviews at home, but learned that this was uncomfortable for most Japanese people. Instead, we did interviews at a market research facility or at a public leisure location, perhaps a local restaurant 2

For more information on methods and results, see Schiano, Elliot and Bellotti (2007) and Bellotti et al. (2008).

Tokyo to Go

91

or café. Topics included demographic information, what they value about leisure time, how much they have, how they use it and why. We also asked for detailed descriptions of a specific recent leisure outing – from start to finish, including any supporting informational resources. Translators were used. The interviews gave us a rich sense of what people valued about their leisure activities, the most common leisure activity types, and how specific leisure outings were conducted, including support resources used. The sessions lasted at least two hours, and ended with discussions of design issues and feedback on early design concepts or from walk-through exercises with more refined prototypes. Online Survey Informed by the findings of the in-depth interviews, a survey comprising approximately 40 subjective ratings, multiple-choice and open-ended items was constructed. We used the survey to look for broad statistical convergence with high-level demographic, attitudinal, and behavioral conclusions drawn from the interviews. Questions addressed leisure priorities, characteristics of typical leisure outings, resources used, and the frequency of various types of leisure activities. We posted it on a popular Japanese online job site, and a total of 697 Tokyo participants (290 male, 407 female) completed it. Street Activity Sampling (SAS) Brief (10–20 minute) on-the-street interviews were conducted with 367 people (164 males, 205 females, modal age 19–25 yrs) in 30 multiuse locations (i.e., city centers with large varieties of shops, places to eat, and things to see, e.g., Roppongi, Shinjuku) around Tokyo and Yokohama. Time of day and day of week were varied. Questions focused on interviewees’ current leisure outing and the specific activities (previous, current, next) comprising it. A focal leisure activity was identified, and further questions focused on it, including whether and how it was planned, how the venue was discovered, resources used or desired, number of companions, mode of transportation, and travel time. Due to time constraints, many participants did not answer all questions. The SAS study provided detailed in situ data on characteristics of specific outings and a sense of the relative prevalence of various types of leisure outings. Mobile Phone Diaries (MPD) Finally, twenty-one 19–25-year-olds (10 males, 11 females) participated in a mobile phone diary study lasting seven days. (See Watts-Perotti et al., this

92

Making Work Visible

volume, for a description of an alternative diary study). This included 10 fulland part-time workers and 11 students from two local universities. Mobile email messages were written using a template prompting for: (1) activities engaged in currently or recently, (2) time, (3) location, (4) number of companions (if any), (5) mode of transportation, and (6) information resources/ media used or desired. Responses were to be made about once per hour except when working, studying, or sleeping. This week’s worth of detailed data on leisure activities and their temporal distributions by people with varied occupations and schedules provided default data that the system required. Findings In this section, we briefly present some field study results on four key topics: (1) leisure priorities, (2) use of information resources, (3) leisure outing characteristics, and (4) relative prevalence and temporal patterns of major leisure-time activity types. We illustrate how converging field evidence was used to draw specific design implications for Magitti. We focus some discussion on a promising new data collection method, the mobile phone diary, used in our final study, and briefly describe how the dataset was coded and analyzed. Finally, we present a small subset of MPD results using a fairly novel visualization technique – normalized area graphs – to illustrate the usefulness of such visualizations in promoting a more comprehensive understanding of patterns within a large dataset, and in facilitating meaningful comparisons within these overall patterns (specifically, temporal distributions of activity types by day of week and participant subgroups), even with fairly sparse data. Leisure Priorities First-round interviewees discussed what they valued about leisure time, and ranked the importance of several leisure qualities (e.g., relaxation, companionship) to them personally. A few top priorities came up for all participants, although ordering differed by age. Relaxation was important for everyone, but especially older participants (26–33 years): “On Sundays, I hardly ever get out of bed before noon, [and] even if I do get out of bed, I’m just relaxing, yeah, during that time.” Companionship and communication with friends were also high priorities for all ages. Companionship – spending time with friends – was the top priority for those aged 19–25. Communication was most important to 16–18 year olds, perhaps reflecting parental restrictions on outside activities. In Japan, the ages 19–25 are sometimes referred to as

Tokyo to Go

93

the “golden years” – a time when pressure to get into a good college has ceased, but climbing up the corporate ladder has not yet begun (Schiano, Elliot, and Bellotti, 2006). Participants in this age group reported spending half their waking day in leisure, almost twice that of both the younger and older groups. Interviews with Japanese publishers suggested that this demographic is particularly receptive to media recommendations, since they are developing more independent lifestyles, and most have fairly large disposable incomes (Iida, 2000). Online survey data provided quantitative confirmation of these results with a much larger sample of the population. d e s i g n i m p l i c a t i o n s Magitti must support a relaxing, enjoyable experience compatible with the spirit of leisure. Companionship and communication should be promoted, perhaps in part by supporting easy exchange of recommendations. Our target customer age-group became 19–25 year olds. Use of Information Resources *

*

Supporting Venue Discovery. Three top resources for discovering leisure venues were consistently mentioned in the interviews and surveys: friends and family, TV (“always on in the background”) and the Internet (via PC). Friend/family are the most trusted sources (“If I know that other people I know have tried out something before, than it becomes like trustworthy information and something that I can rely upon.”). In the SAS study, 64% of participants said they sought venue information for their current activity, suggesting a fairly common need for recommendations. Supporting Planning and Serendipity. SAS results show that 59% of venue choices were planned, 35% were decided “on the spot” and 6% were “habitual.” In addition, the PC/Internet was the top resource used to plan outings, with mobile email used most to coordinate, and the mobile Internet used only rarely for information “on the go.” Results from all four studies converge to confirm this pattern. Interviewees – and focus-group participants – said they planned on PCs for easy search and readability; they coordinated by mobile email since it’s a ubiquitous communication tool, and they used the mobile Internet only rarely, due to usability issues around mobile search. Participants also noted that while leisure outings tend to be planned around one focal activity, prior planning often facilitates serendipity in choice or venue of secondary activities (requiring

94

Making Work Visible

*

venue discovery “on the go”) before or after the primary, focal one (e.g., planning to meet before a concert at a specific train station with many shops and restaurants to choose from). And sometimes one just wants to explore or browse, or fortuitously happens by a promising venue (“I like to walk around town and find stores which usually you can’t find unless you’re walking, so I’m into walking around town. [On the way to the interview] I found a very nice looking bakery and also some nice restaurants . . . so that was interesting.”). Additional Content Desired “On the Go.” MPD results show that the most frequently accessed information “on the go” (via mobile Internet) concerned transit/navigation (especially train timetables). Some participants reported seeking other types of information, (e.g., restaurant/café venues) but not nearly as often. This trend was consistent across all studies. In particular, SAS participants reported desiring additional outing information about 59% of the time. The most commonly desired information concerned: maps/transit/location information, followed by opinions or reviews (by friends or previous customers), prices, and store/venue contents (see Table 5.1).

de si g n i m p l i c a t io n s More often than not, people engage in prior planning and use some information resource to discover leisure venues. They also want information to support their current or next activity. Searching the mobile web is rare even in Japan, despite its ubiquity and potential to provide information any time, any place. Magitti is a new solution that exercises this potential. The wealth of information types desired by participants informed our decision to use web content, published reviews, and end-user comments in our working prototype. Also, since people sometimes discover venues in context (and do not always have PC access), recommendations should be updated frequently with nearby venue information and reviews to amplify the experience of exploring a neighborhood. Magitti should serve as a “Magic Scope” onto what an area has to offer, making what is usually invisible visible. Ideally, Magitti should support some prior planning, since it is so prevalent. One way to do this might be to allow users to set a different time and place from their current ones to explore possibilities. Or perhaps we should develop an easy way for Magitti to access planning content from a user’s PC. A more ambitious (if futuristic) approach might be to allow Magitti – embedded in one’s mobile phone – to update recommendations by searching recent calendar entries, mobile website visits, even mobile email messages, to facilitate outing planning and coordination.

95

Tokyo to Go Table 5.1 Number of participants desiring various types of additional information to support a leisure activity in the SAS study (N = 294) Additional Information Desired (if any)

Percentage (N = 294)

No (121 participants) Yes (173 participants) • Maps, transit and/or location information • Friends’/customers’ opinions • Prices • Store/venue contents • Unusual (one-off) info requests • Item/food quality recommendations • Crowds/queues • Special event info • Recommend venues / provide itinerary • Taste matching / people-like-me • Food menu items • Schedules (trains/movies etc) • Sales/discounts/coupons • New venues/items • Comparisons • Interiors • Hours • Availability/sold out/delays • Ambience • Pictures/photographs • Weather • Parking

41.2 58.8 14.6 8.2 7.8 6.8 4.8 4.4 2.7 2.7 2.7 2.0 2.0 1.7 1.7 1.7 1.4 1.4 1.0 1.0 0.7 0.7 0.7 0.3

of

Participants

Leisure Outing Characteristics *

*

Number of Participants. SAS respondents gave us the number of participants in their outings. Group size was variable, but half (49.6%) had only a single companion. Average group size was 2.9. Only 14.4% of respondents were alone. Finding that leisure outing groups tend to be small, and that solitary leisure outings are fairly uncommon, is highly consistent across all of our field results. Transportation. When SAS participants estimated travel time to get to their focal activity venue (via any mode of transportation), the average response was 33 minutes. Given Tokyo’s excellent public rail system, it is not surprising that train/subway use was most prevalent (72.5%). Walking (19.6%) came in second, with other responses

96

Making Work Visible

*

fairly negligible. This is consistent with in-depth interview descriptions of travel to leisure outings. In addition, interview and focusgroup results show most people commonly use mobile phones single-handedly on the train. Frequency. SAS respondents estimated that they “went out” for leisure about three times that week. This number is consistent with our interview and survey data, but much lower than estimates derived from MPD logs (~ twice per day). MPD participants simply recorded daily activities; no motives, explanations or identifications of leisure outings were given. We thus suspect that a proportion of MPD outings that were classified as “leisure” may actually reflect something quite different (e.g., routine shopping, or eating out for mere convenience). The MPD and SAS data were both generated in context and in close to real-time, but the SAS study focused on leisure outings and included more subjective descriptions. Given the strong convergence between SAS, interview and survey results, we decided to use SAS estimates to ground our sense of the prevalence of true leisure outings. The MPD logs were taken as suggesting temporal patterns of various types of activities (leisure outing or not) while “on the go” during leisure time.

d e s i g n i m p l i c a t i o n s Since most leisure outings are not done solo, Magitti should be able to merge the preferences of at least one other user (preferably more) in its recommendations. The SAS data suggest that recommendations of venues up to 30–40 minutes away by train may be acceptable. Our frequency findings suggest that even Japanese youth in their golden years may not engage in true leisure outings very often. This suggests, on the one hand, that since a leisure-oriented Magitti may not be used every day by everyone, it should especially be useful, easy to use, and affordable. On the other, we might consider expanding our vision of Magitti to also include venues and content that are not strictly leisure-oriented. Prevalence and Temporal Patterns of Major Activity Types In describing recent leisure outings, interviewees most commonly mentioned: Shopping, window-shopping, going to cafés/restaurants, movies, sports events, concerts and hobby/club-related activities (in that order). Special outings included: Visiting “back home,” going to an amusement park, hot springs or music festival, ski trips, school/club field trips, and tourism (mostly within Japan). When online survey respondents rated frequency of engaging in various leisure activities, shopping and going to a

Tokyo to Go

97

restaurant/café topped the list (mean = 2.6 and 2.0, respectively, on a 5-point scale). In the SAS dataset, shopping and dining comprise over 50% of leisure outings (see also Manzenreiter and Horne, 2006). In addition, transitions (waiting or being in transit) were reported very often, perhaps not unexpectedly in a city as large and crowded as Tokyo. The MPD study provides detailed, in situ data on the prevalence and temporal distribution of leisure-time activities over an extended period of time (one week). This data was required to initialize Magitti’s activityinferencing model, but we designed the study to also provide additional information to further inform design recommendations. PARC technologists requested to have the data presented in a specific form of graphical display (normalized stacked area graphs) to promote synoptic views of the data and get a sense of the relative prevalence of major activity types over time. Comparing these graphs for specific days of the week over all participants helped provide initial estimates of the likely timing of various activity types overall. Further comparisons of temporal patterns for specific participant subgroups (e.g., full-time students v. full-time workers, males v. females), helped us to get a sense of where we might need to tweak model parameters for specific participant types. Some of these subgroup comparisons were of necessity based on fairly sparse data (precluding statistical comparisons), but when viewed in the context of the overall data and other, more robust findings, they helped to inform decisions that would otherwise have been made with no empirical grounding. For us, viewing the graphs took some getting used to, but proved extremely helpful in allowing us to literally see – and communicate easily about – patterns in this large, complex dataset. MPD data were initially coded into many specific activities, which were then grouped into four major activity types. These were coded as: EATOUT (e.g., dining at a restaurant, drinking at a bar, eating at a cafe), DO (e.g., engaging in sports, riding a Ferris wheel, taking a trip), SEE (e.g., watching a movie, viewing a ballgame on TV, attending a concert), and SHOP (e.g., buying shoes, going to a flea market, comparison shopping for clothes, window-shopping). Three more categories were added for context: EAT-OTHER (e.g., eating at home or bringing lunch to work; providing more data on eating times.), TRANSITION (e.g., riding on a train, waiting for a bus, queuing up to buy tickets; possible opportune times to use Magitti), and OTHER (activities not easily placed in the prior categories). These data were parsed into 15-minute intervals spanning most of the waking day, normalized by time interval and displayed as stacked area graphs. These graphs show the relative prevalence (in percentages) on the y axis, and time of day on the x axis. Figure 5.1 displays a subset of this data

20

40

60

80

0

5

10

15

20

5

10

15

20

6:

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0%

5

10

15

20

SUNDAY

0 00 :00 :00 :00 :00 :00 :00 :00 :00 :00 :00 :00 :00 :00 :00 :00 :00 :00 :00 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 0

FRIDAY

0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 6: 7: 8: 9: 10: 11: 12: 13: 14: 15: 16: 17: 18: 19: 20: 21: 22: 23: 0:

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

SHOP

EAT OUT

EAT-OTHER

DO

SEE

OTHER

TRANSITION

Figure 5.1 Relative prevalence of various types of “on the go” leisure-time activities over time and by day of week, for all participants in the MPD study (N = 21)

SATURDAY

00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 6: 7: 8: 9: 10: 11: 12: 13: 14: 15: 16: 17: 18: 19: 20: 21: 22: 23: 0:

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 6: 7: 8: 9: 10: 11: 12: 13: 14: 15: 16: 17: 18: 19: 20: 21: 22: 23: 0: MONDAY−THURSDAY

0%

10%

20%

30%

40%

50%

60%

70%

80%

90% Sample Count (Total) Sample Count (Total)

100%

Sample Count (Total) Sample Count (Total)

Tokyo to Go

99

for all participants, with separate plots for Monday–Thursday (combined; i.e., weekdays), and Friday, Saturday and Sunday, respectively). Activity types are represented by patterned bands in the graphs. Since the number of responses per interval varied widely, total sample count (i.e., total number of participants who reported engaging in that type of activity) per interval is shown by white line running across each graph. This is intended to provide a sense of the reliability of the data in each interval. As sample size increases, the pattern of results becomes more reliable. Figure 5.1 thus shows the relative frequency of various types of leisuretime outings over time for all participants throughout the study week. The graphs are complex, but some fairly straightforward conclusions can be drawn. Note that TRANSITION activities (white area with black dots) are most prevalent in all the graphs, presumably reflecting Tokyo’s long commutes and travel times. Of our four major leisure activity types, EATing-OUT (dark gray) was most prevalent by far, followed by SHOPping (off-black). Also, EATing-OUT seems surprisingly dispersed throughout the day. It might be argued that since this coding category included going to a coffee shop for a bite before a movie as well as dining out as a focal activity, it might not be expected to be tightly coupled to canonical meal times. But when we also plotted EAT-OTHER (e.g., eating at home or taking lunch to work; medium gray), the combined curves make much more intuitive sense. EATing (combined) is highly prevalent throughout each day; EAT-OTHER more so than EAT-OUT. Most EATing-OUT takes place midafternoon on weekdays, afternoon-early evenings on Fridays, and later on Sunday evenings. SHOPping is fairly common on weekdays. SHOPing declines somewhat on Friday and Saturday, but seems to resurface on Sunday. Figure 1 shows DOing (light gray) activities as thinly dispersed over weekdays, with a bulge midday Saturday and Sunday morning-to-early-afternoons. SEEing (black on white brick pattern) is the least prevalent activity type overall, occurring mostly on weekends, with Sunday showing the most SEEing activities throughout the day. Separate sets of graphs were made to compare data patterns for several subgroups of participants on specific days of the week. Note that sample size decreases with every subgroup considered. For example, in considering occupational status, we first compared all college students against all nonstudents. However, many college students worked part-time, and many nonstudent workers took a class or two. When we looked at those who were full-time college students only, full-time workers only, and part-time workers only, we were down to three or four qualifying participants per group. Figure 5.2 shows the plots for the temporal distribution of activity

1

2

3

FULL-TIME COLLEGE ONLY (N = 3)

1

2

3

SHOP

EAT OUT

EAT-OTHER

DO

SEE

OTHER

TRANSITION

FULL-TIME WORKERS ONLY (N = 4)

00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 6: 7: 8: 9: 10: 11: 12: 13: 14: 15: 16: 17: 18: 19: 20: 21: 22: 23: 0:

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0

1

2

3

4

Figure 5.2 Relative prevalence of various types of “on the go” leisure-time activities over time on Friday only, for full-time college students only (N = 3), full-time workers only (N = 4), and part-time workers only (N = 3) in the MPD study

PART-TIME WORKERS ONLY (N = 3)

0 0% 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 6: 7: 8: 9: 10: 11: 12: 13: 14: 15: 16: 17: 18: 19: 20: 21: 22: 23: 0:

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0% 0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 6: 7: 8: 9: 10: 11: 12: 13: 14: 15: 16: 17: 18: 19: 20: 21: 22: 23: 0:

10%

20%

30%

40%

50%

60%

70%

80%

90%

Sample Count (Total) Sample Count (Total)

100%

Sample Count (Total)

Tokyo to Go

101

types for these three “pure” occupational groups on Friday only. The resulting graphs are very blocky, since they are based on such small sample sizes. Still – and even without looking very closely at the graphs – it is clear that their activity patterns differed greatly. For example, part-time workers tended to sleep in late, while full-time students and full-time workers were both out fairly early. Full-time workers seemed to engage in EATing-OUT and SEEing more commonly than the other groups, and so on. Overall, the broad temporal patterns for various subgroups visible in the MPD graphs seems reasonable and converges well with our prior findings as well as some recent Japan Statistics Bureau data (2007). MPD participants also recorded media use and type of media content accessed (if any) during leisure time. These data were also coded and plotted in normalized stacked area graphs as described above. We lack space to discuss these results in detail here, but will note that some very suggestive patterns were found. For example, full-time workers (especially males) seem to have the most predictable schedules of media use; part-time workers (especially females) had the least. Full-time college students tended to use the PC/Internet during leisure time much more than any other group. We also found that communication of some sort (email, voice, blogs, social networking sites) – and not access to media content – overwhelmingly dominates Internet use (both PC and mobile) for this age-group. d e s i g n i m p l i c a t i o n s Magitti requires not only relative frequency data but also temporal distribution and other contextual information to infer likely activity types of interest to inform recommendations. Since we were helping to initialize an always-on recommender system that would immediately start learning usage patterns, perfection was not required, and any data – however noisy – was better than none. The MPD results suggest some specific categories that the model could identify and use to conditionalize predictions (e.g., time of day, day of the week, user subgroups), at least initially. Magitti would then implicitly learn typical schedules, activity sequences, habitual locations and favorite venues. Activity type likelihood estimates at given points in time could also be conditionalized on any profile (e.g., age, gender, occupation) or preference information input by the user. Behaviors such as inspection of details of specific venues and any ratings given could also be used to infer what activity type the user is interested in. Research suggests that users willingly rate things they have personally experienced (see Bellotti et al., 2008). So Magitti should encourage reviews and ratings, partly for evidence that a user has just been to a venue if they are nearby when, or shortly before, they create them. From this it could learn

102

Making Work Visible

activity patterns and use that data to influence recommendations. Another approach might be to make broad activity classes explicit in the user interface as a means by which to filter recommendations. Users might be more inclined to provide information on their activities if they get better recommendations as a result. Media use and content access patterns suggest that Magitti should be sensitive to the high value of communication to its users. For example, it could support easy content sharing and social networking, and should enable people to build trust in others’ recommendations through mechanisms that foster social capital (recognition for providing helpful information). Magitti’s “always-on” display and easier access to relevant content should increase mobile internet use and usefulness. Content that is accessed predictably could be automatically displayed at the right time for the right users (e.g., weather and train information early weekday mornings, especially by full-time workers and college students). Moreover, Japanese demographics suggest that given a low enough price-point and good ease-of-use, younger part-time workers (a substantial proportion of whom own advanced mobile phones but not a personal computer) might prefer accessing the web via Magitti rather than investing in a PC.

Discussion One motivation for DNP to work with PARC was to participate in our approach to user-centered design. The market analyses, workshops and brainstorming exercises, and our iterative gathering of feedback on design concepts and prototypes were very useful in informing the Magitti design process. Our most notable contributions derived from the field studies are briefly described here. Converging patterns of results from these studies led to direct design implications for Magitti, most of which were implemented or seriously considered for future versions. The SAS and MPD data provided important activity prevalence and temporal distribution data with which to initialize the model, as well as to suggest major conditionalizing user group and contextual factors. We acknowledge that the field data derives primarily from self-reports, which are notoriously subject to bias, especially when made retrospectively. To lessen the bias, the SAS – and especially the MPD – studies were designed to produce data naturalistically and essentially in situ. We also took care to address key topics with two or more differing methods whenever possible, typically with both qualitative and quantitative data. Using this convergent methods approach, one study’s weaknesses can be complemented

Tokyo to Go

103

by another’s strengths, and vice-versa. For example, interview results often have higher validity than structured survey responses, but reliability issues are better addressed statistically with surveys. Collecting a broad swath of responses from participants in the midst of an ongoing urban outing (as in the SAS study) may tell us a lot about leisure outing characteristics, but will not provide the detailed temporal distribution of participants’ leisure-time activities that the MPD logging data does. The extensive MPD dataset had to be coded, quantified and presented appropriately to yield meaningful results – which required substantial effort, especially with translated text and variable response rates. Still, the results proved invaluable in providing a sense of the potential design impact of specific user characteristics and contextual (especially temporal) factors, in addition to the initializing likelihood data for the model. In the course of this project, we made some process-oriented and practical observations that might be helpful for future fieldwork of this sort. First, our experiences underscore the value of brainstorming and related exercises to promote team cohesion as well as to help quickly generate a wealth of creative design ideas. The importance of facilitating sociable collaboration in cross-cultural teams cannot be overestimated. Secondly, we learned that conducting extensive fieldwork in another country takes a great deal of planning, coordination, and checking in with everyone involved to make sure that all parties agree. For example, recruited participants in Japan were much more wary of having researchers come to their homes than we anticipated, so we needed to find alternative locations. Also, Japanese companies take customer privacy extremely serious, so we learned that to protect DNP’s “privacy mark” – a kind of Good Housekeeping Seal for corporate privacy assurance – we were not permitted to record anything that could ever have the remotest possibility of being used to identify specific people, even with their consent. This wreaked havoc with our original mobile phone photo diary study, which required participants to send us photos of their leisure locations from their mobile phones, since any recording of human faces, even in crowds (impossible to avoid in Tokyo) were unacceptable. This resulted in photos (e.g., of shoes, food, TV screens) that were for the most part uninformative about locations, and thus useless for the purposes of the study. In addition, since we were barred from directly recording participants’ voices, an elaborate and cumbersome system of microphones and recorders had to be devised to ensure that only the translator’s voice – never the participant’s voice – could be heard in any recording. Moreover, despite the complexity and effort required in conducting, coding, and analyzing the MPD results, this fairly novel approach

104

Making Work Visible

to data collection generated a uniquely rich dataset. This facilitated model initialization, and also permitted exploration of potential design implications of the relative prevalence and temporal distribution of various types of “on the go” activities (and media used, content accessed) overall and for specific user groups. Finally, we came to appreciate that a novel form of data visualization (normalized stacked area graphs) can be an extremely useful tool for comprehending, comparing, and communicating broad patterns of findings from a large, heterogeneous, temporally based dataset. This project illustrates a client services engagement at PARC, Inc. Our client, DNP, asked us to help them with the process of designing an innovative product for its consumer market. To do this, we had to work with our client’s needs and expectations, as well as more specific requirements of the technologists on both sides. A client services engagement of this sort can be complex, requiring close collaboration between parties. We were fortunate in working with a set of very interested and agreeable set of managers and engineers, with whom we communicated often and in explicit detail to promote clarity. By its nature, a client services engagement places constraints on research. Researchers must work together with the client in prioritizing goals, addressing research questions, choosing analysis methods, and specifying timetables and deliverables. We helped DNP (and our PARC colleagues) determine likely product concepts and target markets. We brought them a fresh perspective and a willingness to share our techniques. We designed studies that provided the quantitative data needed for statistics and modeling, based on a solid qualitative grounding. They provided us with unparalleled access to Japanese people, cultural insights, and logistical and technical support. They also gave us the opportunity to work on a research project whose deliverable was a prototype for a real-world product. In summary, PARC design fieldworkers played a pivotal role in the Magitti project, conducting initial market analyses, leading workshops, and brainstorming sessions to motivate the design concept, and then gathering extensive field data to further inform design. A converging methods approach (including in-depth interviews, online surveys, street activity sampling (SAS) and mobile phone diary (MPD) studies was used, focusing multiple methods on a set of key topics. We started out with a broad (and more qualitative) exploration of the design space and then narrowed in on increasingly specific and quantitative design implications and requirements. The in-depth interviews gave us a sense of the leisure priorities of Japanese youth of different age-ranges, some rich retrospective descriptions of some specific types of leisure outings and their characteristics, and a nuanced

Tokyo to Go

105

understanding of issues underlying leisure patterns and preferences. The online surveys provided some confirmatory findings and quantitative benchmarks using a large sample of Japanese participants in our target age-range. The SAS study focused in on characterizing current leisure outings of a large and diverse sample of young people at a variety of multiuse locations in the Tokyo metropolitan area. The results informed activity inferencing and provided some important benchmarks for parameterizing recommendations (e.g., likely mode of transportation, acceptable travel times). Finally, the MPD study employed a fairly novel method for gathering detailed, close to real-time data on the leisure-time activities of several participants (in the same age-range, but differing in occupational status) over the course of a week. This provided initialization data for Magitti’s activity inferencing model (i.e., the relative prevalence and temporal distribution of specific types of leisure-time activities), as well as some intriguing results on patterns of leisure-time activities and media access. Visualizing the MPD dataset with normalized stacked area graphs (another fairly novel technique with great potential for researchers doing fieldwork) permitted synoptic comparisons of temporal activity patterns between participant subgroups and on various days of the week. This enhanced our ability to comprehend and communicate about user and contextual factors which might be useful in conditionalizing the model. Taken together, these field studies yielded a remarkably rich and extensive set of findings, some of direct application and others – especially those demonstrating converging patterns across studies – providing a powerful base from which to draw design implications for this innovative project and product.

Part III

Practices around Documents

6

Exploring Documents and the Future of Work Jennifer Watts-Englert, Mary Ann Sprague, Patricia Wall, Catherine McCorkindale, Lisa Purvis, and Gabriele McLaughlin

Rapid sociotechnical change is underway in the world of everyday work. The number of remote and mobile workers is increasing over time (World at Work, 2007). Large corporations are reducing real estate costs by asking employees to work at home. A 2006 report by Forrester noted that 11,400 Hewlett Packard employees work from home and nearly all Jet Blue sales agents work from home. This same report said that 42% of the 320,000 IBM employees worldwide do not have traditional offices. Sun Microsystems allows employees to choose between working in a static office, working exclusively from home, or working in drop-in office spaces (Driver, 2006). Mobile technology has evolved enough to allow people to work almost anytime, anywhere. Millions of people now work in coffee shops, restaurants, cars, airports, and parks (World at Work, 2007). The worldwide shipment of smart phones is exploding (McGrath, 2009). These smart phones allow users to browse the internet from their phone, and have led to an explosion in the number of applications written specifically for these phones (Anderson, 2008). For example, the iPhone is linked to an applications store containing thousands of applications, ranging from games to business management solutions. The world of work has also been impacted by changes in the workforce. The older members of Generation Y have just started to enter the workforce. This generation, born between the years of 1979 and 1995, has grown up with digital technology. As Generation Y enters the workforce, it will be important to watch how they impact the use of technology at work. The convergence of these trends – alternative work settings, increasing mobility, and a new digital savvy workforce – inspired us to conduct an ethnographic study to explore how work is changing. The main goal was to learn more about these trends, and use that information to inform the development of new opportunities for products, services, and solutions. While we were interested in learning more about general changes in work 109

110

Making Work Visible

Figure 6.1 Two central questions in the study

practices, we were also specifically interested in how these changes would affect printing behaviors and the use of paper. An understanding of these issues could help the company brainstorm about how printing should change to fit evolving work practices. For example, how should printing change to support new information structures like wikis and blogs, and new media like audio and video? A better understanding of current printing and paper practices could also spark discussions about how the company could choose new investments to replace paper if the use of paper decreased significantly. If the study found that participants were not currently printing, we could explore why they were not printing, and what they were doing instead (see Figure 6.1). An understanding of the future of paper and printing is vital to the health of our company since printing and paper documents have always been a strong business focus. Therefore, our study findings have significant implications for Xerox strategy.

Methods This chapter presents findings from the first year of a multiyear project that began in 2008. This first-year study was broad and exploratory. We used interviews, collages, and diaries to study a diverse set of participants. We formulated interview questions with input from an internal advisory board. Study results painted a broad picture of the changes that are occurring in the world of work and helped us identify which topics warranted further study. The second year of the project used interviews and extensive observations to focus in more depth on a few of the topics that were identified from the first year study.

Exploring Documents and the Future of Work

111

Participants For the first year study, we interviewed a total of twenty-six participants. Twenty-four worked at home and fifteen engaged in mobile work (work done outside a static office). Study participants did not work for the same company, and they worked from home or from mobile locations (outside traditional office spaces) for 50–75% of their time. We recruited by word of mouth, and participants received compensation for their time. The studies took place in three different cities: Rochester, NY, Grenoble, France, and Washington, DC. Interviews were audiotaped and videotaped, and were transcribed for analysis. Interviews The basis of the first-year study was a set of in-context interviews. The interviews lasted approximately two hours and were conducted in the place where participants did most of their work. If they worked at home, the interviews took place in their home office. If they were mobile workers, we met them in locations that represented a typical place where they often worked. The interviews were conducted in context because it provides rich memory cues for participants, and allows them to show us examples of things that were described in the interview. For example, if a participant used a specific tool for coordinating work with her team, she could open the tool during the interview and show us examples of how she coordinated. Conducting the interviews in context also allowed us to see exactly how participants’ workspaces were arranged. We noted where printers were located, where paper was placed within the workspace, and where the workspace was located. These kinds of observations provided cues to further guide the interview. For example, if a participant had stacks of paper on his desk, we asked him to give us a tour of the kinds of information contained in the stacks, and how these stacks changed over time. The interviews combined a set of prespecified topics together with freeform questions derived from the context of the interview. Prespecified topics were created with input from the internal advisory board, and included questions about where participants did their work, how they coordinated and accomplished their work in various settings, how they managed information (or not), how they incorporated paper and printing into their work, how they balanced work life and personal life, and how they used technology to coordinate their work. These topics were complemented with topics that participants brought up, as well as topics cued by context.

112

Making Work Visible

The prespecified topics allowed us to answer some of the basic questions of the study. By asking all participants a core set of questions, we insured that we could compare the topics across participants. The free-form aspect of the interviews and observations was also very important. This part of the interview allowed us to discover findings that we would not have thought to ask about. For example, one participant talked about critical mass periods in her projects, where team members needed to be available to solve last minute problems to meet a deadline. These periods in her projects required tight collaboration within her team. However, her team was experiencing difficulty because some team members worked from home and were not always available during traditional working hours. Our prespecified questions did not ask about the collaboration cycles inherent in work projects, but this topic arose from the interviews by allowing participants to elaborate on the topic when they brought it up in the context of answering other questions. Diaries Participants prepared for the interviews by completing a diary for a period of seven days. The diary asked participants to use pictures and text to describe who they were, tools and software they used, where they worked each day (see Figures 6.2 and 6.3), who they worked with, and whether/how much they printed. Participants also completed a section using stickers to show the networks of people they worked with (see Figure 6.4), as well as a photo gallery section that requested specific pictures like their primary work spot, tools they used, and how they worked with other people.

Figure 6.2 Example of a diary page asking participants to list the places where they worked

Exploring Documents and the Future of Work

Figure 6.3 Diary entry of one participant’s commute

Figure 6.4 Diary page showing participant’s work-related networks

113

114

Making Work Visible

By completing the diary, participants became more aware of the aspects of their work life that we wanted to discuss with them in the interviews. The diaries focused participants’ attention on the habits and practices they might not have otherwise noticed. This increased awareness allowed us to go deeper into the topics during the interviews, and enabled participants to talk about aspects of their work life that they might not have thought about if they had not completed the diary. In fact, the interviews began by asking participants to give us a tour of their diaries. This allowed participants to reorient to their work practices of the previous week, and allowed them to introduce themselves to us. In addition to facilitating discussions during the interview, the diaries also included basic survey questions asking things like how participants accessed the internet, the brand and model of their printer, cell phone, computer, and other tools, as well as basic demographic information like age, gender, and job title. Collages During the interviews, we complemented the conversation with a collaging technique (Strappers and Sanders, 2004) that explored participants’ ideas about their work environment. Participants used stickers, markers supplied by us, plus any other media they had on hand to create two kinds of pictures: their current work environment and their ideal work environment. When they were done, they described the collage and talked about why they added specific elements to the collage. This technique sparked interesting conversations about what was good and bad about participants’ current work environment, as well as what they wished they could do to improve their work environment. For example, when discussing the collage shown in Figure 6.5, the participant described techniques he used to facilitate time management, and talked about his aspiration to keep his work going from anywhere, even when he was traveling. The collage activity gave participants the opportunity to examine and describe their work environment in nonverbal ways, and therefore led to a richer description of what they liked about their current environment, and what they would like to change. This part of the study provided details about how work and personal life intermingled, and how participants struggled to maintain a balance in their lives.

Key Findings This section provides a sample of some of the findings from our study of future work scenarios, with a focus on working from home, mobile work, and the use of technology, paper, and electronic documents.

Exploring Documents and the Future of Work

115

Figure 6.5 Finished collage

Working at Home Was Highly Valued We found that participants highly valued the ability to work from home. Working at home provided participants with greater flexibility and comfort, and reduced wasted commute time. They could control their own environmental conditions like lighting, sound, and temperature, and they could create a more fluid work/life balance. For example, as Meerwarth et al. (2008) and Jordan (2008) describe, participants in our study could engage in personal tasks during the day when necessary, and sneak in extra work hours when they were not interacting with their family. “Being home is great now that I have kids that are starting to go to school. When they come home, I’m there.” “Sunday morning I wake up, I work for an hour while everybody is kind of having breakfast, whatever.”

Most participants said they would never work in a job that required them to go back to the office. One participant even chose to curtail his career progression so he could continue to work at home. “I don’t want to go back now. I don’t really see myself like trying to advance either, because I’d have to go back to work [in a corporate office]. I don’t see the point in that.”

116

Making Work Visible

“I’ve been working from home for five years. This is not new to me. In fact, I don’t know if I can ever work in an office again.”

Home Workers Still Experienced Problems Though participants loved working at home, they still experienced complexities related to this work style. Most participants shouldered the burden of keeping their technology connected and working properly – having to spend hours on telephone help lines when something went wrong. Most participants found these tasks to be unpleasant at best. The task of repairing faulty equipment usually turned into a huge time sink that took participants away from the work they wanted to be doing. “I spend a lot of time doing my own IT – on help lines, etc.” “(that’s not) professional work. It’s just taking care of nonsense work.”

Jordan (2008) mentions that one of the hidden costs of working remotely is that remote workers must invest a “tremendous amount of energy . . . in maintaining (work) relationships” because “potentially serious problems flow from neglecting their maintenance.” Participants in our study also said they had to be proactive about maintaining relationships with distributed colleagues. One participant (P) we interviewed (I) scheduled events on his calendar to remind him to contact colleagues to check on project status. P: I use the calendar function a lot to set reminders for myself. You know, I just use Gmail for everything, calendar reminder, check with Smith 3 p.m. and set it to remind me every hour until I do it. So it keeps me on top of people, and so I know where they are at. Again, it’s just planning. It’s making sure that I have all of my ducks in a row. I: So you’re basically planning out when you need to check in with people to make sure that you get what you need. P: Right. So there’s no surprises. We’ve all been in those situations where somebody has dropped the ball on you, and you’re the one where the buck is stopping there. Other participants invested considerable effort in maintaining visibility with important colleagues like managers or bosses, making sure they responded quickly to messages from these colleagues to indicate they were available, and engaged in their work. This investment in visibility sometimes caused participants to feel like they were always working. “If the boss doesn’t see you every day, they miss all the good stuff you’ve been doing, but they’ll remember if you forget something.”

Exploring Documents and the Future of Work

117

Mobility Amplified Problems Mobile workers experienced even more complexities than those who worked mostly from home. Time spent traveling squeezed productive work time. Also, participants spent time preparing to leave the office, and reorienting when they returned. These extra tasks could add hours onto participants’ work days. Preparatory activities included gathering materials they would need while they were out of the office, and printing documents that they might have trouble getting to if they did not have internet or printer access. When they returned to their static office, participants reoriented to their work by checking incoming messages to determine what they missed when they were away. Goldmacher (2008) describes how follow-up activities could add hours onto a day of mobile work. “I would spend eight or nine hours a day on a college campus and then return home with hours of follow-up work to do.” Goldmacher’s follow-up activities included responding to voice mails and emails, sales related tasks, and preparing for the next day (Goldmacher, 2008). While they were mobile, participants constantly checked email during micro-breaks to stay in touch with the status of their work (see Figure 6.6). Keeping up with email while on the go helped participants anticipate work as it arrived, helped them stay available to clients and colleagues, and saved participants hours of email review at the end of the day.

Figure 6.6 Participants constantly checked email while they were mobile

118

Making Work Visible

“I think the beauty of these mobile devices is now it’s even much easier to really keep up, and . . . anything that needs my attention, it happens. It has made my life so much easier because I don’t come home at night and have to read through all of the stuff. I easily save a couple of hours every day with the Blackberry.”

However, even though mobile email devices helped keep work at bay, mobile participants still struggled with an overwhelming amount of incoming information, as seen in the following quote: “I have 700 messages in my inbox. They’re there because I have to do something with them.”

Participants Struggled to Maintain Work/Life Balance While mobility amplified problems for participants who worked outside a static office, all participants worked hard to creatively balance work and life. We observed several strategies that participants adopted in an effort to maintain balance. Some participants said they scheduled thinking time on their calendars. Others displayed “do not disturb” signs in an attempt to carve out quiet time when they were working from home. Those who worked from home said their work/life tasks were much more interwoven. They interspersed laundry, childcare, and kids’ extracurricular events with work throughout the day. They often found themselves working late at night or early in the morning to offset the time spent during normal working hours on personal activities. This sometimes left participants feeling like they were constantly working. “My weekends aren’t mine anymore.” “That is one of the challenges of working at home is . . . keeping work life separate from your home life, but it’s good too, because I don’t – I’m not stressed during the week.”

While it was difficult to keep a comfortable work/life balance when working outside a traditional corporate office environment, all participants were very happy to have the opportunity to choose how they blended their work and personal lives. The participants who had previously worked in corporate offices commented that they were much happier working from home. Movement toward Dynamic, Nonlinear Information Structures In addition to the struggle to maintain a comfortable work/life balance, participants also experienced difficulties with current document formats and

Exploring Documents and the Future of Work

119

systems as they were doing their work. It was difficult and time consuming for participants to transfer information across file or application borders. Cutting from one application, like a word processing application, and pasting that information into a different application, like email or a presentation application, led to unexpected formatting changes, which required large amounts of time to correct. In fact, these changes sometimes went undetected until it was too late. One participant said that he often cut information from various documents and pasted it into email notes but encountered problems: “It changes my font sizes. It looks like I used comic books when I assembled the look of the different fonts. I end up with an email that looks like a ransom note in there. What I’m looking at is the draft. Everything appears to be one font and it’s not. When I BCC myself, it looks like I did a ransom note from the funny papers.”

Some participants gave up on the attempt to merge files electronically, and instead resorted to printing disparate files that needed to be merged, and scanning them back into PDF in order to create a file that included all of their desired information. This process allowed them to control the order in which information was presented in documents like client briefs, and allowed for quicker information integration. Future workers will need to easily isolate, chunk, move, and access the right information at the right time, without the current overhead of transferring information across today’s file-based application. To try to get these needs met, some participants were already using alternative tools like DITA (an XML-based tool that facilitates content reuse), wikis, and other information structures that make it easier to use and repurpose smaller chunks of information in a more fluid way. Printing Was a Compensation for Technology Weakness In addition to studying general work trends, we also asked participants about their printing habits and behaviors. The role of paper is changing from information recording and archiving, to a more temporary role of containing transient information (i.e., Sellen and Harper, 2003). In our study, we observed that paper was used to support more transitory tasks like thinking, brainstorming, and to-do lists. Another way to look at paper’s newer transient role is to see printing as a compensation for technology weakness. Participants in our study printed for the following reasons: – –

To remind themselves of what they needed to work on To read on the go

120

Making Work Visible

– To proof the document (i.e., make sure it prints the way it’s supposed to) – To edit documents like PowerPoint files, and manually re-enter corrections – To create leave-behinds (brochures left with clients) – To transfer signatures in accounting systems for things like expense reports All of these reasons for printing are in reaction to technology weaknesses. For example, participants printed to remind themselves of what they needed to work on because things got lost in the vast network of information that existed behind the small screen on their computer. Participants said they printed documents that needed to be processed because they lost track of them when they disappeared into their computer. Participants printed to read in mobile situations to overcome screen size, power management, and portability problems that existed in their laptops. They printed to edit documents because it was easier to hand-write and draw comments on a paper version of a PowerPoint slide than it was to find the menu that allowed them to insert a comment, position the comment box, and type the comment. All of these reasons demonstrate why participants continued to print despite the general movement of information from paper-based archives to digital storage. It is important to note that when technology finds a way to solve the problems that are causing people to print, printing will no longer be necessary to address those problems anymore. If technologists solve these problems quickly, printing could decline faster than we might predict. This finding has significant implications for Xerox’s printing business.

Making an Impact: Facilitating Innovation within the Company The future of work project covered a broad set of topics and had the goal of helping the company explore possibilities for brand new product and business solutions. Through past projects, we have learned that business groups and product teams are often so busy solving current problems that they often do not have the bandwidth to look beyond their current practices and business orientations. This has always been a problem for corporate researchers – to make impactful connections with frenzied development teams who are driven by deadlines and the bottom line. The Work Practice group at Xerox has historically attempted to increase the chances of making an impact within the company by envisioning study

Exploring Documents and the Future of Work

121

implications in ways that would resonate with business groups and by documenting intellectual property derived from these studies (see Brun-Cottan and Wall, this volume, Colombino et al., this volume, and Martin and Wall, this volume). The future of work project continued this emphasis on intellectual property and envisioning implications. We also created an advisory board to garner visibility and support from business groups. Envisioning Implications Qualitative findings of ethnographic studies may not always help a company draw explicit implications. We used several techniques to envision study implications in the future of work project. We brainstormed explicit business-relevant implications, envisioned possible responses to study findings through a design directions document, and created a video podcast to reach a broader audience within the company. Business-relevant Implications Throughout the data analysis, the research team continually held working sessions focused specifically on drawing out business-relevant implications from the study findings. One of the clearest examples of a business-relevant implication was the finding that printing was a compensation for technology weakness. The data that led to this finding were the reasons why participants printed (to remind themselves of what to work on, to easily edit or mark up a document, to read a document while on the go, etc). In examining the nature of why participants printed, the team realized that all of these reasons were attempts to address technology weaknesses (constrained screen size, usability, portability, etc). The implication of this finding for our business was that printing (the company’s core business) could decline rapidly if or when these technological problems were solved. Since this finding was so significant for the company, it received quite a bit of attention from business groups and corporate strategists because it presented the possibility of a direct challenge to the company’s core business. In fact, the following year, the corporate strategy was updated to include businesses that focused on solutions that did not revolve around printing. It is important to note that the specific implication that printing could decline did not appear explicitly in the study findings. The findings were observations about why participants printed. Through analysis of the findings, the team developed the implication that print could decline quickly if technology progressed in specific ways. It was not enough to just present our observations from the study. We had to make the implication explicit to make an impact within the company.

122

Making Work Visible

Active Team Members

Figure 6.7 Virtual collaborative team space: an example of a concept template developed as part of the future of work project deliverables

Design Directions Document To support the process of drawing out explicit business-relevant implications from the study, the team created a design directions document. The goal of this document was to provide a vision for how the company might respond to study findings and implications. This vision was communicated through a set of high level concepts and scenarios, which were mapped to study findings. The design directions document summarized study findings, provided sketches and descriptions of the concepts, and discussed how the concepts could address the difficulties observed in the study. The concepts were not specific recommendations for product ideas, but were instead high level templates that intentionally left room for readers to fill in details that resonated with their business function. The document was designed to be read by many different kinds of people within the company, including researchers working on innovation and new concept development and business groups working on product development. Figure 6.7 shows an example of one of the concepts presented in the design directions document. The basic concept template was a software or service that facilitated remote collaboration. While many systems like this already exist today, we found that our participants were still not able to make them work effectively to support their tasks. We used the concept template to frame the study results about the current problems experienced with remote team collaboration, and added details to the design directions document

Exploring Documents and the Future of Work

123

describing how a collaborative team space might be built to better support remote collaboration. While the image representing the concept template does not include all of the contextual details that we observed in the field, each concept in the Design Directions Document was embedded within a rich description of the fieldwork findings, which included images, quotes, and text. Text was also inserted into the concept template header describing how it might address some of the problems we observed in the field. Video Podcast In addition to the design directions document, the team embodied study findings and implications within a video podcast to reach a broad audience across the company. The podcast was a ten minute stand-alone video that summarized key findings and concepts, included video clips from the field, and referred viewers to the design directions document for more details about the study findings. We created this video with the hope that viewers would better remember study results after having experienced a multisensory representation of our findings. The video was well received throughout the company, and has been adopted by the internal Public Relations group, who is modifying it for external presentation. The Advisory Board: Garnering Support within the Company Since the project had the challenging goal of helping the company look beyond current practices and business offerings, we wanted to start with a better picture of the current state of the business groups: What questions and issues were they struggling with? What new services and solutions were they already envisioning? If we understood their current state, we could help answer their current questions, and help them look beyond these questions to envision new business opportunities together with us. By inviting them to collaborate with us throughout the year, we also elicited their buy-in for the project’s goals and deliverables. To facilitate our interactions with business groups across the company, we created an advisory board consisting of representatives from key business groups, as shown in Figure 6.8. We originally planned a small board, expecting it would be difficult to find people who could commit the time to participate. However, we were surprised to receive many requests for membership on the board. For example, we gained members from internal groups like Human Resources and Corporate Real Estate – who were updating internal policies and practices to support employees of the future. We kept

Business Groups

Corporate

Future of Work Internal Advisory Board

Xerox Information Management

Xerox Innovation Group

Figure 6.8 Diagram showing how the advisory board was distributed across the company

Director of Operations

Corporate Human Resources

Corporate Real Estate & Global Systems

Director Human Resource Communications

Director Corporate General Services

Remote Work Team

Vice President Global Real Estate Strategy

Production Systems Chief Technical Officer

Xerox Global Services Thought Leaders

Xerox Office Group Marketing & Product Development

Program Managers

Project Leaders

Project Leaders

Work Practice Colleagues

Project Leaders

Program Managers

Individual Contributors

Research Fellows

Future Technology Manager

Research Lab 4

Research Lab 3

Research Lab 2

Research Lab1

Vice President

Vice President / Chief Engineer

Exploring Documents and the Future of Work

125

membership open, and ultimately ended up with more than thirty members, representing fourteen groups across Xerox. Functions and Activities of the Advisory Board Advisory board members provided input into the project research questions, tracked the project’s progress through quarterly status meetings, shadowed researchers in the field, and participated in surveys and concept brainstorming sessions. Since advisory board members had the opportunity to shape and contribute to the project, they felt a level of ownership and commitment to the project. A result of this commitment was that advisory board members opened doors within the company to attain more visibility for the project. They introduced the researchers to other potential board members and invited the team to present to groups in the company that we would not have been aware of without having the connections from the board members. In addition to increasing visibility of the project, the advisory board also helped us create resonance for the project within the company. In gaining exposure to questions and concerns from various stakeholders, we were able to not only answer their questions, but push them further than they might have expected, by challenging some of the assumptions that resided behind their questions. For example, many people on the advisory board assumed that people were printing less because of environmental reasons. However, we found that some study participants were printing less because they considered paper to be a hassle. These participants did not want the clutter that came from piles of paper in their office, and they found paper to be difficult to integrate into their digital workflow. By revealing these unexpected reasons for printing less, we challenged advisory board members’ assumptions about printing and opened the possibility of developing solutions to help people integrate paper and digital workflows more easily. Becoming familiar with advisory board members’ questions and assumptions also helped us tailor our findings and concepts in ways that piqued their interest, and drew them into the project, so they could hear not only the answers to their questions, but also gain exposure to other issues they might not have originally thought to ask about. Benefits of the Advisory Board The advisory board turned out to be mutually beneficial. The project team gained visibility and resonance as described previously, and also developed working relationships with other projects that we would not have otherwise

126

Making Work Visible

discovered. For example, during the year, one of the board members initiated a study exploring the possibility of placing a printer in a coffee shop. Given our interest in mobile printing, he invited us to collaborate with him on study planning and allowed us to conduct observations in the coffee shop during the study. This collaboration was mutually beneficial since he provided the framework for us to enhance our studies, and our observations enhanced his study results. Advisory board members also benefitted from participation in the project. The project provided research into questions that board members were already interested in for “free.” Board members told us that they joined the board because they needed to know the answers to the questions the project was exploring (to inform their own projects), but they did not have the resources to pursue the questions on their own. This is the main reason why so many people participated in the advisory board activities. Board members also created new synergies with each other. Some of these synergies eventually led to the creation of a new cross-company team to facilitate collaboration across several of the groups whose members attended the advisory board meetings. The advisory board also provided the opportunity for the project team to influence many levels of the corporation in parallel. Board membership included people at all levels of the company. Therefore, quarterly updates provided an efficient way of connecting with people at different levels within the company. Updates often led to invitations to present to other interested groups within the company. In fact, at the end of the year, the project was considered to be valuable enough that we were asked to present the results directly to our CEO. By staying attuned to various functions across the company through the advisory board meetings, we were ultimately able to focus the study and analysis in ways that resonated enough with the company to garner attention, while also creating evocative/provocative views of what is possible. This then allowed us the opportunity to create a bigger impact within the company. (See Watts-Perotti et al., 2009 for more discussion of the advisory board.) Contributing to Intellectual Property and Further Research Projects The project team created many new concepts and ideas for ways that work could be improved. Many of these concepts were developed into invention proposals aimed at increasing the company patent portfolio and creating new services and products for the company to build (see Colombino et al.,

Exploring Documents and the Future of Work

127

this volume, for another example). Several concepts were already being developed within the research organization at the time this chapter was written, and several others were marked for exploration in upcoming years. In addition to creating intellectual property, the project also contributed to the creation and development of new research projects within the research organization. One project that followed the first study focused more deeply on the question of how printing and electronic information should change to better accommodate mobile work styles. This project was tied closely to a technical team who had the goal of proposing and prototyping new offerings for mobile services (see Schiano and Bellotti, this volume for another example). By working closely with a technical development team, we made an even greater impact on new solutions and services that might be offered by the company. Another subsequent research project was established to better understand how electronic information is changing and how people will need to interact with that information. The project was informed by data from the original future of work study describing mobile and remote work styles, and participants’ struggles with electronic information.

Conclusion The future of work project was designed to explore how work practices are changing in response to evolving work-based trends like the movement toward working in distributed virtual teams, the use of mobile technology, and the ability to work virtually anytime, anywhere. We explored these topics through the use of interviews, observations, diaries, and collages. The study was shaped by our interactions with an advisory board, which in turn increased our visibility and impact within the company. The project gave rise to new intellectual property and research projects examining new business opportunities. The team also began to impact business strategy within the company. The team gave multiple presentations to upper management throughout the year, culminating with a presentation to our CEO and senior management committee. Information from the presentation has since been used to inform corporate strategy planning sessions. By creating buy-in for the project across the company, and moving beyond qualitative findings to envision explicit corporate implications from the study, we have made several kinds of impacts, and therefore facilitated innovation within the company.

7

New Ways of Working The Implications of Work Practice Transitions Mary Ann Sprague, Nathaniel Martin, and Johannes A. Koomen

For decades, Xerox Creative and Technical Communications group (CTC) has been the internal Xerox unit that produces most, if not all, documentation for Xerox products. The organization writes, translates, and publishes documents. They deliver manuals, on-line and embedded help files, and training materials to both equipment end-users and service personnel. This chapter describes a work practice relationship that has endured across multiple studies and an external client engagement. Our relationship with CTC began in 2003 when the research group was studying publishing practices and approached the internal organization as the subject of the study. After an in-depth study of the authoring and publishing areas of the CTC group, we were asked to continue the study by reviewing an authoring system that was being developed and evaluated. Following a long tradition of document innovation, CTC was trying to restructure their document creation and lifecycle management to reduce development times and cost of producing documentation. We uncovered a mismatch between the expectations of the managers and the workers with these new tools and processes. In 2006, members of our team did a follow-up study to assess their progress. At this point, CTC was working diligently to implement an authoring system similar to the one we had seen earlier. As part of our standard practice, we presented our findings to the collaborators within CTC. They found the material unsurprising, as we had hoped, but useful to explain their work to outsiders and management. They had a long history of innovation in documentation development. We learned that they were one of the first organizations to use SGML (Standard Generalized Markup Language), and they were in the process of offering their services beyond Xerox. With the creation of the new service organization, the organization was also looking for ways to provide documentation 128

New Ways of Working

129

and publishing services to outside customers in a cost-efficient manner. The project we were supporting provided input into the development of new technology for authoring, and CTC asked us to help them understand how this new technology might affect their work. We found that there was a divergence of opinion on what this new technology would do. We developed an ongoing relationship with the sales and service organizations based on their offerings. We were asked to provide work practice studies of their customers to help them supply those customers with excellent document services. One of those customers was a major car rental company. Our relationship with CTC continues as of this writing in 2009. Throughout these various phases in our relationship with this organization, several different members of the work practice group were involved in these studies. The studies provided a number of underlying benefits because of the ongoing relationship. Ethnographers within the work practice group were able to leverage the knowledge from team members who worked on previous studies. The amount of time to become familiar with the organization was greatly reduced because of past studies and documented processes, and the ethnographic team members were able to build on the existing rapport and trust that the CTC organization had with the work practice team over the years. Our practice of reviewing our work to ensure accuracy increased that trust. This trust lead them to ask us to return for additional study whenever they considered overhauling their document authoring system. Eventually, they asked us to go along with them to customer sites to help ensure that their solutions would truly support their customer. These are just a few of the internal benefits of having a longstanding relationship with a set of participants that are studied. This chapter provides an overview of the history of the various studies performed and looks at the changes in the organization as well as the transitions that happened with the ethnographic team during those years.

The Relationship Begins In 2003, as part of a study on contemporary publishing practices chartered by research, we engaged CTC to understand their work practices. The research goal was to understand the common workflows in publishing in order to find ways to reduce the size of print runs, thereby increasing the chances that these runs would be printed on digital presses. Our focus was on the customization and reuse of information in the publishing cycle, which we believed would lead to small runs of similar, but not necessarily identical books.

130

Making Work Visible Globalisation & Strategy Authoring Service Documentation Develop text and graphics

Plan documentation to minimize cost Customer Documentation

Translation Customer Services Group Interface to customers

Translate text to desired languages/locales

Publishing

Distribute by CD, Web or Print

Figure 7.1 Management overview of CTC organization

We first engaged CTC through their senior management, relying on their introductions to gain access to the workers. We conducted six openended interviews that ranged from one to one and a half hours of the key members of the CTC management; we focused our attention on the organization, their interactions, their technology use and the group itself. We learned that CTC contains four groups: Strategy, Authoring, Publishing, and Translation. We focused on authoring and publishing to gain a clearer understanding of the work practices needed to write and distribute the documentation. We concluded our study of CTC practices by looking at a single product from end to end. We videotaped all of our interviews and analyzed the videotape. The analysis provided a broad overview of the organization from the management perspective and showed how each of the smaller organizations interacted (See Figure 7.1). The managers we interviewed also suggested people from each of their organizations, laying the groundwork for the next layer of interviews. This information was shown to the management team and corrections and other information was collected.

New Ways of Working

131

During the discussions, the history of the organization and how their role in the company and their technology had changed emerged and was also captured, showing the flexibility of the organization as it responded to the company’s needs. This information provided us with the overview of the process of developing documentation needed to continue to study the processes in depth, but it also helped CTC to communicate their processes and needs with other organizations within the company. The study provided them with a documented process that allowed them to explain their process in visual terms to new upper management after a recent reorganization. We learned the distinction between customer documentation and service documentation, and that two sets of authors support different audiences, end-user customers and service support personnel. We also learned of the consulting work that helped the product teams understand and plan for globalization (including translation), cost-effective format and delivery (e.g., CD vs. paper), cost-effective publishing techniques, and management of service kit complexity. The publishing group was responsible for planning and delivering the final documentation, in the required media, managing external vendors, and fulfillment operations. Documentation often happens late in the product development cycle, since it cannot be completed until the product has reached a stage in which its features and behaviors are stable. The temptation is to wait until the product is completely developed to document its features to ensure that documentation does not need to be redone as features change. However, documentation may take a substantial amount of time, and few organizations can afford to hold up the launch of a product to wait for documentation to be completed. To balance the competing goals of reducing rework and timeliness, planning for documentation needs and expectations are done throughout the development stages. Documentation development costs have been difficult to explain to product development teams often leading to budget cuts. In order to meet a scheduled delivery date and reduce expenses from delivery delays, documentation plans are often reduced or changed. Authoring and Publishing In-depth Studies Because the project’s goals were to develop techniques by which the content of books could be specialized to smaller groups to reduce the size of print runs, we focused on the Authoring and Publishing groups for the next phase of the study. This phase focused on a more detailed understanding of the customer and service documentation process and the publishing and

132

Making Work Visible

fulfillment process. We focused on publishing, since producing and distributing short print runs was the responsibility of the publishers. However, another limitation on shorter runs was the efficient production of material to be published. Authoring produced the content to be printed, so efficient production of material was necessary for short runs or the unit costs would grow. For example, it was prohibitive for an author to spend years producing a book that would be read by only a few people unless those few people were willing to pay an exorbitant price for the information. Twenty-seven additional open-ended interviews and observations were conducted in both the authoring and publishing groups in order to understand their work practices and their connections between groups. We focused on those people who were actually writing the documentation and those people who were responsible for managing the production and delivery of the material. These included people identified by the managers in the previous phase, and the overview of the system helped us make sense of the stories that were heard. The different groups included the parts list group, illustrators, customer and service authors, diagnostic authors, product managers, fulfillment managers, warehouse managers, and test print printers. In each case the interviews were videotaped (Wall, Brun-Cottan, 1995, BrunCottan & Wall, this volume) and transcribed for analysis. Figure 7.2 shows a sample of the interconnections between the different groups within the customer documentation process. Service documentation was delivered electronically. The service technicians used a Portable Work Station (PWS), which was a service specific laptop that contained service documentation test procedures for those machines. Completely new versions of the documentation were delivered on CDs that were loaded onto the service laptop; updates could be delivered over the internet. The documentation groups interacted with the product programs to produce effective documentation for both customers and service staff. These groups specialized in different areas, such as service authoring, customer authoring, parts lists, illustrations, and electronic documentation support. Interactions and dependencies among these groups, the tools they used, the challenges they faced, and the processes they followed were captured in illustrations. The representations (Wall, Mosher, 1994) captured the specializations in software or process for individual products. For example, on one product, the help files were electronically stored within the device for on-device retrieval, rather than delivered on paper. The publishing group, as shown in Figure 7.3, is responsible for producing and distributing physical documentation. Production used negotiated

1

2

Illustrators create needed illustrations for documentation

Illustrations

3

Customer C Documentation goes to publishing

Customer Documentation to be translated

4

Customer Documentation

Curriculum

Authoring

Requested illustrations for customer documentation

Curriculum creates customer documentation in many different tools

Documentation Requirements & verification n

Figure 7.2 Customer documentation workflow

• User guides • Online help • Customer training

Product Development team provide requirements and work closely with curriculum authors to create needed documents:

Product Development Teams

Language Services produces the translated documentation and software screens

Translation

Publishing Services produces the desired media for the product teams

Publishing

Distribution

Document Advisory Office consults on graphic design for printing

Figure 7.3 Publishing organization workflow

Distribute just-in-time kits via Pull, Push or Subscription distribution methods

Publishing bli hi P Project j t Managers work with Program teams to consult on kit complexity and pieces.

Consulting

Warehousing W h i contracts

Production

Only store minimum, Use just in time production to fulfill needs, print on demand

Inventory

End to end print flow expertise ((Bid, quotes, pre-press, preflight, blue line samples, quality checks)

New Ways of Working

135

contracts with offset printers, but other times sent proposals out for bids. Distribution of documentation followed several different methods. For example, distribution of the service documentation is done through subscriptions where a technician subscribes to the documentation for the machines he or she repairs. When information on that machine changes, he or she automatically receives the update. The publishing group contracted with printers to produce the physical documentation (both books and CDs), and with warehouses to distribute the documentation. They coordinated the activities between the printers and warehouse operators. The publishing group was responsible for insuring that there was always sufficient physical documentation on hand to support any potential sales. They needed to insure that the materials were available when needed, but large quantities were not remaining when materials went out of date. Materials could go out of date either through feature changes of the machine or because the machine was discontinued. There were also complex cost trade-offs to make. For example, digitally printed materials could be produced when needed, reducing the risk of over production, but individual printing costs more than traditional offset printing. The complexity was exacerbated by the differences between customer and service documentation. Customer documentation was written and distributed in a manner similar to traditional books with the difference that the books were delivered with the machines. Groups of documents would be ordered for each new batch of machines. Translation into multiple languages meant that books and labels in different languages needed to be delivered for groups producing machines destined for different countries. The Lifecycle of One Product’s User Documents Ultimately, we followed the development of documentation for a large color digital press through the entire process allowing us to make more sense of everything we had seen. We were able to study the entire process, which takes years, by interviewing people who were in different stages of the development for different versions of the machine. The product group continually develops new versions of the machine, with multiple versions in the pipeline. We were able to watch people who were in different stages of the documentation process. By watching people in different stages, we were able to see how they used the information in the documentation for previous versions. We also found a variety of processes and knowledge that were required to provide successful documentation deliveries. An extensive knowledge of the products was developed by many of the authors in their

136

Making Work Visible

respective areas, making them as knowledgeable as the engineers developing the products in some cases. It was this machine knowledge that helped ensure the accuracy of the information that was received from the engineering community and made it efficient and usable by the customers as well as by support personnel, both in the information provided and in the delivery methods. The detailed understanding of the practices in both authoring and publishing was used to support a group that was developing technology for short run books. For example, the complex trade-offs between the cost of storage and the cost of printing digitally when needed was exposed by the study. Digital printing is more expensive than offset printing, but long offset print runs require storage and allow spoilage as documents become outdated. In this phase, we discovered how complicated the process of developing and delivering documentation was from the point of view of the authors and publishers. The authors gather information from subject-matter experts, then tests the information against samples of the product. They then write the documentation, and review it with the development engineers. For example, the authors learned the technical information from the engineers who designed and implemented the products. As different sections of documentation were completed, the authors sent sections back to the engineers for review and revised the documents in light of those comments. Lists of available parts and part numbers, illustrations showing part placements and the related repair procedures were also created, cross-referenced, and reviewed for accuracy. The work practice study produced a detailed description of a stateof-the-art custom document production facility, and provided a means for CTC to share its work practices with outside organizations. The study also enabled better insights on the work practices of the organization and was used to guide the development of technology that would help other facilities improve productivity. Unfortunately, the proposed technology project was terminated because of loss of support from the business group that was supporting that project, so the need for further work practice was deemed unnecessary. However, the resulting technology was transferred to CTC who used the insights from the study to develop a new authoring system, which they hoped would help them roll out their service offering. The baseline information we had gathered earlier in the year provided us with the tools to understand potential risks and rewards for the change in their work practices. The risks of changing work practices were the costs for training and technology and the potential failure of the project leading to

New Ways of Working

137

failed efforts: the workers were used to individual authoring and it was not certain that they could adapt to develop documents by piecing them together. Such adaptation would require writing the individual elements of the document in such a way that the elements could be used in multiple documents. They would also need to be able to find the elements when they were needed to create a new document. On the other hand, if they could adapt to piecing documents together from existing elements written for previous documents, substantial cost savings could accrue. This research had two results: it informed the group that chartered the study and it provided CTC with representations of their work. CTC trusted our methods because of the care we took to ensure that our information was correct before passing it on. Their trust led them to ask us to continue our study for their own purposes. After the conclusion of this study, CTC asked us to study an authoring system they were developing because of our past knowledge of the organization and the prototype authoring system. This phase of the study was over a year after the end of the first phase of the project and shifted to include different members of the ethnographic team, who were brought up to speed on the organization by the previous study results.

A New Way of Authoring Due in part to our past working relationship and understanding of the CTC organization, we were engaged to observe CTC workers as they tried out a prototyped authoring tool that was designed to support CTC as they supported customers outside Xerox in their new service offering. It allowed a writing tool to be connected to a document management tool. By focusing on the connection between authoring and document management rather than the function of the tools, it allowed the system to use whatever tools the customer was already using. Our experience as ethnographers and our understanding of the baseline work practices gave us insight into the goals the workers would have been trying to achieve in a more natural setting and allowed us to explore how the new system would have fit into their new setting. In a more natural setting, the workers would have been trying to develop a document that captured the information and presented it effectively. In the lab, they were asked to perform specific tasks that exercised the features of the system. For example, they were shown how to include a paragraph either by pointing to that paragraph, or by inserting the paragraph into the text. In the first case, if they edited the original text it would change everywhere it was used; in the second, it would change only where it

138

Making Work Visible

was edited. Because we had studied the workers, we could see that the text changed when aspects changed globally, such as the copyright notice or the company logo. On the other hand, text that needed to be manipulated to fit into the document at hand would need to be edited to the particular context. The methods we followed were somewhat different from the methods we used to gather information on the baseline. Instead of interviewing the workers in their place of work as they described what they did, we interviewed workers in a laboratory environment as they followed a script that exercises the various features of the system. An ethnographer with a technical background studied the proposed tool to learn how it operated and then observed the initial user tests. A set of CTC authors ran through a set of test scripts using the prototype tool. The authors’ comments were collected as well as subsequent analysis of the entire prototype test. Fourteen hour-long sessions with authors and their managers were videotaped and analyzed. The results and insights from this controlled observation were summarized and presented to the CTC management team. Several results from this phase of the study surprised us. We had expected the authors who used the system to have concerns about the effects the new authoring tool would have on their work practices. For the most part, the authors were enthusiastic about the system. We had expected that the managers would be enthusiastic about the potential savings from using the system and, as expected, they were. However, the enthusiasm of the authors and their managers came from very different views of the system. Both groups could see that the system changes would help them with their jobs, but the groups envisioned different changes to their work practices. In their current work practices, authors were responsible for creating the documentation from beginning to end of a specific document. In the process of developing the documentation, they researched the product, wrote the documentation, and updated documentation as the product changed. These updates often led to different versions of the documentation with different descriptions of the machine. Managing the information gathered and the multiple versions of the documentation was difficult. For example, because documentation must be written before the design of the machine was stable, features that were intended in the product sometimes changed before the product was launched. Changing features generated a multitude of nearly similar documents making tracking difficult. Even worse, the sources of the information in the multiple documents must also be managed to insure accuracy.

New Ways of Working

139

The original system stored documentation in a linear layout that would later be formatted visually. For example, a Repair Analysis Procedure, a kind of fault tree used in all of the service documentation, was written in a natural outline fashion, and was formatted in the way the service technicians expected. The proposed system provided a centralized server on which documents could be stored coupled with a system for writing the documentation. The new system differed from the current system in that the structure of the documents was stored and the content of the document was associated with places in that structure. This meant that documentation content used across different products could be written once and shared in different documentation structures for multiple products that shared the same features. In addition, the new system allowed authors to access pieces of documentation content written by other authors so they could share content pieces with each other. The authors saw the system as a tool for improving backup and version control. The current system was cumbersome to backup and required authors to manage their own version control manually. The ability to store interim versions of their work in a controlled environment was attractive. Their managers, however, saw the system as a means for increasing reuse of authored material. They saw that the system would allow the authors to write reusable elements that could be shared among the authors and hoped that the ability to share the information would reduce the number of times similar material would need to be rewritten. They saw a much more dramatic change to the authors’ work practices than the authors did. The task of writing the information was seen as separated from the task of collecting the material into a cohesive document. Ultimately, the managers hoped that new documentation elements (or content pieces) would be written only rarely; in most cases the documentation element could be extracted from earlier work. At worst, the new element would need to be lightly edited to fit its new context in the documentation structure. The managers’ work practices would also have to change to reflect the change in workers’ practices. Instead of managing individual workers creating individual documents they would need to manage a team producing a collection of documents. For example, it would be much more difficult to determine what caused a delay in producing a document: was it the author of a piece of the document or the worker piecing together the elements. The work practice research of the proposed tools uncovered a mismatch between the managers and the workers that indicated the need for extensive retraining for both the authors and the managers. Partially as a result of this discovery and partially due to funding restrictions, the CTC system was also terminated. The long-term engagement with the CTC gave them sufficient

140

Making Work Visible

confidence in our results to incorporate them into their decision making. The research and development projects we had been supporting had ended, terminating our study as well.

The Relationship Continues: Technical Publishing as a Service Offering The year 2006 was a pivotal year for CTC. The organization chose and committed to a new technology for internal use, and started delivering technical authoring & publishing services to customers outside Xerox. All of the experience we had developed while creating the baseline and while studying their attempt to move to a new authoring system could be brought to bear to help them engage one of their early customers outside Xerox. In the course of the spring and summer, CTC negotiated a contract for technical document authoring, publishing, and management with a major car rental company (MCRC1). The contract had a start date of January 1, 2007, because MCRC had an existing contract for these services with another company, DCR, and that contract was expiring at the end of 2006. By the time the contract was agreed to and signed by all parties, there were less than four months left to take over these services as well as make significant improvements as stated in the contract. As CTC management had become a strong proponent of work practice, they enlisted our help to understand the current practices and needs of this new customer. We were asked to investigate both the general document needs of the MCRC publishing group, and the services provided by DCR. DCR maintained and published the user manual for MCRC’s mainframe rental-management system, and published a weekly news bulletin distributed to all MCRC’s personnel. DCR had a small number of employees colocated with the MCRC publishing group, as well as a small number located in another part of the country that were dedicated to supporting MCRC. Our work practice team conducted two field studies in October: a threeday visit to MCRC headquarters, and a two-day visit to the DCR location. Although their contract was being terminated, DCR management and their employees were very generous and cooperative by allowing us to come in and “look behind the curtain” to see their actual operations. We interviewed 17 employees and managers at the MCRC site and 8 at the DCR site. 1

Names are fictitious, to protect both the innocent and the others. . . .

New Ways of Working

141

The study focused specifically on the people and processes around creation and management of the weekly news bulletin and the system documentation. We video-recorded all interview and observation sessions, transcribed the audio files, represented the workflows graphically, and documented the practices around these two document processes. Because the weekly news bulletin was the first publication to go “live” after Xerox’s takeover, a significant portion of our analysis revolved around documenting the processes and people involved in publishing it. Not surprisingly, although the number of people directly involved in its creation was small (three at MCRC, two at DCR), the process was very time consuming and entailed quite a bit of iteration between the two companies. To balance our study of the news bulletin creation process, we also found it important to collect the end users’ point of view. Process improvements designed to help the three people involved in its creation might negatively impact the thousands of readers in the company. So in November we visited one of MCRC’s many airport rental counter and back offices, where we interviewed managers, counter agents, back office, and maintenance personnel. The newsletter consisted of a number of news items, each one typically in the form of a short summary with a link to the full story. The items were divided into three groups: items that required an action on the part of the reader, items that required immediate attention, and items for information only. The newsletter was also targeted for three different groups of employees, namely managers, employees providing customer support, and employees providing vehicle support. Not all items in a particular newsletter edition were necessarily intended for all three groups. A newsletter edition was published as a web page with dynamic content depending on variations in the URL to accommodate the three unique audiences. The URL was sent to about 5,000 employees via corporate email. Each rental office (such as in airports) was required to print and post the newsletter in a place where all the local employees without access to corporate email (another 7,000+) could read the bulletin. Corporate policy dictated that all targeted employees read the bulletin. On the MCRC side, a single “coordinator” spent the better part of a day every week receiving news items from various business departments, ordering and grouping them, and deciding what part of each would go in the newsletter vs. in a “full story.” When the list of items was ready, the coordinator sent them to a DCR employee, the “creator,” who then crafted a web page with the given content. The creator typically started with the previous week’s edition, removing that week’s items, and inserting the current

142

Making Work Visible

MCRC

DCR 15

3 2

4

Sources 1 Coordinator

11 Creator

5

12

8 16

6

13

9 7

Link Checkers

10

14

Audience Review & Approval

Publisher

Figure 7.4 The weekly news bulletin creation workflow

week’s. The creator worked in a simple text editor, modifying raw HTML. When the draft page was complete, the creator would email it back to the coordinator, who then printed it and distributed it to two managers who were responsible for review and approval. The coordinator, having received corrections either via conversation or email, sent them to the creator for revision of the draft. This cycle was repeated until the draft was approved. The coordinator notified the creator, who then sent the approved draft via email to another DCR employee, the “publisher” (See Figure 7.4). The publisher posted the draft page on the test server, and linked it into a manually maintained table of contents, also on the test website. Then, whoever was available on the DCR team (both local and remote) was enlisted as “link checker.” Link checkers went through all online issues of the news bulletin and verified that all the links in the news items, which typically referred to files (Word, PDF, etc.) on a shared drive, were still valid. The web page (and the internal web site) was under control of DCR, but the files on the shared drive were owned by the business department where the new items originated. Over time the departments were known to move, rename or delete these files, which resulted in broken links in the newsletters. Once the link checkers reported back to the creator, problem links were fixed or disabled, and the publisher was notified of the final revision. The publisher posted the page and the revised index on the live website, and notified the creator with a URL of the new page. All of these iterations were completed preferably by Thursday at 5 pm, but often much later than that. The creator then notified the coordinator of the completion, who sent a message to a corporate email list to announce the availability of the new issue.

143

New Ways of Working Table 7.1 Ethnographic findings and their impact Issue / Finding

Impact

Stations print and post the Manager’s version of the bulletin, including articles not intended for customer and vehicle support personnel;

This could compromise confidentiality of certain items intended for managers, and could lead customer or vehicle support personnel to lose interest or skip over items intended for them. Manual and time-consuming process could lead to inadvertently skipped articles.

It is not easy to print the bulletin for posting – all the associated “full story” articles must be opened and printed separately; Bulletin pages may take a very long time to load into the browser, which we attributed to occasional high-resolution (and therefore, large) images in the newsletter; many field stations have slow internet connections. Also, limited time was available to read the bulletin when it was released on Friday mornings, since Friday and Monday are the busiest days at a typical airport station; Old equipment makes for slow browsers, and screens are often too small to show the full width of the bulletin, leading to a lot of horizontal scrolling; Search is impossible – the only way to access an old issue because of a particular topic is to remember on what date it was published and use the table of contents. Personnel in the field are not always very skilled with computers, and introducing changes in the format, organization and/or the delivery of the bulletin may encounter resistance and steeper than expected learning curves.

Frustration and time pressures could lead to lack of compliance with corporate mandate to read the bulletins.

Slow and cumbersome equipment exacerbates time pressures.

Customer support personnel often need to look up facts in response to a customer question. They end up either admitting they do not have the answer or fabricating one. A significantly changed newsletter could be disruptive, lead to a lot of questions to management or IT support, or reduce reading compliance.

From our interviews and observations at headquarters we identified a number of issues related to the creation process, some of which (such as the broken links issue) were already known to management. Several other problems were heard and observed during our very limited exposure to end users (see Table 7.1).

144

Making Work Visible

In addition to a report of our findings, we also managed to put together a prototype for both constructing and presenting the news bulletin, built on top of DocuShare, the Xerox Document Management System. Using a custom object to represent news item summaries, that also contained links to associated documents and begin and end dates for publication, the bulletin was presented as a separate web page, independent of the standard DocuShare interface. It looked identical to the existing bulletin, with the same trifold organization, but it had built-in search capabilities, and was dynamically generated with standardized graphics and layout, eliminating the typical errors that often occurred with the old manual editing processes. The news item objects could easily be created and managed directly in DocuShare by the editing team without the help of technical personnel. The newsletter reader application (web service) retrieved appropriate newsitem objects, based on date or keyword search, and automatically assembled and formatted them in the way the end users were accustomed to. Ideally, a work practice study provides sufficient time to do a thorough job of the analysis of all the data collected during the site visits, and prepare and deliver a report in time to adequately inform and support the design and implementation of a service solution. Alas, in the corporate consulting world, project schedules and budgets are generally based on how much time and money is available, rather than on what needs to be done to do the job properly. Even though we processed well over 40 hours of field data in a short period of time, by the time the report and the prototype were delivered in December, many decisions about the final Xerox solution had already been made. The project implementation manager also decided that there was insufficient time to complete, test, and launch the prototype newsletter solution. MCRC did use DocuShare to manage all their documents, including the creation and consumption of news bulletins. Fortunately, the implementation engineer assigned to the project was able to incorporate many of the study’s suggestions and recommendations, and eliminated some of the identified pitfalls over the course of the first few months of operation. One lesson from the MCRC experience for CTC was that it’s a major challenge to provide authoring services to external clients in a competitive way. Although CTC has many talented and experienced authors, it was clear that more than mastery in English and a production-oriented technology environment was needed to be successful. As documented earlier, CTC authors work closely with the product engineers in order to develop relevant and correct documentation, and they are very much Subject Matter Experts (SMEs) themselves. In order to provide authoring services, CTC has a dilemma: either turn CTC authors into SMEs in the client’s area of business, or take over (some of )

New Ways of Working

145

the client’s own authors and make them CTC employees. Either way, there is a steep learning curve to learn the client’s subject matter or the CTC culture and technology, which service contracts typically do not accommodate. Interestingly, CTC has had far more success providing its translation expertise as an external service. No doubt this was due, at least partially, to the fact that unlike authoring, translation does not require as much domain knowledge or expertise beyond terminology. With the growing business naturally came additional challenges, such as a trend from large jobs involving many documents and many languages, to small jobs of just a few pages and only one or two languages. CTC recently asked us to perform a work practice study investigating their practices around these small translation jobs, which are difficult to do profitably using the same project-management overhead as with larger jobs.

Conclusion Initially, we were chartered by the research group to understand how shortrun publishing could be managed, but the nature of the information shared with the CTC led the participant organization to ask for help with their own processes as they transitioned to new technology. This led to a beneficial relationship over the years as the CTC relied on our ethnographic expertise to give them insight into their process and technology changes. Our support for these studies turned out to be somewhat paradoxical because we uncovered unexpected impacts for the people who would be using the technology. That is, though the technology was intended to support those who would use it, the changes to CTC practices were found to cause significant process changes that had not been anticipated or planned for. The entire work process and the resulting impacts to the organization had not been fully understood by either the developers or the CTC before our study. Partially as a result of our report, the study of the technology transition we were to support was terminated. It seemed that the project was over. However, once again, the information we had uncovered proved to be crucial. CTC continues to work towards developing the authoring system, this time by focusing on the work practices of their authors and the nature of the artifacts they produce. By organizing the information in their documents in a more regular fashion, they developed a system that was more amenable to the kind of authoring the system designers had envisioned. Understanding these practices helped as they moved to offer their authoring and publishing services to customers outside Xerox. Our group was invited

146

Making Work Visible

along as partners as they worked with external customers as well. Since then, we have begun training several of their employees in this ethnographic technique to expand the interaction with external customers (see Chapter 19, this volume). Our initial study in 2003 introduced the CTC to the work practice methodology and enabled the CTC management to use those representations to explain their work to other people. From this study, a working relationship with CTC management team for several subsequent projects helped to further the relationship in which CTC contacted the work practice group for additional studies. The development of the interactions enabled the work practice team to easily step into reestablish connections and update previous studies with new information and document new processes. The CTC was able to rely on the work practice team for a level of analysis that enabled them to improve their own technological efforts. Since our initial project in 2003, the work practice team was able to respond to the requests of the CTC with a firm understanding of the history and workings of the organization. Previous work practice studies provided a better grasp of the history of the group and helped the work practice team to evaluate the current work in that context. Such a strong rapport and understanding of an organization aided with the quality of the subsequent studies and reported insights. The ethnographic approach has provided important insights into the changing work practices that were a fundamental part of this organization. These insights were applicable to a wide variety of goals over the years, allowing the information to be redirected as needs changed. Our relationship with CTC became well established over the years. CTC has come to see the value of ethnographic methods and the information they provide. They have continued to look to our group for assistance as they developed new processes and technologies over the last few years. Throughout, the experience, the techniques we employed, the understanding of the CTC organization, and the knowledge the work practice studies produced provided insights that could not have been reached in other ways.

8

Behind the Scenes The Business Side of Medical Records Nathaniel Martin and Patricia Wall1

In 2007, Electronic Medical Records (EMR) systems, though not new, were clearly the future. Yet, most medical records in U.S. physicians’ offices were on paper. Hing et al. (2007) report that only 29.2% of 2,117 survey respondents had any EMR system and only 12.4% had fully implemented an EMR system (i.e., no part was paper). An earlier study found that EMR use was higher in hospital Emergency Departments (31%) and Outpatient Departments (29%) than in physician practices (17%) (Burt and Hing, 2005), but penetration was low in all areas. Many have studied EMR systems and their use in doctor–patient interactions. Heath and Luff (1996) found that medical practitioners continued to use paper records along with a newly introduced EMR. Other studies (Clarke et al., 2001) have examined the use of an EMR in medical exams and issues an EMR poses in physician–patient interactions (Ventres et al., 2005). Martin et al. (2005) studied the issues involved in the integration and implementation of an EMR system in a large hospital. Workflow changes required by physicians using an EMR have also been discussed (Puffer et al., 2007). Fitzpatrick (2000) studied the implications of the use of paper records for EMR systems from the point of view of the providers. No studies focused on the work of practice management. Practice management deals with the business side of the clinic arranging interactions between the patients and clinical staff and insures payment for those interactions. This typically involves personnel at a front desk who schedule and process records for patient visits and back office personnel who ensure payment for the visits.

1

We would like to acknowledge and express our appreciation to the healthcare sites who allowed us to study their workplaces. We would also like to thank our colleague Maryann Fuhrman who accompanied us into the field and helped us analyze the data, and our colleagues Paul Austin and Mike Kehoe who translated our insights into technology.

147

148

Making Work Visible

Figure 8.1 Formation flow across the clinical enterprise

This chapter describes results from a two year ethnographic study of work practices around medical records at a Community Health Clinic. We studied two groups of workers: those who worked at the front desk and those who worked in the Accounts Receivable department. We found that although practice management required the coordination of the two groups, the demands of each group’s tasks complicated their interactions (see Figure 8.1). By describing the work practices of these two groups, we highlight a few ways in which their transition to EMR would impact change.

Methods It took several months to find a fieldsite, but a breakthrough came when we met the Director of Medical Records at a local Community Health Clinic2 at the New York Health Information Management Association (NYHIMA) conference. She invited us to meet the clinic’s senior staff, and they agreed to let us study them. They thought that an improved understanding of their

2

The clinic is a free-standing primary health clinic, which includes Family Medicine, Obstetrics/Gynecology, Pediatrics, Dental, Vision, and WIC services. The clinic, founded in 1973, has 140 full and part- time staff members, including seven physicians, four dentists, two optometrists, two opticians, four nurse practitioners/physician assistants, two certified nurse midwives, two dental hygienists, eighteen licensed practical nurses, eight dental assistants and other allied health professionals, including nutritionists and outreach workers. They have around 13,000 active patients, each of which had a paper medical record; approximately 6,000 inactive patients also have paper medical records on site.

Behind the Scenes

149

work practices would help them transition to their new EMR system.3 We assembled a team consisting of ethnographers and technologists. In our experience, including technologists in fieldwork enriches our results and streamlines the transfer of findings to those exploring technology solutions. When we shared our findings from our initial study of the Medical Records department with the management team, they suggested that we study the practices at the front desk. They said that this was one of the most difficult areas in the clinic. During the study we determined that the difficulties they were talking about were errors introduced by interruption. The study of the front desk in turn, led us to study the Accounts Receivable department which depends on activities that take place at the front desk. The primary fieldwork consisted of open-ended interviews and observations of staff of the front desk and the Accounts Receivable department; more than ten hours of videotaped observation were collected. The videotapes were transcribed and used to create representations to aid in the analysis of the work. (Figures 8.3 and 8.4 are examples of these types of representation.) These representations were shared with the research team to promote exploration of potential technology solutions for records management. We also took the representations back to the site in order to share, verify, and update our findings (Brun-Cottan and Wall, 1995; Brun-Cottan and Wall, this volume), and also to explore and prioritize areas for further technology development. We reviewed the representation to ensure that our descriptions were unremarkable to the participants (Tolmie, this volume) without imposing a particular perspective on the work (Sharrock and Button, this volume).

Observations The front desk and Accounts Receivable communicate using the encounter form and the schedule. The encounter form is prepared and managed by the front desk, but it supports communication between the physicians who provide care and the Accounts Receivable department that charges for this care. It contains demographic information on the patient, the procedures 3

We studied the paper medical records first. We found a complex ecosystem surrounding these records, which requires significant effort to maintain but which was invisible to most in the clinic. The clerks who manage the records balanced a set of tasks involving medical knowledge and intuition in a fast- paced environment. (Martin and Wall, 2008). As a result of this work, Xerox initiated a more targeted project to scan and capture data from paper medical records for insertion into EMR systems.

150

Making Work Visible

Figure 8.2 Patient Encounter Form (Care Slip)

performed during the visit, the diagnoses reached during the visit, additional appointments the patient must make either to follow up or as a referral to a different provider, and billing information (see Figure 8.2). The encounter form was, in part, a result of the clinic’s hybrid system in which the medical records were paper but the practice management documents were computerized. An encounter form records the diagnoses and treatments given the patient and was used by the Accounts Receivable department to prepare insurance claims and bills. This information should match the notes the clinicians make in the medical chart during a visit (called a progress note). The Practice Management Software (PMS) supported work at both the front desk and Accounts Receivable. The front desk clerks used the schedule in a number of different formats: a daily schedule of the entire clinic to check people in, a list of pending patients to prepare for the visit, and a long-term schedule for doctors to arrange new appointments. The Accounts Receivable clerks used the PMS to search for a particular patient at a particular time in order to associate that visit with the encounter form and track payments against that visit.

The Front Desk The front desk ensures the smooth flow of patients into and out of the clinic, capturing information critical to the operation of the clinic. But the workers

Behind the Scenes

151

cannot control the flow of this work; their activities are subject to interruptions4 from people calling or arriving at the front window. Physically, the front desk is arranged with five work stations facing sliding windows that open to the waiting room. There is a phone on the desk next to each work station and a printer on a counter against the back wall. The arrangement is designed to accept interruptions: people arrive at the window and call on the phone. At the work station, the workers at the front desk schedule patient visits, prepare the paperwork for the visit, check the patients in when they arrive for the visit, and check them out at its conclusion. The encounter form reflects the process from the patient’s point of view, however, the worker’s experience is different. Here, we first describe the process view, then describe the worker’s view. It turned out that the difficulties we were told about were caused by challenges that were apparent only from the worker’s view. Scheduling a visit prompts the creation of the encounter form. To schedule an appointment, the worker at the front desk first calls up the patient’s record in the PMS and enters an item in the schedule of the patients Primary Care Physician (PCP) as indicated on the patient’s record. Same-day appointments (or sick visits) are scheduled by the triage nurse who notifies the front desk of the appointment by printing a form on the printer (this is referred to as a “tickle”) at the front desk. About half of the day’s appointments are scheduled that same day. The front desk prints the encounter form and verifies the patient’s insurance from web sites provided by the insurance company to prepare for the patient visit on the day of the appointment (see Figure 8.3). Verifying insurance for a same-day appointment could be difficult due to lack of lead time. If the patient arrives before the paperwork is ready, the front desk worker has to scramble to print out the encounter form and complete the insurance verification in time for the patient to make their appointment. The encounter form and the insurance verification are stored at the front desk in preparation for the patient’s arrival. To check in the patient, the front desk worker pulls the encounter form and insurance verification and notes the patient’s arrival time in the PMS, in the schedule book and on the top of the encounter form. If the patient has not visited the clinic before, the worker asks the patient to fill out registration forms. If the patient has new insurance, they ask for their insurance card, make a copy, and update the information in the PMS with the new information. The worker then collects the copay and any outstanding balance and enters 4

See Rouncefield et al., 1994 for a description of work interruptions.

152

Making Work Visible

Figure 8.3 Front desk activities are a constant flow of interactions with patients and others in the clinic

the payment into the PMS. If the patient’s bills are in collections, the clinic will only treat emergencies. The worker at the front desk explains the patient’s outstanding bills and the clinic’s policy on limiting patient’s access until their bills are paid. Finally, the worker gives the patient a plastic sleeve containing the encounter form and the insurance verification for the patient to take to the nurses’ station. The physical arrival of the paper forms signals the physical arrival of the patient. New methods to indicate the arrival of the patient would be required when the clinic initiated its EMR. When the exam is completed, the patient brings the plastic sleeve back to the front desk so a worker can check the patient out. The worker looks on the encounter form for follow-up instructions and schedules a new appointment if needed. If the patient has no insurance (i.e., is designated “self pay”), the worker asks for payment for the visit and enters the payment in the PMS. The worker then files the encounter form by placing it in the bottom shelf of the forms bin on the front desk. The rhythm of front desk work differs significantly from the orderly process by which the patient visit is managed. At the beginning of the day, when there are fewer phone calls and patients arriving, front desk workers focus on preparing the documents for the patients who have appointments scheduled for the following day. As patients begin to arrive, the work becomes more fragmented. They answer the phone, attend to patients who come up to the counter for either check-in or check-out, and prepare the documents for the patients who have called for a same-day appointment. At the end of the day, when the number of patients contacting the clinic declines, they process the encounter forms that have been filled out that day and the money they have collected. The forms and money are collected and put in the safe.

153

Behind the Scenes Excerpt 8.1 Five minutes at the front desk

1



2 8:15:50: W asks coworker, “Did you print?” 3 8:15:55: W checks printer for new printout 4 8:15:57: W finds a “tickle,” a message indicating a same day appointment 5 8:16:27: W searches through documents on printer for additional tickles 6 8:16:27: W starts arranging print outs 7



8 9 10 11 12 13

8:16:28 W runs back to desk to catch phone 8:16:34 W checks print outs while talking on the phone 8:17:00 W transfers phone call to dental 8:17:17 W enters same day appointment in PMS 8:17:34 W enters same day appointment in appointment book 8:17:55 W prints encounter form for same day appointment and picks it up at printer 14 8:18:28 W checks insurance information on web for same day appointment 15



16 8:19:22 W interrupted by phone call; puts it on hold 17



18 8:19:48: W interrupted by second phone call; transfers it 19



20 8:19:48: W interrupted by form delivery through the window; accepts it 21 8:20:21: W prints insurance verification form and picks it up from the printer 22 8:20:43: W staples encounter form and insurance verification 23 8:20:46: W slips stapled form into bin to complete task For example, Excerpt 8.1 shows a sequence of activities in which a worker, W, is processing a same day appointment. The task is triggered when the worker heard the printer running alerting her to a possible same day appointment request. She asks her coworker “did

154

Making Work Visible

you print” to try to determine if the print is a tickle. Dealing with the tickle that has indeed been sent, is interrupted in line 7 by the phone ringing. She finishes the phone call and returns to processing the same day appointment. The task is interrupted again in line 15 by another phone call, followed by a third phone call in line 17. These phone calls are put on hold, but the process is again interrupted in line 19 by a person at the window. Finally in line 23, W completes the task almost five minutes and four interruptions later. Front desk work requires constant communication to ensure that all of the tasks were done. Notice how in the excerpt above, the worker asks her colleague if she printed; answering “no” indicates that the printer’s activity might be a notice of a same day appointment. While this checking requires minimal effort, it adds an interactional dimension to the work of both these front desk workers. Due to the interdependence of the departments, mistakes made at the front desk are felt in the Accounts Receivable department. Accounts Receivable Accounts Receivable manages patient bills and insurance claims. They interface with many different insurance companies, each with several distinct insurance plans. Accurate claims have to be submitted, and when the claims are paid, the money has to be posted against the patient’s bill. Physically, the Accounts Receivable department is arranged as ten desks separated by dividers. The job requires significant knowledge of individual insurance companies and policies. Interruption comes only from other Accounts Receivable workers. Their difficulties come from the many insurance companies each with a multitude of policies, the variability of rules imposed by these policies, and the frequent changes to these rules. Their process involves creating a claim from the encounter form, processing rejections resulting from errors on the claim, responding to denials of coverage resulting from errors in the service provided, and tracking the money received (see Figure 8.4). Claims are created from the encounter forms and the schedule. The biller enters the encounter form number into the PMS bringing up the patient’s name and address and checks them against the information printed on the encounter form. The biller checks the insurance verification to verify the patient’s name, the termination date of the patient’s insurance, and the PCP. The PCP is the gatekeeper for healthcare, without a referral appointments can only be made with this provider. If the patient sees a physician who is not the patient’s primary care physician, the insurance company may deny the claim.

Behind the Scenes

155

Figure 8.4 Accounts Receivable activities

Once the visit is found in the PMS and the insurance information is verified, the biller adds the information about the visit to the claim being created. The provider writes the diagnosis by hand and checks the type of visit and the procedures performed on the encounter form. The biller has to translate the provider’s hand written notes into IDC-9 (International Classification of Disease version 9) codes, which indicates the diagnoses reached during the visit. These CPT procedure codes are printed on the form so they can be typed more easily. Though the clinic is reimbursed for the procedures done, the diagnosis is important because the insurance company will only pay for appropriate treatment. A substantial portion of the work in Accounts Receivable is fixing two types of problems with claims: rejections and denials. Claims are rejected if there is an error on the form. Before the claim is submitted, the PMS reports formatting errors and the clerks correct them. The claims are submitted electronically on Monday. On Tuesday, a confirmation report, which lists the accepted and rejected electronic claims, comes back from the insurance companies. These claims are fixed using the problem codes on the report and resubmitted. The clinic invested in a direct electronic interface that allows the transfer of data to the insurance carriers their patients commonly use. However, insurance claims to the less frequently used insurance companies have to be submitted on paper, because the cost of implementing an electronic interface was deemed too high. Most of the paper claims are submitted on HCFA5 forms. These claims are prepared the same way the electronic claims are prepared, but they are printed onto preprinted stock. Of course, paper submission has a longer turn around time than electronic submission, occasionally violating the 120 day submission window. 5

Health Care Financing Administration is the old name of the Centers for Medicare and Medicaid Services (CMS). Everyone still refers to them as HCFA 1500 forms.

156

Making Work Visible

Once the claims are accepted – that is, all formatting errors are fixed – the insurance company evaluates them and returns the evaluations as a remittance report. The remittance report lists all of the claims that are paid, the claims that are denied, and the amount of money deposited into the clinic’s account. One of the Accounts Receivable workers posts the money returned in the PMS system and updates each patient account with the money paid. Any claims that are denied are highlighted on a printout of the remittance report. For each denied claim, the insurance company returns an Explanation of Benefits (EOB). This list of denied claims is returned to a specialist for that insurance company in the Account Receivable department. There are many reasons why a claim might be denied, including clerical errors. For example, if the incorrect ICD-9 code is entered for a billed CPT code, the claim could be denied as inappropriate treatment for a disease. Another common problem is billing the incorrect insurance company; often this error could push the claim towards the 120 day time limit for submission. If a claim is denied, it has to be resubmitted on paper because it has to be resubmitted with the EOB. The requirement for a paper EOB indicates a breakdown in the electronic process. Exception processes, being rare, often require nonstandard techniques. In addition to submitting claims, the Accounts Receivable department also compiles and posts the payments received in the general ledger. This process is similar to Accounts Receivable processes in other small and medium businesses, except that much of the clinic’s income involves third party payers. Therefore, the biller has to reconcile the patient’s bill against money collected at the front desk, the checks that come in through the mail from the patient and bank reports of payment that arrive from insurance companies. Since Accounts Receivable does not have to deal directly with patients, their work is more predictable than the front desk, however, they have to negotiate a vast array of information and relationships. Billers are assigned to become experts in one of the seven or eight insurance companies to which most claims are sent. These companies offer many different policies, and their policies change yearly. Maintaining this expertise is a big part of the biller’s job. In addition, the billers have to be able to research the less common companies and policies. Consequently, billers spend a lot of their time reviewing information on the web and talking to insurance company representatives by phone. The billers also need medical knowledge. For example, they need to know which ICD-9 codes correspond to the handwritten notes on the

Behind the Scenes

157

encounter form. The PMS offer a search capability for ICD-9 codes, but the terminology used by the PMS and the doctors can differ. Moreover, the billers must choose the ICD-9 codes designating the diagnosis that matches the CPT code for which they are billing. That the particular ICD-9 codes required for a particular procedure can differ between insurance policies complicates matters even further. The biller’s knowledge increases their value to the clinic. The billers also need to deal with software problems. Dealing with software problems is frustrating because the insurance company can claim that the PMS was providing invalid codes whereas the PMS vendor can claim that the insurance company is rejecting valid claims. During our observations, there was a redo feature in the software for repairing rejected claims, but it was not working, so instead of using the edit function in the software, they edited the file directly and resubmitted it. The claims that were edited using the feature were rejected with errors that were not caught originally, but the PMS vendor claimed that the rejection was due to changes at the insurance company. An interface change would require payment to the software company. So instead, the billers worked around the problem. The billers’ negotiate an array of relationships. They interface with the providers to get clarifications about the correct CPT and ICD-9 codes to put on the claims. They need to stay in contact with the front desk and other office staff to insure that the information collected there is accurate and timely. Finally, they need to maintain contact with the people at the insurance companies from whom they get most of their information.

Reflecting Findings to Stakeholders One of the key aspects of our methodology is sharing our emerging observations and findings with the study participants. Sharing findings with participants is important for building an ongoing relationship with the site and providing an opportunity for the participants to see how we interpret and represent our observations of their work. When sharing findings, particularly at an interim stage in the study, participants correct misunderstandings we may have and refer us to other people and activities that could help us develop a better understanding of the work. For example, we discovered as we reflected the billing procedures back to them, that we had confused rejected claims with denied claims, so we scheduled additional sessions to clarify our misunderstanding. The findings provide the organization under study with an outside perspective on their work practices which they can use to make decisions about

158

Making Work Visible

changes in practices and technology. For example, by sharing the experience of the front desk with the rest of the clinic, we were able to show them the difficult environment in which they work and give them insights into the source of errors by the staff there. There is no easy solution here, because interruptions are intrinsic in the work and it is too costly to hire enough clerks to handle peak demand. Perhaps the most profound result from the feedback was increased sympathy for “stupid mistakes” at the front desk. When we shared our findings with the clinic, it allowed us to suggest a technology change that might be helpful for the clinic. The front desk listens for notices of same day appointments arriving at the printer, so we suggested that they introduce a second printer at the front desk that would be used only for these notices. Whenever this printer ran, the staff would know that a new same day appointment had been scheduled, allowing them to address it more quickly and reduce the chance that a tickle would be lost among other documents.

Conclusions The study of the front desk and Accounts Payable department rounded out the research team’s understanding of clinic practices by providing insights into the business side of the clinic. We learned the complexity of humanfacing work and the types of errors that occur under the interruptions necessitated by this type of work. We also learned how interdependent departments in an office are and how difficult it is for departments to understand the demands and work constraints on other departments. Despite these difficulties, we saw that people compensate for inadequacies in their work tools with practices such as listening for a tickle or manually editing the software system. We came to understand the intricate interdependence of the practice management and the records management systems (whether the latter is paper-based or electronic). As healthcare providers consider the transition to electronic record systems, it will be important to evaluate how essential practices will be supported, eliminated, or transformed. This study captured the practice surrounding the business side of the clinic and gave us a foundation to project how these practices could change as electronic records are introduced. For example, the paper encounter form indicates the arrival of a patient; new procedures will be needed when the encounter form is replaced by an electronic form. A critical element of the work that is uncovered by ethnography is the difference between the process point of view and the workers point of view.

Behind the Scenes

159

For example, from the process point of view, the front desk work consists of sequentially organized tasks. However, from the workers’ point of view, they work on multiple tasks, at the same time. Though both points of view are necessary for successful technology design, the workers’ view can be missed without ethnographic study. This research gave Xerox insight into the use of medical systems that guides research and development. For example, physicians are used to filling out encounter forms, but some are unfamiliar with computers. We have designed, but not implemented, a system whereby the physicians can document their patient encounters using paper forms similar to the existing encounter form, which can then be scanned and the data entered into an EMR. We continue to study the clinic as they complete their transition to an EMR system, and provide them feedback in order to help them shape their technology solutions and work practice process. While some of the particular practices described here have changed, the challenges remain the same: the front desk must still manage interruptions from patients and the Accounts Receivable department must still manage a large and changing array of insurance policies. By making their work practices visible, we hope to improve the technology that helps them manage these challenges.

9

Seeing the Right Color Technical and Practical Solutions to the Problem of Accurate Colour Reproduction in the Digital Print Industry Tommaso Colombino, David B. Martin, Jacki O’Neill, Mary Ann Sprague, Jennifer Watts-Englert, Jutta Willamowski, Frederic Roulland, and Antonietta Grasso

This chapter presents the findings of a series of cross-continental, crossresearch-centre ethnographic studies of Xerox customer print shops and design agencies. This was undertaken to examine the technical challenge of reproducing colour consistently across different devices (such as digital cameras, computer screens, and printers), and how that challenge is differently addressed by professionals involved on one side in the creation of colour documents (designers), and on the other side their (re)production (digital printers). The studies were prompted by Xerox Voice of Customer (VOC) feedback that indicated that while our digital production presses are capable of excellent colour quality, achieving it can require extended effort in document preparation and printer adjustments. Given that the major competitive advantage of digital print technology is the ability to print short-run and on-demand at an affordable cost and with a quick turn-around, any excess time and labour involved in achieving acceptable print quality is perceived as a problem. While the VOC complaint was clear, the reasons behind it were less so. Professional printers have at their disposal hardware and software tools which should, in theory, allow users to manage the issue of colour consistency across devices in a digital production print workflow. These tools rely on the industry-standard technical system of Color Management. Color Management (CM), was developed by The International Color Consortium (ICC),1 to enable translation between different colour spaces and colour 1

160

http://www.color.org/index.xalter

Seeing the Right Color

161

devices (monitors, printers, etc.). Essentially this means that when a file is transferred from one device to another, it should come with a profile that indicates how to interpret its colour information. This profile is then translated into a device-independent colour space defined by the ICC, before it can be reinterpreted correctly in terms of the new device. CM can be described as a technically correct solution, in that the system, when used properly, can translate colours between, for example, a computer screen and a print device such that colours seen on print appear essentially like those on the screen. CM technology was adopted to avoid relying on manual adjustments of the document and the printer settings in order to achieve good quality colour reproduction. The VOC data reported the need to undergo several proofing cycles to achieve the desired colour suggests that this technology was failing. The work practice studies were therefore undertaken with the goal of understanding what exactly the extra work involved in achieving acceptable colour quality consisted of, where it originated, and how it might be addressed.

Background This paper builds on previous research in Computer Supported Cooperative Work (CSCW) which has considered the design of “sociotechnical systems,” bringing the work into a previously unexplored domain. It shows how an understanding of the social is just as central when designing technical infrastructures as in systems design. The importance of considering the social aspects of work in the design of the technology has been established for some time. For example, in the early 1990’s, Gentner and Grudin described how computer use, and therefore interface design, was changing, moving from an interface for engineers, representing the engineering model of the machine, to single “everyman” users, where the interface took in concepts from psychology, to groups of users, where it needed to take into account the social and organisational context of use. At the same time, workflow systems – systems that manage the coordination of work along a process – were becoming popular. They were commonly designed by abstractly modelling the processes of the work (e.g., using data flow diagrams). The process description is at the heart of workflow technology, describing the sequence of actions to be undertaken so that the process is carried out correctly. The system then drives user actions according to the sequence embedded in the system. It was in critique of workflow systems that the idea of sociotechnical systems really came into

162

Making Work Visible

its own. Researchers examining workflow systems that were inadequate and those working on workflow system redesign reported how people interpret processes very flexibly to get the work done (Bowers et al., 1995; Carstensen et al., 1999; Dourish et al., 1996). Process models commonly omit vital details of how real work is organized. Hence the workflow systems built on these process models either restricted workers’ activity, generated workarounds, or were relegated to mere accounting rather than coordination tools (Jordan and Lambert, 2009). This led the call for creating systems which were designed with a clear understanding of work practices rather than on the basis of an idealised or rationalised, process model (Dourish et al., 1996; Klein et al., 2000). This research speaks directly to the findings from our studies of printing and graphic design. CM, as it is currently instantiated, does not fit into document creation and production workflows. It gives the appearance of being a wholly technical solution to a sociotechnical problem (O’Neill et al., 2008). Instead, the practical solutions currently adopted in the industry to manage colour are “good enough” (quality in relation to price and effort), but rely on direct communication between print shops and designers and ad hoc workflows rather than accepted standards. The acknowledgment of these practical solutions and of the communication exchanges required in the workflow is the starting point for proposing new forms of technology support to enable digital colour production workflows.

Settings and Method The project consisted of two sets of multisited ethnographies in Europe, the United States, and Canada of 1) print shops and 2) graphic design houses. This work was carried out by Work Practice Technology teams from Xerox’s European and North American research centres. The teams consisted of ethnographers and technologists, including colour experts. Six print shops were visited: four commercial print shops and two Xerox printer testing sites. The print shops varied in size, customers, core business, and workflow organisation. Three graphic design houses were visited (two in the United Kingdom one in the United States) – they varied in size and client base. Interestingly, we started the project with the studies of print shops and only turned to studying graphic designers when we realised that many of the problems that we saw producing colour in print shops originated upstream at file creation. This led us to turn our attention to graphic design houses to understand the problem from their perspective.

Seeing the Right Color

163

The field studies involved site visits of between one day and three weeks, with a total study time of more than 20 weeks. Although the team working on the project was large (six ethnographers and five technologists) the size of the teams visiting the various sites varied a lot. Visits to the United States sites tended to be shorter with a larger mixed team (e.g., three or four people), whereas visits to the European sites tended to be longer and involved only individuals or pairs of ethnographers. Largely this was for practical reasons of time, money, travel, and site availability. The different durations and locations of these ethnographies gave us both an in-depth insight into the practices of design and document production and an understanding of a range of “types” of print jobs, workflows, their problems, and solutions according to the different organisational and customer characteristics. The joint nature of the work between the European and North American Work Practice Technology teams enabled us to visit a greater range of sites, with different characteristics, for a greater total time, meaning the study was in-depth, and had good diverse coverage. However, as is common with ethnographies, this access was more opportunistic than systematically planned – you cannot fully dictate which fieldsites, which people, and for how long you can have access. The key to analytic claims arising from ethnography is that they must be legitimately rooted in and demonstrated through the fieldsite observations – the more you see, the more you understand, the more detailed and stronger your analysis can be – and it can be comparative when other sites are visited. Ethnography is not about scientific method, statistical sampling, and so forth, but is about analytic rigour (demonstrating legitimacy, warrant, etc. for claims). “Completeness” is always a judgement about whether one has enough data for the desired purposes. Our primary research method was observation, supplemented by in situ interviewing. Materials, such as examples of print jobs, were collected as well as field notes, photographs, and audio and video recordings to create a rich data set for analysis. An ethnomethodological approach (Garfinkel, 1967 and Randall et al., 2007 for a recent discussion of the use in CSCW and related disciplines) was taken to analysis, paying special attention to understanding and describing the methods that practitioners used to design and produce documents, with a particular focus on colour and aesthetics. We were interested in how design work and printing were organised socially, how they involved working on and reasoning about colour, communication, and what the production and management of colour, aesthetics, and document requirements consisted of in practice, in relation to files, and

164

Making Work Visible

other materials. We also examined how this work related to ICC Color Management – if it was used, how it was used and understood – and we were interested in other local ways by which colour was managed. This in-depth analysis of the problem was used to inform the design of some technology solutions which aim to better fit with actual design and production environments.

Setting the Scene: What Is Colour Management? To understand the problem of colour reproduction in the print industry and therefore the relevance of our findings it is necessary to have a basic understanding of CM. The technologies involved in colour work – input (e.g., scanners and cameras), display (e.g., monitors), and output (e.g., printers) – have device centric ways of producing and encoding colour (e.g., different colour spaces). Colour is produced differently on screen, by combining red, green, and blue (RGB) light sources, than in print, made of cyan, magenta, yellow, and black (CMYK) ink or toner pigments which are applied to some substrate (e.g., paper). On top of this, different devices (even different devices of the same order, e.g., two printers) have different gamuts2 (ranges of colours) they can produce. This clearly sets up an issue for how files or documents, passed between devices, are represented. There needs to be some method of translating between these different colour spaces, particularly for deciding how colour definitions within one gamut should map on to those in a different gamut. ICC Color Management was designed to reproduce colour accurately across different devices by translating the colour profile of a document in one device into the equivalent profile for the next device. Colour profiles can be thought of as part of an infrastructure that “sits behind” the interface level and operates on its own according to preestablished defaults. In the case of professional tools for the creation and production of digital documents and images (such as Adobe CS, QuarkXPress, and the print drivers of digital presses), CM controls are prominent. However, their function is not self-explanatory without some understanding of colour science, see for example the choices offered when importing a photo into a document with a different colour profile (Figure 9.1). 2

The range of colours – gamut – a device can produce are define mathematically in a multidimensional space. Colour spaces both overlap and diverge. Translating colours from one space to another involves mapping colours in one space to the other according to various algorithms. Various compromises are made for mapping colours that occur in one space but are outside the gamut of the other.

Seeing the Right Color

165

Figure 9.1 Colour Management choices when opening a file in Photoshop®. Adobe product screenshot reprinted with permission from Adobe Systems Incorporated

Every screen-to-print workflow uses some CM technology components whether it is the knowing intention of designers and print workers or not. Although CM is not in itself a workflow system, to use it successfully, necessarily implies a workflow where a specific instantiation is rigorously followed and understood by all of those involved in the process. That is, achieving the desired outcome requires successful coordination and communication of the proper colour-management practices. In practice, however, CM is rarely used as intended. This is a known problem (Riordan, 2005), which we confirmed in our ethnography. What was not known, however, were the reasons for this. The ethnomethodological analysis of the fieldwork data, revealed the underlying reasons for the nonadherence to CM, from which we were able to derive some suggestions for technology solutions.

Field Study Findings Our studies confirmed that ICC CM was not being used as intended but rather graphic designers and print shops employed a variety of ad hoc workflows to manage colour (Martin et al., 2008; O’Neill et al., 2007). These workflows gave a variety of levels of success, but were often costly – with the print shops often having to put in extra uncharged-for work

166

Making Work Visible

(O’Neill et al., 2007). Since CM is an underlying infrastructure it is always on (colour values must be translated for colour to be produced) and thus cannot be worked around (Martin et al., 2008, 2010). The way that practitioners work with colour – through tangible, practical, visual understandings, and manipulations – is at odds with CM, which uses abstract, mathematical manipulations of colour (O’Neill et al., 2008, Martin et al., 2008). Thus whilst CM may be a technically correct solution, it is rarely used as intended because it has a number of fundamental limitations, which lead to it being either ignored or misused. In this chapter we will explore some of these limitations to illustrate how ethnographic research can provide insights into even highly complex, (apparently) technical problems. These limitations include the fact that: *

*

*

CM is not transparent. Colour translation takes place, irrespective of whether CM is supposedly switched on or off, or whether the practitioners know what the settings are. When no information is provided to the system, a default profile has to be assumed and this is unlikely to be known by the user. This can be incredibly confusing and can lead to all sorts of problems when files are printed. CM is not fault tolerant. If for any reason, at any stage of the document lifecycle, colour profiles are not correctly attached to a document, there is no means to know from the latest document version what should be done to go back to its initial specification, or even to reach a state that is closer to the original specification. Therefore, to guarantee the integrity of a global colour managed solution in a digital-production workflow a rigorous CM process is required from the start of – and must be adhered to by all the parties during the whole workflow. This means that from a user perspective, this technology is either trusted – that is, considering that all previous parties appropriately applied CM – thus useful – or not trusted, thus useless. CM assumes a strict specification of colours, that is, that the colour encoded in the document is the exact colour to preserve. In reality, a variety of ranges of tolerance apply depending on the document requirements. This information cannot be captured in the ICC colour specification and can therefore not be used when the colour cannot be preserved (e.g., where device gamuts do not match adequately). With ICC CM, colour substitutions are left to heuristics (which may be better or worse matches for user preferences for that document) reducing the benefit of using CM.

Seeing the Right Color

167

We will illustrate these findings with examples from the print shops and graphic designers, to show how we came to these conclusions and their impact on the production printing workflow. We will, however, focus on a particular part of the workflow and a single theme: the use of colour controls on print machines in the print shop and the theme of transparency.

Problems at the Print Shop: Transparent and Nontransparent Colour Controls A classic design principle for producing usable user interfaces has been that of transparency. Various definitions exist, but in this context we are using the term to mean the clarity of: (1) the operation of a system; and (2) the features of feedback presented on the interface. An interface is deemed more or less transparent depending on the information (usually visual) it provides in relation to the actions of the user. For example, in the information retrieval domain the term transparent has been used to refer to interfaces which gave visible feedback on the underlying transformations which the system performed on users’ queries before returning results (Koenemann and Belkin, 1996). It is this version of transparency that we are applying to the domain of colour adjustments in production printing, that is, does the system make its underlying transformations available to the user and can the user see what effect their actions have? It should be clear that it makes intuitive sense that transparency should be important for making systems usable – after all, if a user does not know how a system works and does not get feedback on his actions surely the system will be a usability nightmare. However, as is usual with design, the case may not be so clear cut. What constitutes the right type and level of feedback? Indeed, an attempt to design a transparent query system (Muramatsu and Pratt, 2001) for the comparatively simple domain of search engine queries, met with somewhat mixed results, suggesting that designing usable transparent interfaces is no easy task. In many cases, when we talk about transparency it is really about the transparency of functionality and of the interaction model – such as whether the user can understand what the effects of particular actions will turn out to be. Users do not want to understand (and many would be unable to) the operation of code or the algorithms behind an interface and its controls and parameters. With a highly complex system, the point of the interface is often to present things in a way that makes successful operation possible without the requirement to understand just how it is structured or how exactly it works. To examine how this concept of transparency plays out with CM tools, we will look at the work that print operators do when proofing. Proofing is

168

Making Work Visible

the process of printing off single copies of a document “run” in advance of the entire run to ensure the output is of good quality – whether the proof is matched to a hardcopy example or not. Two tools commonly used at the Digital Front End (DFE) – what might be considered the interface to the printer – are emulations and Tone Reproduction Curves (TRC). Using Emulations for Aesthetic Adjustments Emulation settings allow the operator to control the interpretation of the colour encodings in the document file. They emulate the look and feel of other print machines, paper types, and so forth. To be used in a predictable way they need to be used in a controlled CM workflow, however they were frequently used by the print operators in non-CM workflows as an uncontrolled way of manipulating the aesthetic properties of the document – “try it and see.” To illustrate, one print shop systematically used emulations as aesthetic tools. They had developed, over time, a heuristic understanding of what emulations produced the best overall results, and had recently decided to use one of the available preloaded emulations, Euroscale, as a default transformation on most jobs. However, when printing a commemorative calendar, this caused them unexpected problems. The calendar came in two versions, with a calendar on one side and an image on the other. The main difficulty, as it turned out, was balancing the quality of the image on the back while achieving a satisfactory background colour on the calendar side. The problem turned out to be because the DFE had recently had a software upgrade. The upgrade had changed the way the emulation worked, which meant they could no longer rely on their acquired understanding of how their default emulation (Euroscale) would render the colours in this case. The test file produced unexpected variations in colour on the calendar side of one version (being noticeably more pinkish than the expected orange). Previously, Euroscale had never produced such a marked deviation from the expected colour. In the other (purple) version, there was a noticeable deviation in colour but it was not as marked as with the orange version. However, when the run was tried with no emulation, the colour of the calendar was better, but the picture on the back of the second (purple) version was noticeably pixilated, whereas with Euroscale both pictures were of higher quality. In the end the print shop compromised by printing the purple version using Euroscale and the orange using no emulation. It is important to remember that all this experimentation, while in this case necessary to achieve acceptable print quality, takes time and money that

Seeing the Right Color

169

can be ill afforded in jobs that need to be turned around quickly and efficiently in order to be profitable. Production printers are not designed or optimised to be used as test printers – that is, print one-off copies of a particular job. Their print engines take up to fifteen minutes to power up and down between tests, easily lose their calibration, and in this way become tied up for long periods of time, which is rarely charged to the customer, in inefficient testing cycles when they should be used to print production runs. The issue of image quality with colour accuracy could have been approached in a variety of manners, for example, changing the colour composition of the background colours on the calendar using the software it was created on. However, adjustments done in prepress are also time consuming and costly, and if a file is submitted as “ready to print” such adjustments will not have been included in the cost estimate. Also, if the file is not colour managed there is no guarantee that prepress adjustments will produce the desired effect. There is a clear convenience in being able to make adjustments at the DFE but this means that print operators must use transforms that apply to the whole document in an attempt to find a good balance between different areas and this can prove difficult. Emulations used for aesthetic control in non-CM workflows are clearly nontransparent – the user cannot be sure of the result in advance. The effect of applying any single emulation is difficult to predict and categorical (changes all the colours in the file in complex ways) rather than directional (e.g., a percentage increase, a sliding scale). Therefore the outcome of one test emulation, if not fully satisfactory, will not inform the user of which steps should be taken next. All of the other print shops also used emulations for aesthetic adjustments, to a greater or lesser extent, and had similar problems in predicting their results, but nevertheless attempted to acquire some heuristic that might inform their use (even though these can only ever work in a temporary approximate manner). Using TRCs (Tone Reproduction Curves) Another tool that was used in the aesthetic adjustment of images was the TRC. The TRC can be used to change the levels of one or all of the four toner pigments (CMYK) that make up the printed image. This type of parameter modification could be described as relatively transparent to the user, in that the action that one performs has, or can be expected to have, a direct correspondence, in direction and magnitude, to an observable change in the perceptual quality of the printed document (i.e., changing the shape of the curve for a particular colour separation directly affects how much of that

170

Making Work Visible

Figure 9.2 Tone Reproduction Curves can be directly manipulated by the user

colour will be applied in proportion to the other colours). The process of interpreting the CMYK values is mathematically complex (i.e., because it involves understanding how, for example, changing part of a magenta (M) TRC will affect colours that are composed of CMY and K), but for the TRC (as for other parametric-type selections) it may be possible to understand, in theory, what transformation one is effecting, although the net result only becomes obvious once the file is printed. Furthermore, the interface, in the case of TRCs, offers a direct visual representation of the manipulation that is being effected. The visual relation between parametric modification and intended effect allows for incremental modifications of a continuous variable, and it might therefore be thought that this could guide the development of heuristic understandings of the general effects of manipulations. However, this example illustrates that “transparency” (of the form outlined previously – direct, incremental, reversible “graph” manipulation) does not equal easy to use because it is very difficult to envisage how the results will play out in a multicomposite coloured document. When one considers that many colours are produced by CMYK combinations of specific percentages one will realise some of the difficulties in using this tool, and the fact that its transparency can be problematised (i.e., in effect it is only partly transparent; the interface shows the user what they are manipulating and by what magnitude but the effect of this on the document as a whole is often difficult to predict).

Seeing the Right Color

171

We found that operators in a “testing,” rather than a commercial site who used the TRCs (along with emulations) for every print run had moreor-less useful heuristics of the effect. For example, suggesting about one print that more red was needed in a number of places (e.g., “I like your skin tone better here but you need more red”) and that the black should be “brought up.” Even these expert users often completed numerous adjustments to the TRC to produce satisfactory results and crucially their manipulations were made at prepress where the tools (such as Photoshop or Illustrator) provide direct visual feedback mechanisms of their effect by showing the results on screen. Implications for Colour Management Tools We can see then that there are different nuances to transparency. Emulations are nontransparent, revealing neither the operation of the system nor providing feedback at the interface level. They perform a number of complex mathematical transformations whose effects will depend on the document’s particular (unseen) settings. TRC curves have a level of transparency in that the changes they perform are known (e.g., reducing magenta) and there is feedback at the interface level (in that one can see the curve one is manipulating). However, this transparency is limited in the tools at the DFE, as it is not easy for the user to understand what effect these manipulations will have on the complex (rather than singular) colours within the document itself. TRC’s are mainly used in prepress with tools which provide real time dynamic visual feedback of the effect of the manipulations. While knowledge of the underlying mechanisms is part of the competence of an experienced designer/photographer it is undoubted that a crucial part of learning and understanding is the visual feedback of a real-time dynamic display that can be directly related to the spatial manipulations operated on a graphical user interface (GUI). This type of visualization is not available on the DFE – whether using emulations or TRC’s. It is clear that people can understand and develop methods for successfully completing aesthetic enhancements and making colour changes when they can more directly perceive the visual effect of their actions rather than via deep reasoning about the way a tool works. Since they are not designed to support the task of aesthetic adjustment, DFE controls give little opportunity for developing broadly generalisable and reliable heuristics, which often leaves the printers working with often somewhat folkloristic “rules of thumb” or going through inefficient trial and error proofing cycles. In the next section we briefly turn to the work of

172

Making Work Visible

graphic designers in order to understand how features of CM relate poorly to how they organise their work.

Problems in Document Design: Colour Specification in Practice CM technology is designed to be used from the beginning of the document creation lifecycle through to production. However, document design may not always be done with printing in mind. Often graphic designers are designing “concepts,” which is some illustrated ideas about the product, which may or may not be taken up by the client. Only over some iterations, often involving parts selected from different concepts as well as additional work, is a product settled upon. Once the product has been chosen, designers move into the artwork phase (also known as mechanical design), where the final product is drafted to exact specification (size, colour, text etc.). Therefore, early in the design process the focus is not on design for print, but on concept design, thus it is unlikely that this is a point where designers would want to do extra work to ensure accurate or high-quality printing, when much of what they design may never be included in the final printed product. Further, the aim of CM is to communicate the document creator’s intent across the document lifecycle, through various devices and colour spaces to production. However, design is a process where compromises are common and “intent” is not preformulated, rather it is realized during the process of doing and is shaped by the interaction between designers and customers and print shops around artifacts. As we saw with the calendar, compromises often have to be made between parts of the document during printing and intent cannot be encoded without prior knowledge of these compromises. As stated earlier, graphic designers have a visual, practical, and tangible orientation to colour. They are interested in colour as they see it – here. They (but it must also be said pretty much everyone apart from colour scientists and certain photographers) have very little understanding of the workings of CM. They deal with colour for printing when it is decided their design will be printed. They deal with it practically – how much will my client spend, will this be digital or colour, can we order specially mixed inks, what sort of proofing can we do, which printer gets a good orange?, and so on. Using colour-swatch books to pick colours then encode those colours in files (allows tangible visual matching between screen and sample) and working in CMYK emulation settings (rather than native RGB) on their screen are also methods that enable nonexact but practical, visual and tangible colour

Seeing the Right Color

173

manipulation and reasoning that enables good enough results. This orientation to colour is clearly paradigmatically (everyday, experientially) different to CM (abstract, mathematical) and thus points to a central problem with the CM infrastructure that needs to be bridged (or solved) to make it more usable.

Recommendations for Improving the Management of Colour Understanding how practitioners deal with colour in practice leads us to make a number of recommendations. The first two focus on adjustments to the current technology and tools, the third leads off from the second but is a more radical suggestion for managing (visualising, assessing, recovering, correcting) colour in non-CM workflows. All of our material has been widely disseminated within Xerox and has attracted interest, as too it has been reported in the academic community. We took the choice of looking to develop a prototype solution – Print Mediator (discussed further later in this chapter) – based on the third option and have been developing this for the last two years. We have yet to impact the international community of colour scientists, ICC, and so forth, but this is an ongoing project. Enabling colour adjustments at the DFE is likely to remain beneficial because: (1) this is where the colour problems are often noticed and therefore, if effective, adjustments here can be more timely than in prepress; (2) “Ready to print” files do not include prepress time, yet the printers often can not take “ready to print” literally, as to do so would be to lose clients (O’Neill et al., 2007). We therefore propose that print operators need to be given tools which are designed for colour adjustments in noncolour managed workflows. The answer to the problem is not to load up the DFE with more and more tools, rather it is to make available simple tools for which the effects are immediately visible, for example, through dynamic, real-time soft proofing. This would enable a heuristic understanding of the effect something has – a form of transparency – rather than deep reasoning about the way it works (its underlying mechanisms). Since graphic designers are, in many cases, not concerned with colour in relation to printing during concept creation, it makes sense to allow CM to be applied or recovered with a document later than initial creation in the production process. Certainly default settings should offer something that allows good recovery at a later stage or that preserves an audit trail. We identified a need to bridge the gap between the current CM infrastructure and the actors of the workflow to make achieving good

174

Making Work Visible

Figure 9.3 Print Mediator provides comparative soft proofing and guided natural language annotation features in Acrobat®. Adobe product screenshot reprinted with permission from Adobe Systems Incorporated

color more tangible and understandable for users. Currently the knowledge and perspectives of graphic design and print shop practitioners are far from those required for the CM infrastructure to work correctly. Our recommendation, therefore, is that there’s a need to complement current CM technology with a set of tools that can assess, highlight, and offer solutions to colour problems and assist in communication and mediation between designers and print shops. To that purpose, we have been developing an AcrobatTM plug-in – Print Mediator (PM) – that provides an enhanced comparative soft-proof on documents to be printed. The features of PM will enable a more relaxed workflow, and provide some key capabilities: *

Intelligible integration with colour processing tools. Its softproofing is based on a calculation of the differential between the document on screen and how its colours will change when printed (taking information from the print queue settings). It will be enhanced (under current development) by a number of detectors that can point out various colour issues (most obviously when a colour is out of gamut, or will change markedly) and correctors offering solutions to issues (where possible).

Seeing the Right Color *

*

175

Aesthetic Document Annotation. PM will also enable designers to specify information regarding intent (I’m looking for a really bright red here), tolerances (as close to Pantone xx as possible), priorities (the most important aspect is the royal blue) and expectations with regards to colour rendering by tagging areas of documents. Such aesthetic document annotations can elaborate the colour values specified in the document file. They will be particularly useful when these colour values cannot be trusted because of a failure in using the ICC model during the document workflow. Aesthetic document annotations will also be useful to reach a compromise between constraints and expectations when perfect visual consistency cannot technically be achieved on a printer. Mediation. The use of detectors and correctors and the provision of aesthetic document annotations attached to a document implies the need for tools supporting the creation, tracking, and communication of annotations from the document originator to the printer. Thus the expectations associated with the document can be constructed, accessed, and understood. Our study showed that one difficulty with the “ready-to-print” business model is that it assigns the entire responsibility of accurately reproducing a submitted job to the print shop. An accurate reproduction may sometimes be challenging given the device limitations. We believe that PM fits into the specifics of this type of workflow well as it should work as a pre-submission or prepress tool. Since in many cases “intent” is something that develops, this allows users to assess documents just before submission, taking into account print queue information, with any changes and annotations being communicated to the print shop. Another nice feature of it is that it places some more responsibility and control for colour choices back in the hands of the designer.

Lessons Learnt This project stands as a demonstrator of the usefulness of ethnographic studies, where these provide the material for a thorough and detailed analysis – in our case with a clear ethnomethodological orientation, of the underlying work, assumptions, and reasoning and their relation to technology in use. There are a number “assumptions of use” embodied in the CM infrastructure – end-to-end workflows; nontransparent tools – because the complexity of its workings was designed to be hidden in the workflow; strict specification of colours; rigid adherence to protocols by all parties, with no

176

Making Work Visible

allowance for the inevitable need to compromise. These can then be compared with the methods and reasoning that underlie the actual practices of the users – in doing which we revealed a large gap between what would be required to make the CM infrastructure work, and the professional skills and practices of the practitioners. From this understanding we can begin to envisage solutions. This was a challenging area of investigation, requiring that all parties involved – the ethnographers and the technologists – became familiar with the mechanisms of colour management and, at the very least, the basics of colour science. It was therefore vital to the team that we had colour scientists involved, who were not only part of the project but came out into the field themselves. This created a unique situation – they were able to teach us, to help us to interpret our findings and to be, initially at least, closely involved in technology design. In addition, being in the field provided groundbreaking revelations for the colour scientists themselves: having worked for many years on what was a good technical solution, they came to understand that this did not make it a good sociotechnical solution and, indeed, that alternatives were desperately required. What might be seen from the remove of the laboratory as “stupid” or recalcitrant users was revealed on the ground as in fact a gulf in understanding – between the abstract, mathematical manipulations of CM and the practical, visual, and tangible manipulations that practitioners perform when managing colour. The long-term nature of the project (the studies were carried out over two years, analysis over a number more) was essential given the complex nature of the domain – as was the freedom to move in the second year upstream to understand the work of the graphic designers. If the scope of this project would have been prespecified in advance (i.e., to look at the work of print shops) we would have had only a very limited and print shop-centric view of the work. This is a good example of one of the principles of ethnographic study – do not decide in advance what you might find or even where you might need to look to find it – rather the research needs to follow where the data takes it.

Conclusion We started the chapter, and our studies, with a perceived problem around performance gaps with production colour, especially around the accuracy and consistency of colour and the complexity and efficiency of the workflow. What the Work Practice studies ultimately showed is that a technically correct solution such as ICC Color Management, designed to be the

Seeing the Right Color

177

standard technology to translate and reproduce colour “consonantly” across devices and formats, fell short in its implementation because the problem of producing colour across screen-print workflows, rather than being a purely technical problem requiring a purely technical solution, needs to be viewed as a sociotechnical problem. This is a realization that, in turn, opens the door for sociotechnical solutions. The need for design to orient not just to the technical nature of a problem but also to the social nature of the use of technology is not a novel discovery. What is always novel is what might be discovered when one goes to the trouble of trying to understand and describe the methods that practitioners actually use to accomplish the activities, goals, and tasks that make up their work. It certainly provided us with the perspective to realise that it might be useful to step outside the current paradigm of CM somewhat and look to design a practical tool that would chime better with practitioners understandings and practices. Print Mediator attempts to do this. The development and initial testing is progressing encouragingly, and we have now submitted four patent applications, and we will report on development as they happen. Note: Acrobat® and Photoshop® are either registered trademarks or trademarks of Adobe Systems Incorporated in the United States and/or other countries.

Part IV

The Customer Front

10

Integrated Customer Service Reinventing a Workscape Jack Whalen and Marilyn Whalen

It is shortly after eight a.m. on Monday morning, and Nan has just settled into her cubicle at a large call center operated by INK Corporation – a Fortune 500 company – that is dedicated to customer service and support. Work at her center is divided into three different organizations that are responsible for customer support. One organization handles requests for technical service on machines purchased or leased from the company; another deals with ordering supplies for those machines; and the third, Nan’s organization, is responsible for customer account administration – questions about invoices, contracts, and so on. Together they employ close to 4,000 people, located in four North American sites. Soon after Nan puts on her headset, a call is routed to her station. She hears a soft beep in her ear that announces the incoming call and selects “answer” on her call management screen. Excerpt 10.1 INK customer service phone call

Nan: Thank you for calling INK., this is Nan. Cust: Yes. This is Stephanie Dales from M. A. Wilkin’s office in Louisville, Kentucky. (Nan writes “Stephanie” on a piece of paper) Nan: Huh hum. Cust: And, uhm, last week we had a problem with our copier, uhm copying really dark Nan: Uhum Cust: and not gettin’ all the – not copying everything. ‘n some man came out here and supposedly fixed it. But it’s getting worse. Now there’s like black lines across the page. ‘n we can’t even send anything out to the clients because it looks so bad. 181

182 Nan: Cust: Nan: Cust: Nan:

Cust: Nan: Cust: Nan: Cust: Nan: Deb: Nan: Deb: Nan: Deb: Nan: Deb: Nan: Deb: Nan: Deb: Nan: Deb: Nan: Deb: Nan: Deb: Nan:

Making Work Visible Hm. . . Is there any way he could come as soon as possible today? Uhm. I think you got customer billing instead of customer service. Oh I did? (laughing) Yeah. (return laughter) Let me transfer you back over to- do you have your serial number? I can go and give ‘em Yeah. It’s sixty two H, Uhhum. (writing on paper) four nine three, two four three. Okay. Let me transfer you back over there. Okay. Thanks. Huhhum. (Dialing sounds) I.N.K. services. This is Deb Smart. How can I help you? Hi Deb. This is Nan with the Customer Account Center Huhum. (very softly) I have Stephanie who needs to place a- um- a service call but she got us instead. Okay You want ‘er s – You want her serial number? Sure. It’s sixty two H, four nine three, two four three. Mary Wilkins? Uh. . .Stephanie’s her name. Uh oh! Give me the serial number again. Sixty-two H, four nine three, two four three. I’m gettin’ Mary Wilkins C.P.A. Oh, for the- customer? Uh huh. Let me uh – talk let me talk to ‘em.= Okay Thanks Uhhuh (transfers the call)

Integrated Customer Service

183

This call illustrates many issues with customer service – issues that have been troubling Nan’s company. From INK’s perspective, it was a call that should have been made to their equipment service function, not to Nan’s account administration unit. The three support organizations in INK not only have their own 800 numbers, but separate customer databases and information systems, technology infrastructures, work processes, and management. Nan has no access to the equipment service organization’s database and cannot place a service call for this customer or deal with any issues regarding the faulty machine. As a consequence, the call must be handed off. Note that before being transferred, the customer had already identified herself by name, described the problem she was having with the machine, and reported the urgency of the situation. Worse, for the customer this situation is especially taxing because this was a call-back: the service technician had already been out to her business to “supposedly fix it,” meaning that the machine had been nonoperational for some time. Also, an important part of the customer’s efforts to provide information – the particulars of the problem description – is more or less lost in this call due to the transfer process (cf. M. Whalen and Zimmerman, 1987). Even the machine’s serial number becomes a source of some confusion in the hand-off. After discovering that the caller’s business is her copy machine, Nan asks her for the serial number, knowing that the service organization will want it. Of course, had Nan typed that serial number into the database, she would have the related customer identification information, including the name of the business and the main contact person, right in front of her. When Deb from the equipment service function identified the customer name as “Wilkins,” Nan – even if she did not remember what the customer said earlier – would have been in position to confirm or to correct the information. As it was, though, Nan had written down the caller’s personal name (Stephanie), not that of her company (Wilkin’s), and so the customer had to start all over again with Deb, confirming her business identity and retelling her fairly detailed story.

Good Business Reasons to Change the Process When you consider that approximately 22 million customer requests came through the different support functions for INK each year, and recognize how much time was wasted – for the customers and employees alike – there were good business reasons to consider making a change in their customer service and support process. Certainly, each functional organization could claim success in terms of traditional measures like volume of calls handled and brevity of calls, but the customer experience could be quite different. The reality is that

184

Making Work Visible

customers often experienced uncertainty or confusion about the appropriate organization in INK to contact or which number to call, and, more important, their problems often had more than one dimension and may have required the involvement of more than one support function. Being transferred and having to retell your story two or even three times was frustrating. The experience was also problematic for employees, who felt unable to meet the customer’s needs at the initial point of contact. And this was not simply an issue when the customer called the “wrong” organization, as was the case in Excerpt 10.1; it happened even when the customer reached the “right” organization. For example, if a customer called the equipment service function about a problem with their machine but the telephone representative determined that the problem was actually supplies or partsrelated, the customer would have to be transferred to Supplies. If, however, it turned out that there was a problem or unresolved issue with the customer’s account that prevented Supplies from placing the order, then they would be transferred yet again to Customer Administration. When, and if, the problem with the account got resolved, the customer would then be transferred back to Supplies to finally place that order. As a result, many employees became dissatisfied with the care that they were providing for their customers. Because of functional specialization, they did not have a full understanding of how customer needs were handled within the company; additionally, when they did understand what needed to be done to solve the customer’s problem, their decision-making capacity was severely circumscribed by functional barriers. Since employees could not take the actions they recognized as necessary, they often had to deal with frustrated customers from multiple hand-offs, and the cumulative effect was that employees did not have a sense of completion, of having truly taken care of the customer’s needs. Besides these issues, INK also believed that they were missing out on revenue-enhancing opportunities because of their functionally divided support process. With the functional approach, less than 20% of the telephone customer service employees were selling products and services, and only 12–15% of the corporation’s 22 million annual telephone customer contacts were handled by those employees, most of whom worked in the Supplies. If nearly 100% of your customer contacts were, in fact, potential opportunities for selling supplies, and you could involve 100% of your employees in trying to take advantage of those opportunities you would then be able to incrementally increase your supplies sales revenue. However, to do this, INK required an approach where all of the employees would have the capacity to sell products and services.

Integrated Customer Service

185

The “Integrated Services” Alternative – And a Learning Challenge INK decided to consider a radical reengineering of their customer telephone support: a “one face to the customer, one-stop shopping” operation. This would mean a new role for their call center employees, where each of them would be proficient in all three customer support areas – equipment service, supply ordering, and account administration – and able to handle any problems customers might have (Sprague, 2009). The approach would be called “Integrated Customer Services,” or ICS. To develop this integrated approach, they established a reengineering team (later termed the “program team”) of some ten call-center employees, including two managers drawn from across all three of the functional organizations. A one-year pilot project – essentially, a field experiment – was planned to test the ICS model in an operational context. The model would be tested under real business conditions by having the fifty or so employees from the three organizations handle real customer calls in all three areas of work. The performance of the pilot participants in each area would be measured against that of functional specialists. In addition, the integrated model of work had to be scalable beyond the fifty pilot participants to a work force of 4,000 employees located at several different customer support centers in the North America; this kind of scaling-up was the expected next step should the pilot prove successful. We were consultants on the pilot, working as researchers in this instance for the Institute for Research on Learning (IRL), and joining the reengineering/program team to employ ethnographic research as a resource for understanding in detail the nature of all the tasks that an ICS employee would need to master in order to do truly integrated work, including the tools they would need to effectively support it. (We were brought into the project by Gitti Jordan, another IRL researcher, who led a project at the same facility the prior year; see Jordan, Chapter 20 in this volume.) Before the pilot even began, INK recognized that any major change in organizational strategy like ICS would have significant start up costs associated with the transition; it would cost money, to save money down the line. For instance, to enable their expanded roles and responsibilities, improvements in the company’s technology infrastructure were needed so employees could access all its information systems. More than this, though, the ICS approach would require more knowledgeable employees. Simply giving them access to systems could not ensure that. Employees would have to learn what to do with the information they could now access, and they would have to master the processes for doing it – for supporting customers while

186

Making Work Visible

interacting with them in a time-critical environment on the phone. Solving this learning challenge in a cost-effective way would turn out to be the ICS project’s most valuable contribution to the company. INK’s natural response to the learning challenge was to engage their Corporate Training organization. Training was asked to design, develop, and execute a curriculum that would meet the needs of the Integrated Customer Services initiative – enabling employees who currently knew how to do one job to now do three – without breaking the pilot project’s budget. Unfortunately, the structure of Corporate Training almost immediately emerged as a critical barrier to developing an effective solution. Similar to other corporate offices and functions, Corporate Training was organized into functional divisions: one for gathering requirements, a second to design the curriculum, a third to develop or write the curriculum, and a fourth to deliver or teach the curriculum. The entire corporate training process required nearly forty hours of preparation in order to deliver one hour of curriculum. This made the use of their services rather expensive, and their reliance on classroom instruction also made the delivery itself very time-consuming. Moreover, this process often did not produce results that line managers found satisfactory. A common criticism was voiced by a manager in the company’s billing organization, who had decided to bear the time and expense of training his own people rather than use the training organization: You ask me what I want and you give me what you have. You cost too much and take too long. And by the way, when my people come back from training four months later they still can’t process a bill.

These time and expense factors were a problem for the ICS pilot, which needed to move quickly and efficiently in designing and executing a learning plan to meet its one-year schedule. A survey of the existing classroom-based, specialist curricula for the three functional areas suggested that it might take an ICS employee as much as fifty weeks to cover the necessary material if using the traditional delivery method. This was not a practical learning solution for ICS, as the company had to continue to meet business and operational objectives even as the nature of the operational environment and structure changed dramatically. It had been determined that employees participating in the pilot had to take with them fifty percent of their operational workload from their previous functionally specialized organization. Therefore, the learning had to take place while pilot participants continued to perform, at least in the beginning, their usual work tasks while simultaneously learning new ones. In this situation,

Integrated Customer Service

187

the participants could not spend an extensive amount of time in the corporate classroom and hope to retain the necessary knowledge and skills to be immediately effective on the floor. Moreover, the tight schedule of the pilot required that the employees learn most of what they needed to know in just fifteen weeks, not fifty. It became clear that any effective solution would require Corporate Training to work very differently; in effect, to reengineer its own work process. Moreover, not only would Corporate Training have to change its ways of working, the traditional kind of curriculum as well as the methods of teaching and learning traditionally used by the company would have to be replaced by something very different. The learning needed to be functionally integrated while still significantly reducing the cycle time from the requirements gathering process to the actual delivery of instruction. It would also have to reduce the huge learning curve currently required for employees to master the curricula. To deal with this last challenge, we recognized that the curriculum had to be based on a detailed understanding of the actual work practices and the everyday, practical knowledge that employees relied upon to get the job done. That is, the focus of the curricula had to be grounded in real-world engagement: what the ICS employee would actually do on the phone with a customer when performing any specific function and using a particular system (e.g., canceling a maintenance contract, changing an address, selling supplies, handling a request for equipment service). It was apparent from our own participant observation as learners in some of INK’s conventional training courses that most of the course content missed any systematic focus on learning about the requirements of INK’s customers and their practices. As a consequence, we recommended that the preparation by Corporate Training staff for all learning events begin with observations of the actual INK work environment. In this way, customers’ requests could be comprehended and represented in whatever curriculum was being prepared for the pilot. This was our preliminary (and informal) attempt to transfer some work practice analysis capability to the Corporate Training team. Later in ICS, pilot participants – those who were veteran employees of their functional organizations – actually helped to write most of the streamlined curriculum, as they knew from experience what people most needed to learn and why. The elementary work practice analysis by the Corporate Training team was helpful in designing and developing the curriculum for customer account administration. It was in this context that pilot participants had reported they were having trouble encoding the customers’ requests on the

188

Making Work Visible

telephone after the classroom learning events (e.g., “I have trouble understanding what the customer is even saying to me”). Consequently, we recommended that the Corporate Training team build a catalog of customer requests and prepare course materials to understand those requests. This would make customers’ words and activities intelligible to new learners.

A Radically Different Learning Strategy For the first few months, participants had to learn the new integrated job while still doing fifty percent of the workload from their old jobs. This problem, equivalent to “changing the tire while the car is moving,” required a very different strategy than that usually followed in corporate training programs: learning had to be moved out of the classroom and into the workplace as much as possible. Classroom instruction would still be an important delivery method for learning, but it could no longer be the primary method. This alternative strategy was based on a social, interactive model of how meaningful and effective learning takes place, recognizing that “knowing how” depends on engagement in real-world practice (cf. Lave and Wenger 1991; Jordan 1997). In this regard, we made use of our ethnographic observations (carried out almost every day over three months), including some 40 hours of video recordings of work practice in the functional organizations. This research revealed how employees naturally and routinely collaborate with and learn from each other while engaging in everyday work. The data also showed how new employees regularly learn from experienced workers. Taken as a whole, our ethnographic data and recordings demonstrated that people learn by observing others, by doing real work with experienced peers, and by monitoring others in order to collaborate, solve problems, and seek help. During our very first days in the field, we discovered – through the ethnography – a decisive event that informed the new learning-strategy design. Stella, a customer services representative (CSR) from the equipment service organization, had taken an ordinary customer call for machine repair service. The work environment in her section of the facility was fairly typical of call center operations, with 250 people located in cubicles with low walls, tethered to computer monitors and telephone with headsets on short cords. On this particular day many people were moving to a new work area – packed up cardboard boxes filled the cubicle in the series of pictures in Figure 10.1. Stella was working in a cube that was not her usual space, so she did not have her normal work support documents with her – resources that she commonly used to help answer questions or describe policies and procedures. The customer is requesting service on his machine, but it is quite

Integrated Customer Service

189

Figure 10.1 Natural monitoring and collaboration

late in the day. The problem for Stella is to determine whether or not the customer wants to pay the extra $225.00 fee to have the technician service the machine after-hours. In frame 1, Stella does not have the necessary document to help the customer, the pricing information for after-hours service. In frame 2, the colleague on her left notices that she is having a problem. She is saying: “Let me put you on hold for a minute.” In frame 3, she unplugs, and then in frame 4 she walks away from the workstation to retrieve the document from her own workstation. Then her colleague extends the needed document to her and hands it over as she returns to her workstation (frames 5 and 6). Even though her colleague was busy at his own workstation when Stella received the customer call, he is able to overhear Stella’s side of the exchange with the customer, figures out what she needs to answer the question, sees that she does not have the information at hand and, without being asked or even noticed with respect to the initiative he now has underway, offers to Stella precisely the document she requires. We quickly identified this sequence as the natural monitoring and collaborative activity ubiquitous in call center operations, and this episode became the design inspiration for the learning strategy we later developed with the corporate training department and the ICS program team (cf. Whalen and Zimmerman 2005; see also Jordan 1997; Goodwin and Goodwin 1996). Observing repeated events like this led us to conclude: *

The integration of learning and work is crucial for performance on the floor, and the work environment must become a true learning environment.

190 *

*

Making Work Visible Peer coaching is critical in the transfer of skills and knowledge from peer to peer and from group to group. The design of the physical work environment must support learning, and enable quick collaboration between peers and among groups.

This research resulted in the ICS learning strategy which was organized around two essential elements: learning by observing and learning by doing. These two elements were part of the learning process for each content area and allowed pilot participants to continue to perform their current functional tasks while developing integrated work practices. Learning by Observing “Learning by observing” was based on the recognition that learners need a real-world context for understanding any instruction or materials they receive in a particular function. Our research suggested that it would be helpful to expose learners – even before they begin any classroom activity – to the natural work activity for that function. In this way, ICS learners would gain a rudimentary understanding of the most common types of calls that employees receive in that area. They would also have some idea of what these calls looked or sounded like, and how they usually unfolded and got resolved. In short, this observing activity exposed the new learner to customer requirements and practices, which then formed the bedrock for learning the processes for responding to and satisfying those requirements. Learning by observing included both unstructured and structured methods. Unstructured observation simply means giving employees, while they are working on tasks within their own functional domain, an opportunity to watch and listen to other employees doing other functional work. This unstructured observing allow employees to learn some of the most common and important features of that other function and continues as the employee begins to perform the new work tasks; it is a constant source of practical knowledge. Pilot participants found this process quite valuable. As one person commented, Whenever I’m sitting here doing my [wrap-up function], I can be listening to somebody else on the phone call and they’re in their work area working on their functional thing, and just hearing their verbiage and the kind of things they ask and go through helps me. The next time I get a call like that I can say the same kind of thing. You know, we’re kind of like parrots.

And as another pilot participant summarized it,

Integrated Customer Service

191

You learn, sitting there in “available” [not on the phone and available for the next call]. Just listening you can learn!

It became clear that an essential condition for unstructured observing had to be the deliberate organization of the pilot participants into fully cross-functional and colocated work groups: each group needed to have at least one member from each of the three functional areas and group members would have to work side-by-side, giving them immediate and easy access to each other. Structured observation involved having learners sit with and observe, on a scheduled and thus more formal basis, another employee performing the tasks that they themselves needed to learn. During their structured observing activity, learners were expected to develop a list of questions and concerns, as well as suggested areas of concentration (“what I think I really need to learn or understand”). Structured observation was a way to familiarize the learner with the nature and scope of a new task before a classroom-learning event. We came to emphasize this type of “before class” observing after receiving employee feedback to some early learning events in the pilot, where the employees stated that the instruction they received did not have any practical grounding in the actual work task being taught. This feedback also led us to recommend that structured observation should be used to help the learner gain reinforcement after the classroom-learning event, by scheduling time to observe another employee who would then demonstrate the new work task that had been covered in class. The Program Manager for the ICS pilot was enthusiastic about the approach: By taking the learning process out of the classroom and putting it into the work environment, employees were able to learn faster, more comprehensively and they are able to put their learning into practice quicker.

Pilot participants were also enthusiastic about the change: I can’t believe that we learned as much as we did as fast as we did. The typical [training situation in our company] was classroom – eight hours a day for several days. By the fourth hour of the first day you are already zoned out. The way we trained here was you were in class for four hours and on the phone for four hours so that could put into practice what you were learning that day so you didn’t forget. You absorbed it and you got to use it!

Learning by Doing “Learning by doing” was based on the recognition that employees naturally and routinely collaborated with and learned from each other while engaging

192

Making Work Visible

in everyday work. It has long been understood that most employee learning – perhaps as much as 80% – takes place on the job. The problem is discovering precisely how this learning-while-working occurs in each setting, among specific groups of workers; our research in the functional areas was designed to uncover precisely that. A crucial component of both “learning by observing” and “learning by doing” in ICS was the taking of live calls from the very start of the pilot. Initially, the plan was for pilot participants to learn through simulated calls, classroom training, work “group” discussions, and the like. However, we recommended that live calls be used as soon as possible in the pilot, and that the learning associated with this activity be linked to the step-by-step integration of work processes. This required a process integration strategy that used call routing and distribution technologies to initially distribute calls, so that employees had a dedicated workload that was closely matched to what they were learning (for example, calls to “800 Supplies” in the first stages of the pilot would be routed only to employees with prior Supplies experience). This would enable pilot participants to learn by observing fellow work group members until they were qualified to begin handling some live calls in functions other than their own. In addition, in order to make the learning process more effective during this early stage, the number of live calls that the pilot received was reduced to allow sufficient time for review and discussion by the employees; moreover, the hours during which employees received calls was also reduced to allow sufficient time for classroom work and other activities. Once employees were at the stage where they could begin to handle calls in a function different from their own, the “learning by doing” approach was used to systematically introduce, in an orderly function-by-function fashion, new processes and tasks. Since service calls make up close to 70% of all the customer calls received by INK, we recommended that they be the first process integrated into the pilot, because mastering this process and volume was a crucial first step in testing and validating the ICS concept. Related to this, we proposed that supplies selling inbound calls should be the second process to be learned and integrated, followed by the more complicated – with respect to the nature of the work processes as well as the IT system (and its ninety different screens) that had to be mastered – customer administration inquiries, both written and telephoned.

Phased Interactive Learning We named this conjoining of learning-while-working with the step-by-step, carefully phased introduction of new work processes Phased Interactive

Integrated Customer Service

193

Learning, or “PhIL.” It is “phased” because learners continually shift between distinct learning situations (such as classroom activities, or group discussions) and work situations where they can immediately put that learning into practice in the field; each new area of learning is introduced in the ordered or phased manner that best supported the learning-while-working approach. As a participant in the ICS pilot put it, “The way we did it, we learned a little, we did some practical [work], we learned a little more, we had reinforcement.” The PhIL process is intensely “interactive” because employees learn from and teach each other, or learn together with instructors, through social interaction. This is true whether the learning activity is taking place in the classroom or on the floor. Thus, classroom activities cannot be organized largely as lecture instruction, as this delivery method rarely engages the learners as fully active participants in the activity, as true collaborators. In this sense, Phased Interactive Learning takes as its foundational principle that learning is, at its core, a fundamentally social process. It should be emphasized that peer coaching, both in class and, especially, on the floor, was a crucial component of PhIL. It was essential that employees have the time, while doing their work, to coach and teach each other; that is to say, if peer coaching is to be successful, it requires operational support and structure. Pilot participants describe the uniqueness of this coaching experience and its value: The peer coaching’s so different than anything I’ve ever seen within INK, or frankly anywhere. . . . we had very little classroom, maybe two to three days per subject, max. Then you hit the floor and you start getting the repetition to use what you’ve learned. And on the same side of that you’ve got peer coach on one side and peer coach on the other side. Somebody’s always there if you’ve got a problem, you’ve got your answer right there and you learned – you learned while doing it. One of the biggest pluses in the pilot was the ability to tap into the knowledge base of the other people around you. . . .It was “on the job” more than classroom and it was peer coaching.

Because we could not realistically expect that everyone would automatically know how to best teach others, we created a performance aid for employees to use whenever they needed to instruct one of their peers in a new process. New Roles through PhIL If most learning takes place while engaged in real work and depends on employees teaching and learning from each other, a true Learning

194

Making Work Visible

Table 10.1 Seven-step peer coaching model 1. Demonstration – show whole task, relevance, decision points 2. Review Key Points – key points: use, relevance, steps in process, systems used. Q&A 3. Second Demonstration – identify key points, analyze actions, Q & A 4. Learner performs simple parts – complete component part of task, know what to ask customer, and so on. 5. Coach guides learner to perform whole task – start to finish with plenty of guidance 6. Learner performs whole task with coach observing – coach close by for assistance and feedback 7. Learner performs independently

Organization must find ways to recognize, officially support, and significantly enhance this essential but eminently natural activity. We realized that as the organization transformed its educational strategy from one based on a training paradigm to one based on a continuous learning paradigm, the traditional role of “trainer” would also have to change. The question was, how could people whose previous role was tied to the traditional, classroomcentered training paradigm change that role to support the new, “integration of learning with work” approach? We argued that instructors should become learning activity facilitators (LAFs) in all operational settings, in addition to their responsibility for classroom teaching. The concept of “facilitator” suggests a more supportive and collaborative (rather than directive) approach to learning than does “trainer.” Most important, LAFs needed to play a major, proactive role in implementing new programs (including the integration of new technologies) and serve as catalysts for behavioral change in the organization, rather than have their role be limited to traditional (and largely reactive) “knowledge transfer.” Further, the LAF role should not be limited to people who were previously trainers; ICS Pilot Program Team members should serve in this capacity as well. Because these Program Team members serving as LAFs, were based in an operational environment, we felt they should continue to perform operational functions; that is, engage in real work, at least for some portion of their time, in order to sustain and develop their skills and knowledge base. Most important, this would allow them to maintain their membership in the local “communities of practice” (Wenger 1998) – the work communities in which they had long participated (cf. Bobrow and Whalen 2002). This

Integrated Customer Service

195

community membership would be the vehicle through which their facilitating role was most effectively carried out. Finally, it was agreed that learning facilitation responsibilities could not be limited solely to LAFs. Although the role of managers as coaches was still conceptually underdeveloped at the time of pilot launch, an essential part of the PhIL strategy was for manager-coaches to proactively facilitate and support learning in their work groups rather than simply manage them in an operational sense.

Configuring the Physical Environment for PhIL It was important to organize the physical environment, the actual work setting of the employees and their managers, so that its configuration would effectively support collaborative learning and the new work practices. To address this need, we came up with a “space for learning in call centers” design that was based on having as few barriers to sharing information and working together as possible. For example, work area walls would not exceed 48 inches which would allow for seated privacy without restricting interaction across them. Most important, telephone representatives would not be located in individual cubes; rather walls would be used primarily to provide a boundary to support the work groups. This arrangement of space supported the cross-functional work group concept because the collocated group provided the organizational and environmental framework for collaborative work and learning. In addition, locating managers in cubicles on the floor, fully integrated into their work group areas, rather than in offices around the periphery of the center had “learning facilitation” advantages. Pilot participants, including the managers, found this configuration especially helpful to their work and learning: I love the way it’s set up with the low walls where you’re in your work group, and you feel more of a family sense. You’re not in a cubicle all by yourself where your workgroup is all spread out in this whole long row of pods. You are all right there together; you have that unity. You can bounce off a lot of ideas – you can call impromptu little chat sessions to say: “Okay this just happened, it’s a great success story and I want to share it with y’all right quick, ‘cause it is really going to be a big performance builder.” [Also] having all the different functional people with us has been such a good base of knowledge to draw from. The way we had the physical environment set up here really facilitated the people in the work groups to work together because the walls were low

196

Making Work Visible

enough . . . to interface. The whole fact was that they could turn around and ask the question, or they could see the person and say, “Come here.” Or just mouth it and have the person come over, that person could understand what was happening and point to the screen or actually punch the buttons on their PC. You couldn’t do it in a separate office; you almost have to have lower cubes.

Further, to enable easy and effective collaboration – and thus peer-to-peer teaching – we gave ICS pilot participants extra long telephone cords. This allowed them to roll their wheel-equipped chairs around their work group’s shared area, which now had no barriers between its individual members, whenever another member asked for help. They could easily see the other person’s computer screen and work side-by-side with them while still connected to their own call-management system. The organizational cost of collaboration was minimized without hampering the peer-to-peer learning process. From Task Work to Knowledge Work The larger and more global goal of the empowered learning and work group strategy was not to simply acquire additional skills and process capabilities. Instead, the goal was to support participants in their desire to do a different kind of work: knowledge work rather than simply task work. Their opportunity was to become an INK businessperson rather than a narrowly

Figure 10.2 Phased Interactive Learning in action

Integrated Customer Service

197

functional employee. The ICS project accomplished this by transforming the organization’s work process, captured in the quintessential moment of Figure 10.2. In this illustration, three of the ICS work group members, each from a different functional area, are in the large pod that surrounds them as a group. They also have a visitor who is observing them. Denise, seated front right just behind the monitor, is from the Customer Administration area, and has just received a call requesting service on a machine. Terry, standing back right, is a service expert and overhearing Denise taking a call, rises from his workstation to coach her in an impromptu fashion. Audra, seated at the far back left, is from the Supplies area and is listening in and observing both the coaching and the customer service call. Carolyn, a Vice President from the Training Organization who is seated front left, is in the first step of the Phased Interactive Learning Process; she is learning by observing the work group as they process customer calls from all three functional areas.

Integration in Action We began this chapter by recounting a customer call to INK that nicely illustrated many of the problems with a functionally divided customer support process. The experimental fully integrated ICS process-in-action gave a much different picture of what was possible with customer calls. The ICS call in Excerpt 10.2 begins normally enough, in that the caller has telephoned the Supplies organization’s 800 number (remember that during the ICS pilot, calls from the 800 numbers of all three functions were routed to the ICS facility), wishing to place an order. To this point, the call appears quite similar in trajectory and tone to the one at the beginning of the chapter. The questioning by Karen – an ICS pilot participant – following the caller’s explanation about wanting to place an order is aimed at meeting the standard information requirements for supplies. Once the requested information has been exchanged, Karen prompts the caller to describe what supplies she needs. The talk that follows is equally “standard” for supplies order calls: The caller says she needs lift-off tape for a particular series of typewriter; Karen responds by telling her how many tapes there are per box and then asks, “How many boxes?”; the caller answers she just needs one; Karen confirms the typewriter series number but also points out to the customer that INK records show (here Karen is looking at information displayed on her computer screen) that her office has another typewriter series as well, and then

198

Making Work Visible

Excerpt 10.2 INK customer service order

Karen: Cust: Karen: Cust: Karen: Cust: Karen: Cust: Karen: Cust: Karen: Cust: Karen: Cust:

Good afternoon, INK, this is Karen. How may I help you? I needed to place an order please And I need your customer number, please Eight seven zero four, zero eight six six five. And your name? Linda Wagoner (.) Okay Linda? (.) Is there a purchase order number I can reference? Just my name’s fine. Okay, you’re at 805 North Main? Muhhum? And your phone is six six eight, four two two two That’s right. (.) Okay Linda, tell me what you’d like to order today. Okay I need . . . (begins to explain what supplies she would like to order)

asks if she would like to “mix and match” tape boxes for both kinds of machines in order to “get some [quantity-based] price breaks?” Employees working in the supplies call center are taught this sort of “upgrade” sales pitch – where you try to get the customer to increase the order amount by explaining the financial advantages. Such pitches are more likely to succeed when you know all the equipment the customer has on site and their order history, which then allows you to craft the pitch to actual opportunities, to what the customer probably needs. This knowledge can only be gleaned from careful, albeit very quick, reading of the customer profile and order history in the supplies system database – which is dependent on obtaining the basic customer information Karen requested at the start of the call. In other words, Karen has demonstrated the knowledge and skill expected of a good functional employee of the supplies organization, and her interaction with the caller continues to exhibit the expected features of a typical supplies order call. But this is what we would expect of her, as prior to the ICS pilot she was a top performer from the supplies organization. Before she can respond to Karen’s upgrade pitch, the caller explains that she has to check her current supplies status. When she returns, she notifies

Integrated Customer Service

199

Excerpt 10.2 (Continued A)

Karen: Cust: Karen: Cust:

Eight and a half by eleven? By fourteen? Eleven by seventeen? Ahhh(hhh)- a li- a- a little of- a little of both here. (laughs) (laughs) Um. . ., I need some of the eight and a half by eleven three hole punched. (0.5) Karen: Okay. (.) Karen: That’s two one seven nine three three. Cust: And umCust: I’m not sure of the number (referring to that re-order number) Karen: I have it. Cust: I also need . . . (goes on to order other sizes and types of paper)

Karen that she actually does not need any lift-off tape: “Actually, when I went out and looked I found another box [of tape],” she explains, and then adds, “so I’m gonna hold off on and wait until I get a little low on both at the same time.” So the upgrade offer worked, insofar as the caller will try to take advantage of the quantity price break in the future. But the caller has also decided to now order some paper for her photocopier. Karen inquires about the paper size, and once again, a fairly “standard” supplies order exchange takes place. Here too Karen readily leads the caller through the necessary steps for placing her order, presenting all the information required for decision making on the caller’s side and order-entry on the INK side. Once again, a more-or-less typical supplies call. It is only when the caller states that her order is complete that the call takes a very different, “integrated” turn. And it is Karen who introduces the caller to the opportunity for such integration, almost immediately following her explanation of when the caller can expect her paper order delivery. Karen does not simply make a cursory attempt to extend further services to the caller; rather, in the face of the caller’s initial “that’s all I need” declination she reformulates the offer, moving from a generic offering of services to suggesting what additional kinds of services are available – “Account information . . .” – while also making clear it could actually be

200

Making Work Visible

Excerpt 10.2 (Continued B)

Karen: Cust: Karen: Cust: Karen: Cust:

Karen: Karen: Cust: Karen: Cust: Karen: Cust: Karen: Cust: Karen: Karen: Cust: Karen: Cust: Karen: Cust:

Karen: Cust:

Okay, well Linda, these are going to be delivered for you within three to five days or less Okay Anything else within INK, any other services I can provide for you today? hmm, that’s all I need Account information, anything at all? Oh, actually, there was one thing, yeah. (soft giggle) I need – I don’t know if you can help me or if I need to be transferred, but I do need a service call on one typewriter Oh ok. (.) And, is it the C eight zero one five eight four three three? No, actually it’s the eighty R oh oh three five oh four. (1.0) Hmm. Which one is that It should be a sixty two series Eighty R? yeah Zero zero three five oh four? Yeah (.) Well that works And the problem you’re having? Well, Maintenance, cleaning, or – He had been out a while back because some of my keys were stickin’ and he just completely replaced the whole keyboard? Mnhm And, hhhh he was out just the other day for some other reason and I just happened to notice while he was here – somehow now when you use the correction thing to erase (a letter) it doesn’t quite erase the very top of it and then when you go and try to type something over it doesn’t look very good. And, I had him look at it when he was here. I, you know, thought it was fixed but it’s doin’ it again. Okay. So it won’t erase the top of the letters? Yeah. (0.5) I think it’s just probably minor adjustment

Integrated Customer Service

201

Karen: And that’s on your liftoff? Cust: You don’t think you need to replace them maybe, you know how they can be defective, have you tried putting a new one on there? Cust: No. . .actually no I hadn’t tried putting a new one on Karen: I’m going to go ahead and place a service call by the time he calls maybe you will have tried? Cust: Huhum ISCR: and if that’s the case then we can certainly accommodate you Cust: Okay. Karen: and send you another one to replace one if it was defective Cust: Okay. (call concludes soon after) “anything at all.” This action by Karen goes to the very heart of ICS: INK’s customers can get support with and for “anything” they need in just one call, speaking with just one INK representative. As ICS is just an experiment, callers have no advance awareness of this opportunity. They have dialed the usual 800 number for whatever functional organization they wish to reach and so the ICS pilot participant has to find a way to reach out to them at the conclusion of that functional business, to then make clear their opportunity to do “anything.” While this seems to be an unexpected opportunity for the caller, in the face of Karen’s persistence, she does in fact begin to say what else she needs by telling a story about her typewriter. Notice that in responding to this story, Karen very quickly negotiates the service-call handling system, and after a few seconds of searching identifies the specific machine in question by the machine number the customer provided. Then Karen suggests a simple fix: try replacing the liftoff tape cartridge, as sometimes they can be defective. When the caller indicates that she has not tried it, Karen quickly offers some reassurance that she will “go ahead and place the service call,” and suggests that she try to replace the cartridge. In this last, integrated activity of the call we can see Karen’s demonstrated skill in now processing a service call in the expected and “complete” manner. This observation is paramount; a key test for the ICS pilot was whether participating employees like Karen could successfully learn and perform the work required for all of the functions at a level that was at least equal to that of single function employees. Plainly, Karen has now mastered all the skills expected of an equipment services employee. Moreover, in handling a fairly complicated upgraded order for several different types of paper, placing a

202

Making Work Visible

service call, and also making suggestions that might fix a machine over the phone, Karen has achieved the ICS objective: processing customer calls in a completely and seamlessly integrated manner. It is important to emphasize that Karen’s skills were quite representative of all employees in the ICS pilot. It was our experience that employees from each of the functional organizations were able to make the transition to processing integrated calls. While there were executive-level doubts, we found that the background knowledge and skills that people brought to the pilot actually enhanced their abilities in other areas. For instance, those with customer account experience sold supplies by framing it in terms of solving the customer’s problem, as one participant from accounting later explained to us near the conclusion of the pilot: This particular month I have the highest selling average in our work group. I’ve surpassed the goal. And it amazes me because I just didn’t think I was the selling type person. But, when you’re on the call, it’s not a sell or it’s not saying, “Give me your money.” It’s [rather] “this is the customer’s needs,” whether it’s placing a service call or selling them supplies. It’s not, “I’m trying to give you something you don’t need [or] you don’t want and push it off on you.” It’s “I know you need this. I’ve looked at your history, you haven’t ordered dry ink in seven months and from your average copy volume, you need to order. So let me help you and make sure you have it stocked on your shelves.”

And for the employees who came from equipment service, it was their organic knowledge of the machines that enhanced their selling capabilities. They knew that certain problems with machines were related to improper supplies and knew what the customer needed to keep their machines running properly. In other words, what some corporate-level managers thought would be an obstacle to the employees learning a new and possibly more challenging job turned out to be an asset. The learning strategy for ICS was able to take advantage of this in ways that simply would not have been possible if a traditional training approach had been used.

The Legacy and Significance of ICS Unfortunately, at the conclusion of the pilot, the management of the functional customer support organizations in INK decided against implementing ICS, due almost exclusively to difficult (and inevitably politicized) organizational transformation issues. That is, despite the demonstrable success of the ICS pilot, it was simply too daunting to face the transformation of

Integrated Customer Service

203

three functional organizations, each with a long history and deeply vested interests, into just one organization; to take three functional vice-presidents and select one integrated VP, for example. Moreover, the corporate champion for ICS, the executive who first proposed the idea and fought to get the pilot experiment funded, had left the company. Thus, without a champion, the ICS idea hit organizational and political barriers it could not overcome, regardless of the way pilot participants had performed at their new integrated jobs. The failure of INK to carry out an organizational transformation based on the ICS pilot was in no way due to work performance problems among pilot participants. Nor was it due to the project’s reliance on a radical learning strategy, PhIL. Indeed, PhIL was enormously successful and the performance of pilot participants was exemplary. The same can be said for the key role in the pilot experiment played by ethnographic research. It would certainly be worthwhile to undertake a detailed analysis of why organizational transformation efforts that show such promise early on, at the pilot or experimental stage, are never carried forward. That would require a careful look at organizational politics and power relationships. As the ICS pilot was drawing to a close, and before we learned that INK was going to decide against implementation, we had many planning sessions with the pilot Implementation Team about how an implementation should be organized. From our view, success going forward depended on following the same learning principles embodied in PhIL, with an emphasis on peer-to-peer sharing of experience and practical expertise across the different support centers rather than any conventional “roll out” that was directed from the top. The true legacy of ICS can be seen in the Phased Interactive Learning Strategy that did actually move forward along several important dimensions in INK. On a local scale, the pilot participants took PhIL back to their functional organizations and used it to enhance their colleagues’ learning capabilities, especially as corporate training budgets were reduced. In addition, the PhIL model from the call center environment was adapted to train INK field technicians – employees who service and repair the company’s products in the field at customer sites – which resulted in savings to INK of close to $20 million annually for a number of years. Further, the new, customer-focused curriculum development process helped transform the way the training organization did its work. And most important of all, this led to a new learning strategy for the entire corporation. After seeing the results of the pilot and personally hearing the accounts of the participants regarding their learning experiences, the company CEO pronounced, “This is the way [all INK employees] should learn now.”

204

Making Work Visible

The significance of ICS can be found in these practical results, both during the pilot and afterward. There have been many studies where ethnographic methods and “workscape analysis” are demonstrated or where the capacity of this approach for generating valuable findings about work practice and the knowledge underlying that practice is proven. We have chosen to tell a story about how such findings can be translated into real organizational change – how work environments can be rearranged (the physical together with the social), how employees can learn to do their jobs and master the knowledge, how conventional training strategies and techniques can basically be turned upside down – in eminently practical ways, in the hope that this story would inspire others to put research into action.

11

Interactions at a Reprographics Store Erik Vinkhuyzen

Around the turn of the century, the Knowledge Interaction and Practice Area was a newly formed group of social scientists at PARC with a background in conversation analysis and ethnomethodology aiming to conduct workplace studies for Xerox (see Sharrock and Button, this volume for an overview of the theoretical underpinnings of this approach to studying work). We were keen to pursue a team project so that we could create synergies among our different backgrounds and research interests. Our goal was to secure a business fieldsite for a long-term research project that provided opportunities for interesting research and value to our parent company. Of course, there is a tension between the long-term open-ended nature of ethnographic research and common business practice. In business, cycles are short because the performance of organizations – and by extension the performance of the sponsors of research – are publicly scrutinized every quarter. The pressure to make headway in some measurable way in a short amount of time – months, sometimes weeks – is great. We were adamant, however, to try and define a project that would satisfy both research and business criteria. Eastside reprographics, a local chain of copy shops in the San Francisco Bay area,1 seemed to match our needs well. First, it provided us with ample opportunity to observe people interact with Xerox machines and thus to report on any usability issues. Second, and of interest to Xerox sales, the research gave us an opportunity to observe the inner workings of Eastside stores and thus insight into what additional services Xerox could offer them. Third, since Eastside was one of Xerox’s largest clients, our offer to conduct the research study for free fit well with Xerox’s desire to keep Eastside 1

Eastside reprographics is a pseudonym; Eastside is a local competitor to outfits such as Office Depot and CopyMax.

205

206

Making Work Visible

happy. As one sales person put it: “What is good for Eastside, is good for Xerox.” Fourth, as a retail shop with a variety of services, Eastside offered exciting possibilities for doing workplace ethnography. The Xerox sales person responsible for the Eastside account set up a meeting with their management and they jumped at the opportunity to have us conduct ethnographic research in their stores, as they yearned for insights from the field. We promised them no specific deliverables, only that we would regularly report our findings, and so we embarked on an open-ended study of Eastside’s operations. We worked with Eastside for nearly three years and it proved a most rewarding research site. This case study recounts the project, how we conducted the work, what data we collected, the research topics we explored, and some of our findings, and the intervention we codesigned for Eastside.

The Research Team Our research team expanded and then contracted over the course of the project – we started with four core researchers, grew to nine researchers, but due to layoffs and attrition, we ended the project with the same researchers we started with. We were an interdisciplinary group with diverse backgrounds in the social sciences: sociology, psychology, and linguistics. In previous work, the group members had studied a variety of workplaces, including call centers, 9-1-1 operators, bilingual classrooms, survey centers, pharmaceutical organizations, the work of service technicians, insurance agents, and mission control officers (Whalen, Zimmerman and Whalen, 1988; Vinkhuyzen, 1999; Vinkhuyzen and Whalen, 2007; Whalen, Whalen and Henderson, 2002; Szymanski, 1999).

Method The researchers shared a commitment to the naturalistic study of work and workplaces; we preferred the direct observation of work practices over interviews or other research instruments. And we preferred video recordings of natural practices “for they serve as an important control on the limitations and fallibilities of intuition and recollection.” (Whalen and Whalen, 2004). Our focus was ethnomethodological in that we tried to understand people’s work from their own perspective (Schegloff and Sacks, 1973), it focused on a wide variety of issues, from employee’s interactions with customers to their interactions with artifacts and technologies in the course of their work. We came to call this methodology “studying

Interactions at a Reprographics Store

207

workscapes” (Whalen and Whalen, 2004, Sharrock and Button, this volume; Whalen and Whalen, this volume). Although in hindsight it is possible to discern a number of distinct and logically sequenced phases to the project, we proceeded in a rather ad hoc manner, and always adjusted to the opportunities available and the evolving research interests of the team members. Key to the opportunities was our relationships with the Eastside store managers which were challenging to maintain as several times during the project store management changed. Some managers were more amenable to working with us than others and this determined what stores remained as research sites and which stores were dropped – we conducted fieldwork in five stores, but eventually worked with only two. Roughly, the work had the following phases. Initially we observed several stores, shadowing employees in different roles. We wrote notes and took an occasional picture. We developed relationships with some of the coworkers and store managers. Then we did some exploratory videotaping in different parts of the store, and followed that with an analysis phase in which this data was transcribed, edited, and made into presentation materials. In a subsequent phase we collected video data and other artifacts more systematically. This phase was followed by a long period in which we processed and analyzed these materials, with only occasional visits to the stores chiefly to maintain our relationships with the employees and store managers. We developed ideas about what ought to be changed and improved in the stores and in a codesign phase we worked together with a small group of employees to develop and refine a solution. In retrospect, our research method evolved quite naturally from exploratory ethnographic observations to gain a broad understanding of the store’s operations to doing more and more detailed video analysis to pursue specific topics of interests.

Background: Eastside Copy Shops Eastside copy shops vary in size and layout depending on the space in the building, but they offer the same wide-ranging services: from simple blackand-white copies to brilliantly colored brochures, customized calendars, photographs and large posters, and a range of bindings and finishes. Each store has a do-it-yourself area in which customers can make their own copies. This area is usually attended by an employee who helps the customers with their projects and troubleshoots any problems with the machines. There is another area of the store in which customers can work on

208

Making Work Visible

computers provided by Eastside. Frequently, these computers are used to send files over to Eastside’s printers. Customers can also have Eastside create their documents for them. In this case, they have to go to the front counter to place an order with an employee; the production staff that runs the faster, more complex printers behind the front counter processes these orders. Typically the orders that are left at the front counter are the larger and more complicated orders that can take hours to process, although when the store is not very busy, the front counter personnel may accept small orders as well. Eastside makes most of its money from production work. Curiously, however, Eastside charges the same price per copy in production as it does in the do-it-yourself area. Hence, many customers would like Eastside to make their copies for them, but Eastside does not want to do small jobs in production (unless they are low on work) so they try to send customers with small jobs to the do-it-yourself area (Vinkhuyzen and Szymanski, 2005). Since our project, Eastside has developed a web site on which customers can submit their print jobs but at the time of the study, most work was submitted by walk-in customers. The store managers and the employees are all Eastside employees – it is not a franchise – but the store managers enjoy a great deal of independence. They hire and fire their own people and get paid according to the profit of their store. Each shop is a relatively flat organization, however in bigger stores there is a production manager that tracks the jobs and makes sure they get done on time. Among the employees there is a fair bit of turn over, although not much more than in other low-wage occupations. It was striking to us how different the stores are; each has a palpably different culture, a reflection of the store manager’s style and approach.

Initial Observations After gaining access to several local stores, we began by observing employees in different roles in the stores for several hours to half a day at a time. We observed employees in production that operated the large copiers in the back of the store, employees that took orders and rung people up when they came in to pick up their orders, employees that worked on digital files, employees that maintained the machines in the self-service area, the production manager, and so forth. We wrote field notes about our observations and began to get a feel for the rhythm of the work while developing relationships with the stores’ employees. After several months of doing observations,

Interactions at a Reprographics Store

209

we were allowed to make some videorecordings in the store. We had already come to see the front counter, where customers and employees interact to place and pick up orders, as a pivotal place in the store so we set up our cameras to capture front counter interactions. Upon reviewing these recorded materials, we found a problematic interaction between a counter worker and a customer who came in to pick up her documents. Her documents did not look like she had expected. When she complained the employee told her that the documents were done exactly as specified. The interaction was an example of a common problem and the encounter became a canonical case for our project. Excerpt 11.1 is a partial transcript of the original interaction that has been simplified to enhance readability; the transcript starts after the opening in which the customer gave her name and the employee retrieved the copies. Excerpt 11.1 Half-size pickup

01 EMP 02 CUS 03 EMP 04 CUS 05 CUS 06 07 CUS 08 09 EMP 10 CUS 11 EMP 12 CUS 13 EMP 14 CUS 15 EMP 16 17 18 19 CUS 20 21 22 23 EMP

Here’s the copies. Great. Here’s your originals? Okay, (.) Oh (.) huh huh huh ((laughter)) I- yeah half the size but I needed themI guess I can do that myself But i (.) but it seems like it’s (.) smaller than halfsize though. It’s at fifty percent reduction. It is? Mhmm. It’s too small. That’s theWhat else can I have. Well if you have a specific measurement you wanna give us, we can work with that. But the order form said fifty percent so that’s what we did Well it’s not your fault, I’m just saying that this is- I need something- eh- you know. I want is s- something like this but to fill the page like that I’m not sure what you mean because these already

210 24 25 CUS 26 EMP 27 CUS 28 EMP 29 30 CUS 31 32 EMP 33 CUS

Making Work Visible fill the page Right an’ these don’t. huh huh huh ((laughter)) This is a hundred percent. Uhhuh So you don’t wan’ us- (0.1) you don’t want us to reduce them then I want you to reduce them but I- I (0.3) want it to be like this. So you want us to trim No it seems like it’s too small. ((conversation continues))

Clearly, Eastside had not produced the copies the customer wanted, but how had she ordered them to be done? We searched our video data to find the customer’s order placement. Excerpt 11.2 contains the order taking interaction that had preceded the pickup a few hours earlier: Excerpt 11.2 Half-size order placement

01 EMP 02 CUS 03 EMP 04 CUS 05 CUS 06 07 EMP 08 CUS 09 CUS 10 11 EMP 12 CUS 13 EMP

Hi, can I help you Hi . . . I’d like ten of these please In color? Yes please and then (.) this (.) these these I’d like ehm (.) ((looks at EMP)) half this size? mkay And I just want a copy of each I’m not sure how many there are (.) do you want me to count them? In color also? No eh yeah (.) same color And your name is? (conversation continues)

In lines 5–6, the customer requests the copies, “half this size?” She does so with a questioning intonation as she looks at the employee. This intonation contour is what conversation analysts call “try-marked” (Sacks and Schegloff, 1979) a way in which speakers mark their speech with an upward, questioning intonation designed to elicit a response to confirm that the recipient understands a term of reference or description. The delivery of

Interactions at a Reprographics Store

211

the question thus gives a clue to the employee that the customer is testing whether there is common ground; that is, whether, for the Eastside employee, her description is institutionally adequate. The employee accepts the description without further inquiry: “mkay” (line 7). She then notes on the order form that the copies should be made at 50% reduction and the conversation continues. The employee thus missed an opportunity to explore further what the customer may have been uncertain about: what, exactly, she meant when she said “half this size.” It should be noted that in the document production business the description “half this size” is ambiguous; reduction and enlargements must be specified in percentages on copiers, but when a copy machine is set up to make copies at 50% reduction it reduces both the length and the width of the original by 50% thus making the picture area one fourth the area of the original (see Figure 11.1b) which is smaller than many would expect when they want a document copied to “half this size.” In order to create a picture that is half the area of the original, a reduction of around 71% is more accurate for U.S. letter-sized originals (see Figure 11.1c). If the customer wanted a picture that would fill a piece of paper cut in half a setting of 65% should be used (Figure 11.1d). Given the ambiguities of the “half this size” instructions it would have been prudent for the employee to explore just what the customer meant, especially since the customer herself indicated that she was less than certain about the adequacy of her description. However, the mistake is understandable as well, mathematically “half” is the same as 50% which also is an adequate instruction for the copier. While the order taking was flawed, the pickup interaction was no less problematic. The defiant stance of the employee (e.g., “It’s at fifty percent reduction,” line 9) – holding the customer responsible for what it says on the order form and hence that Eastside is not to blame (the customer accepts responsibility in line 19) – is abysmal customer service. In the subsequent interaction the customer resubmits the order and is charged again for the copies. This example highlighted for us the crucial importance of counter interactions in the overall operation of the copy shop and the importance of the employees’ interactional skills at the counter. The case also provided a vivid illustration of what the organization knew to be the case statistically: a high redo rate of orders and very low customer satisfaction. It pointed to a great opportunity: improving the interactional skills of the employees working at the front counter would impact the heart of Eastside’s business, improving both its profitability and its customer service.

212

Making Work Visible

Figure 11.1 “Half this size”

We presented our findings to Eastside management as well as to the store managers we were working with. They agreed wholeheartedly with our analysis and some store managers were happy to cooperate with us on the next phase in which we proposed to deepen our analysis and codesign a solution.

Systematic Data Collection In order to investigate the challenges at the front counter systematically we needed to collect more interactions between employees and customers. We obtained the permission from two managers to record extensively in their stores. We set up multiple cameras to capture the front counter – which was divided into different areas for order taking and picking up orders – and used additional cameras to capture other aspects of the work in the store. One camera captured the area where the customers stood in line, another camera with a wide-angle lens was directed at the production area, and one member of our team roamed the production area to record details about the order

Interactions at a Reprographics Store

213

fulfillment process, that is, how the production staff translated the instructions on the order form into instructions for the machine. Four employees wore wireless microphones so we could capture their interactions with the customers as well as their interactions with other employees when they moved around the different areas of the store. We fed the signal from the receivers of the wireless microphones into an audio mixer and recorded the output onto the left and right audio tracks of the video cameras. Data such as the exemplar discussed above had taught us how important it is to collect both order taking and pickup interactions for the same order. Since many orders at Eastside are not picked up until the next day we asked permission to record the shop’s operation for two consecutive days from 7am to 7pm. During the filming we also tagged and scanned every order form as we were keen on examining how the interactions with the customers were captured onto the order form and then communicated to the employees in the production area. Proceeding thus, we captured nearly 300 hours of video in DV format.2 To process this enormous data set we digitized all the video in Mpeg 1 format and stored it on large disc raid arrays connected to servers. The whole research team was involved in watching all the hour-long video files and cutting out small segments – usually brief interactions at the counter when customers came to place or pick up their orders, or interactions between employees. Using the order forms we had tagged and scanned we could now link order-placement interactions with order pick-up interactions as well as any video-clips of production personnel actually creating the documents. The information thus organized was captured in an online spreadsheet application called Sparrow (Chang, 1998), a wiki for structured data developed at PARC, that could be edited and viewed by the entire research team simultaneously (see Figure 11.2). In this way we created a large, yet well-organized, data set of interactions easily accessible by all the researchers over PARC’s internal computing network. Most importantly, the database allowed us to study not just individual interactions, but multiple interactions concerning an order as it traversed through the store from order taking to pick up and payment. 2

Given the large number of customers that frequent Eastside’s stores, it would have been impossible to obtain permission from every customer for the recordings without interfering with the regular operation of the store. To inform the customers about our activities we set up many signs throughout the store that stated that if they objected to being recorded we would turn off the cameras. A fair number of customers did not want to be recorded. Because individual consent forms were not acquired, we always disguise the people on the video for presentations.

214

Making Work Visible

Figure 11.2 The Sparrow page with the video clips

Research Themes During the subsequent analysis phase we watched the video clips and made transcripts of particularly striking ones. The large data set allowed everybody in the team to develop his or her own research topics. In the following, I highlight some of the findings in these research themes.

The Grammar of Customer Requests When customers come to the counter to place their order, the opening of the interaction is usually followed by the customer’s request similar to that of other service encounters (Zimmerman, 1992) in which customers describe what they want. There are two main ways in which customers format their request: some started with “I need” as in “I need three copies of this” and others start with “Can you” as in “Can you make three copies of this.” The grammatical formulation of the request turned out to be consequential particularly in case the Eastside employee wanted to have the customers make their own copies in the do-it-yourself area of the store.

Interactions at a Reprographics Store

215

We found that when customers started their request with “I need” it was easier for the employee to respond by sending them to the do-it-yourself area than if the customer formulated the initial request beginning with “Can you.” In the latter case, the employee had to counter the implication embedded in the request that Eastside make the copies, in effect not granting the customer’s request, whereas when the customer used an “I need” formulation the counter worker could send the customer to the self-service area as part of a granting of their request (Vinkhuyzen and Szymanski, 2005). Due Time Negotiations When placing orders at the counter, there is an expectation that the work will take some time to finish and that the customer will have to come back to pick up their order. Therefore, the employee must ask when the customer would like to come back and pick up the order, and note a time on the order form, which would aid the production personnel in prioritizing their work. We noted that, frequently, the employee’s question would not engender the response they were aiming for – an exact time. Here are a few examples that highlight the problem: Excerpt 11.3 Sometime this afternoon

01 EMP When did you need this by? 02 CUS We wanted to pick it up (.) sometime 03 (.) this afternoon? Excerpt 11.4 How soon can you get it?

01 EMP 02 CUS 03 EMP 04 CUS 05 06 EMP 07 08 CUS 09 EMP

And when do you need this? How soon can you get it? Uh, when do you need it? Okay, well I’d like it for this afternoon. Three o’clock, four o’clock, (0.2) five o’clock? Uh three, Okay, uh::m

It was very common for customers to answer the question, some version of “When do you need this?” not with a specific time but rather with a broad

216

Making Work Visible

time frame: “This afternoon,” “Today,” “Tomorrow,” and do so with a questioning intonation (Excerpt 11.3, lines 2–3). Another common response was a “counter” (Schegloff, 2007), some form of, “How soon can it be ready?” which put the onus of answering the question back on the employee who, as a representative of Eastside, ought to have a better sense for what a reasonable time would be (Excerpts 11.3 and 11.4, line 2). These responses highlight the practical problem: customers do not necessarily have a specific time in mind, moreover, they don’t know how long the job will take, and therefore they don’t know whether the time at which they would like their documents is a time that can be met. Although in most cases, it does not lead to an impasse (although note the tension in Excerpt 11.3 as the customer’s counter is met with yet another counter from the employee, in line 3) it is nevertheless obvious that the formulation of the question consistently gives rise to interactional trouble that alternative formulations would likely avoid. Queuing When we recorded the data we had taken care to capture both the details of the interactions and to get an overview of the store. This allowed us to observe customers while they stood in line, waiting to be served. Given the variations in the stores’ layout, the lines were organized differently at each. However, there were also similarities; all stores had common signage and an open layout with the production personnel working in full view of the customers. We observed that whenever people were waiting in a line and no one was at the counter to help them, they would use a similar technique to get help: they would try to solicit eye contact with an employee behind the counter by looking at them.3 Although not always successful, we noted that whenever a customer was able to secure a reciprocating glance the chance that a production worker would leave their machine to help the customer at the counter increased greatly. Perhaps because of the increased pressure a reciprocated eye gaze engendered, many production employees quite consciously avoided looking at the people in line, something that required considerable effort, for instance always turning away from, not towards, the front counter when moving around.

3

Eye-gaze is also used by patients to get a doctor who is reading the medical record to start the consultation (Heath, 1986).

Interactions at a Reprographics Store

217

Gestures When customers describe their orders, they often use gestures in conjunction with talk. Moore (2008) focused his research on these gestures and their organization in order-taking interactions. He found that when referring to documents in counter interactions, there is a preference for minimization similar to the preference for using a single word descriptor when referring to persons (Sacks and Schegloff, 1979). Excerpt 11.5 is an example: Excerpt 11.5 Coated (Moore, 2008, p. 395). © Cambridge University Press. Reprinted with the permission of Cambridge University Press

01 Cus 02 03 04 EMP 05 CUS 06 07 EMP 08 09 CUS 10 11 EMP 12 13 CUS

Well let’s do eleven by seventeen and then I need four of ‘em? (0.6) [Wha-] [An’I] need ‘em coated. (0.9) Wha’ d’ ya mean by coated. (0.4) You know ya put ‘em between the two plastic (0.1) deals You wan’ it laminated? (0.3) Laminated, thank you

In his initial attempt to describe what he wants, the customer says he wants his copies “coated” (line 5). When this term is not understood by the employee (line 7), the man elaborates with a description produced in conjunction with a gesture (lines 9–10, Figure 11.3): “You know ya put ‘em between the two plastic deals.” Moore (2008) noted that only when customers do not know the name of the services they want to request, or when their description causes interactional trouble, will they resort to the use of gestures. The Order Form We also looked at the relationship between what was said in the order taking interaction, what was written down on the order form, and how production employees made sense of what was marked on the order form (Moore and

218

Making Work Visible

Figure 11.3 “You know you put ‘em between the two plastic deals”

Whalen, in this volume). In particular, we looked at how the standardized order form was often annotated in nonstandard ways to clarify just what the customer wanted. Often, the interaction between the counter worker and the customer was reasonably captured on the order form. However, when it was examined by the production staff that had not been part of the initial conversation, the notes on the order form were often ambiguous and could lead to production errors. Money Talk In some order taking interactions, the employee discusses the cost of the services requested. But in many others, the actual cost of the order is not discussed at all; the price only comes up when the order is picked up. This can have unfortunate consequence as was the case when a customer came in to pick up her order of three custom-made wine labels she intended to put on three bottles of wine to give as a wedding gift – they would help the couple celebrate their 1st, 5th, and 10th anniversary. The labels were on three different files and Eastside charged $10 to print each file. Additionally, the customer asked for the labels to be cut professionally. Eastside charged $1 for each of the 12 cuts so the total cost of the order was well over $50. The woman was outraged and asked to speak to the manager. In the end

Interactions at a Reprographics Store

219

Excerpt 11.6 Good grief

01 EMP 02 03 CUS 04 EMP 05 06 CUS 07 EMP 08 CUS 09 EMP 10 CUS 11 EMP

Ten of this, and ten of this. If you want us to collate it so its ten sets of yeah Everything like that is 10 cents to collate – each page. Mkay, did you want us to staple it? Yes please Staples – I believe – hand stapling is ten cents also. Oh good grief! (lowers head) Then forget it. Okay. Just forget about that and forget about that. Okay

she refused to pay for the order and instead printed the files herself from the computers in the self-service area. The scene could have been avoided had the cost of the order been part of the order-taking interaction. Discussing the price in the order-taking interaction is not without its challenges, either. We examined a selection of order-taking interactions in which the cost of the service was discussed and found that how the employee talks about the cost greatly impacts how the cost is perceived; that is, whether the service is expensive or not. Excerpt 11.6 provides an illustration. When the employee asked whether the customer wanted to collate the copies, she answered affirmatively (lines 1–3). After the customer agreed to the service the employee revealed how much it would cost (line 4). In the next sequence the employee asked whether the customer also wanted her copies stapled (line 5). The customer again accepted the offer (line 6) after which the employee produced the price of the service (line 7). This set off the customer who exclaimed “oh, good grief” and retracted her requests for both services (line 10). Interestingly, in this case the collating and stapling services would only have cost the customer a few dollars. Hence, more than likely, the customer’s response was not the result of a rational analysis of the economics of the service, but instead was a response to the way in which the employee presented the price. Note in that regard that the employee offered the price in an expansion of the base sequence in which she has secured acceptance for the service she offered. Combined with the way she formulates the price as costing ten cents “each page” – thus emphasizing not the total cost but rather the size of the multiplication factor that will be used to calculate the total – this raises the possibility that the customer’s acceptance

220

Making Work Visible

should, perhaps, be reconsidered. When in the next sequence she stresses that there will be additional cost it only reinforces the sense that in her view, these services are expensive. It is not difficult to see that a more favorable outcome for both the customer and Eastside could have been achieved had the employee designed her turns differently. For example, she could have asked: “For a few dollars more we can collate these copies so that they will look just like the originals. Would you like us to do this for you?” which would likely have engendered a different response. Requesting Help in Do-It-Yourself The do-it-yourself area of the shop was designed to enable customers to make copies on their own. One employee was assigned to monitor the do-ityourself area, and the employee’s responsibilities included maintaining the equipment, keeping paper trays stocked, and ringing customers up when they finished making their copies. Despite its designation as a do-it-yourself area many customers needed help either with the machine or simply to figure out how to navigate the do-it-yourself area. The challenge for the doit-yourself employee was to balance how much help they should give to a single customer given the other demands of their job. We looked at some of the strategies that customers used to seek help from employees, and how employees responded to these requests for help. For example, in Excerpt 11.7, an elderly man has come into the do-ityourself area and picked up a copy key. The employee was talking to another customer, and the elderly man waited. When the employee finished he turned towards the man:

Excerpt 11.7 Do-it-yourself

01 EMP 02 CUS 03 04 EMP 05 CUS 06 EMP 07 08

How ya doin’? Pretty good, I just need u:h (0.2)/((CUS looks at documents in hand then down to the left)) need a machine to work on, huh ((scans room)) something Let’s go over here, ((points at nearby machine)) let’s use this machine over here. ((EMP and CUS walk to machine))

Interactions at a Reprographics Store

221

In line 2, in response to the employee’s greeting the customer produced the beginning of a request but then trailed off. This trail-off invited the employee to come in and offer help. The employee finished the customer’s sentence for him: “need a machine to work on,” which makes clear that what the customer himself would be could expect is to be working on a machine; that is, that the customer himself would be doing the work, not the employee. The customer’s response, “something” (line 5) displayed that he was unsure that the employee’s suggestion was actually what he needed and a further indication that he needed help. The employee then offered to accompany the customer to the machine, in effect setting in motion a helping sequence at the machine. Thus, we can see the delicate ways in which the divergent goals of customers and employees were negotiated turn-by-turn in the interactions of the do-it-yourself area.

Participant Observation Our investigations of the interactions between employees and customers frequently pointed to interactional practices that were far from optimal. Instead of going native, a problem quite common in ethnography, we were quite critical of the natives. We wondered whether the analysis of recorded materials had blinded us to the complexity and lived-reality of the work of the employees so that we could no longer appreciate the employees’ point of view. To guard against this, we decided to ask the store managers for the opportunity to experience first-hand what it would be like to work as an Eastside employee. They were delighted with our offer of free labor. Hence, for a few days a week for several months, three researchers in the group worked as regular Eastside employees, at the front counter, in the selfservice area, in digital production. The experience helped us understand the many intricacies of the work and how an individual’s work was an interdependent part of the overall functioning of Eastside as a business. I remember still very vividly a time when the production manager came to look for me in the back of the store with an order envelope in her hand and she asked me if I could recall just how many copies the customer had wanted. As it happened, I had forgotten to ask the customer this crucial question, so the production manager had to call the customer back. I felt profoundly embarrassed to have made such a basic error and the experience gave me a better understanding for why counter workers were always so keen to drive the interaction with the customer from the structure of the order form, sometimes giving that activity precedence over listening to the customer.

222

Making Work Visible

Working in the store also gave us an appreciation of the occasionally antagonistic relationship between employees and customers as at times we felt that customers took advantage of Eastside’s services. Most importantly, however, our experience as employees in the store strengthened our sense that the single most important improvement to Eastside’s business would be to improve the interactions at the counter. If we could improve the interactive skills of the employees at the counter, there would be fewer mistakes made in the orders, hence the redo rate would decrease, and customer satisfaction would increase. With the aid of one of the store managers, we investigated the potential savings that would result from a reduction in the redo rate. The results were encouraging: even a slight reduction in redo orders would save the stores millions of dollars per year. If we could achieve this in one store and then replicate it throughout the stores in the area, we could have significant business impact. And that was quite apart from the improved customer service with all its potential benefits. Thus, we set out to develop a training program, to improve the employee’s interactive skills at the counter.

The Customer Services Skills Set Our training in interaction analysis had taught us that the problems at the counter were not primarily a matter of the employees’ attitude, but rather of their interactional “moves” when dealing with customers. We believed that counter workers could become better order takers if they had better awareness of their own interactional moves and their consequences. We wanted them to become reflective practitioners (Schon, 1995). That is, they should attain such an awareness of their own interactions that immediately following an interaction they could analyze themselves what went amiss, and how, in the future, they could adjust how they responded to the customer in order to make the interaction go better. We located the recurring problem areas in the interactional structure of the order-taking encounter; like many institutional interactions (Drew and Heritage, 1992a), order-taking interactions at Eastside appeared to have a more or less stable structure. We identified the following phases in the order-taking interactions: * * * *

Opening Customer request Requirement elicitation Due-time negotiation

Interactions at a Reprographics Store * *

223

Order confirmation Closing

In all of the phases of the order taking interactions we tried to identify different practices and their consequences. For instance, there were two main ways in which order takers opened the interaction: sometimes they just said “Hi” whereas in other cases they said “How may/can I help you?” We analyzed the subsequent turns of the interaction and looked to see if there were consequential differences in the way customers formulated their next turn (in this case, there were differences but they did not seem to be consequential). For each of the phases in the order taking we identified some recurring troubles and made suggestions as to how they could be handled differently.

The Learning Solutions Design Team With some ideas about how to improve the interactions at the counter, we initiated a codesign phase in which we asked the local store managers if we could recruit some employees who worked at the counter to codevelop a solution for some of the interactional challenges we had identified. They each agreed to let two of their employees participate in what we called a Learning Solutions Design Team (LSDT). Once a week for several months, we would meet with the four members (two from each store) of the LSDT over lunch. In the sessions with the LSDT we presented videos of the interactional troubles we had observed and asked the employees for their opinion, and for possible solutions, that is, what could they say that might alleviate the interactional problems? In between the sessions, the LSDT members tried out the suggestions and reported back on how effective they were in the next session. This usually led to some further refinements of the previous week’s solutions. Thus we made steady progress on the development of the customer services skills set. Unfortunately, in the midst of our work with the LSDT, the project came to an abrupt end. Eastside had recently been acquired by an investment organization and the new owners were wary of the possibility that our work might engender negative exposure for the organization. They asked the store managers to stop working with us. We tried to convince the new owners of the value of our work, but it was in vain; our project did not fit with their priority of cost cutting, which contrasted with our goal to improve quality through employee development. It was disappointing to have to end

224

Making Work Visible

our work on the customer services skills set and not have a chance to see whether our ideas would have the impact on Eastside’s bottom line we anticipated.

Lessons Learned We learned many valuable lessons from the project with Eastside. First, our newly acquired expertise in service encounters and document production centers helped us a great deal in our next engagement with a government print shop in which we pursued similar research. The Eastside project solidified our team’s expertise in analyzing interactions “at the customer front.” Second, we learned how to conduct large-scale video shoots and how to organize a large video collection. Over the course of the project we refined the technological infrastructure needed to store and make accessible nearly a terabyte of video data (a lot at the time). We made conscious choices about how to name and organize files and since everybody stuck to it, this paid off. Third, the experience of having to abandon the project in the midst of developing the solution taught us to be more careful in how we set up our engagements with external customers. The project with Eastside was a great research project but our inability to deliver our solution to Eastside made it harder for us to obtain large-scale long-term project of this kind. As a result, most of our engagements now are shorter and more focused on specific research objectives and deliverables and executed by one or two researchers. The project with Eastside was unique in that it offered the research team an opportunity to do research in a purely data-driven fashion. The project’s legacy is an exceptional data set that we continue to mine for work practice insights.

12

Ethnographically Informed Technology for Remote Help-giving Jacki O’Neill, Peter Tolmie, Stefania Castellani, Antonietta Grasso, and Frederic Roulland

Ethnomethodological ethnographies have played an important role in design since Suchman’s (1987) seminal work revealed that the cognitive models used as a basis for system design failed to take into account situated use and thus could lead to systems behaviour, which was incomprehensible to users. Ethnomethodological ethnographies aim to reveal the situated accomplishment of action. This in turn makes the social organisation of action visible and available to design reasoning (Button, 2000). The idea is to enhance design by enabling designers and ethnographers to explore “the practical implications for design of the incarnate social organisation of human action and how it may be supported, automated, or enhanced by a system” (Crabtree et al., 2009). The exact role to be played by ethnography has been subject to a long, and, at times heated, debate – whether used for advancing the research field of Computer Supported Cooperative Work (CSCW) or requirements engineering (e.g., Viller and Sommerville, 1999). However, there is a strong consensus that ethnographies provide invaluable insights into how the orderliness of work is achieved. As Schmidt put it: “[to] understand how orderliness is accomplished in cooperative endeavors; we need to uncover the practices through which the myriad distributed and yet interdependent activities are meshed, aligned, integrated, because it is the very practices through which such orderliness is accomplished that must be supported” (Schmidt, 2000).

An important use of these ethnographic studies is the critique of existing systems. For example, Whalen and Vinkhuyzen (2001) critiqued an expert system for call operators, demonstrating how the system embodied misconceptions about knowledge and expertise. These included an assumption that expertise could reside within the system, utilised by nonexpert operators, and that the operator–customer interaction would then be an 225

226

Making Work Visible

unproblematic one. Whalen and Vinkhuyzen’s study contrasted this to the common-sense practices used by operators who often had to circumvent the system to do practical troubleshooting. Studies may also be used as rich informational resources for system designers (Martin, 2006). In this conception the ethnographer is generally seen as providing a service; producing rich descriptions and analyses of work that computer scientists wish to support. There has been a wealth of studies in CSCW using this approach (Hughes et al., 1994; Martin and Sommerville, 2004; Button and Sharrock, 1994; Whalen, Whalen and Henderson, 2002). An example of how such studies can influence design comes from our own lab, where developers used an internal report about an ethnographic study of large-scale printing to inspire technology design, including the support of negotiation of print jobs across print shops (Andreoli et al., 2003) and a distributed scheduling system (Castellani and Grasso, 2002). Exactly how studies might be used to inform design remains something of a contentious issue, since they produce detailed descriptions of the work, whilst design requires considerable abstraction. In the literature, the most common link between these studies and design is through “implications for design”: be they bulleted lists of design recommendations or more discursive suggestions. The aim of design implications is to draw out the practical ways in which the rich, detailed fieldwork findings might feed into the design of new technologies or the redesign of existing ones. That most papers now have such design implications might be seen as an attempt to address the disparity between “accounts of sociality generated by field studies and the way information can be of practical use to system developers” (Plowman, Rogers and Ramage, 1995). However there has been criticism of this model (e.g., Schmidt, 2000, Dourish, 2006, 2007), with the basic question being whether ethnographies can be used to directly inform the design of specific CSCW systems for specific settings or whether their contribution to design should be at a more conceptual level, through their ability to reveal how social order is constructed. Schmidt makes a distinction between workplace studies with a view to the design of specific CSCW systems for the same settings (i.e., requirements analysis) versus workplace studies as contributions to the development of the conceptual foundation for CSCW. He further specifies this distinction as being between the application of existing technologies to the requirements of specific settings and the design of novel CSCW technologies. Whilst we agree wholeheartedly that studies should not be made to stand or fall on the presence or absence of design requirements, the focus in this chapter is very much upon how ethnography can serve as a practical inspiration for design.

Ethnographically Informed Technology for Remote Help-giving

227

Ethnography to Design: Our Approach Perhaps because of our setting – design in an industrial research lab – much of our work sits outside of the distinction Schmidt makes between the designers of novel CSCW technologies and the application of existing technologies to specific settings. Instead, innovation arises from studies of specific settings; we design novel CSCW (and other) technologies that are founded upon what they offer for specific settings. Central to this enterprise is an appreciation that it takes a detailed understanding of the social organisation of the work in those settings to provide the kind of inspiration that can support both of these goals. That is, by understanding the methods of social action and interaction in any setting we can conceptualise technologies which fit with those methods. In a design setting, when we understand the underlying methods through which work is achieved, we can abstract away from current technologies and try to project what new technologies (and new assemblies) might mean for the activities undertaken, wherever those methods may be brought to bear. This is not the same thing as requirements analysis as the outcomes are of a different order. Design is undertaken in a multidisciplinary team, with the ethnographers as integral members enabling the ongoing design to be consistently grounded in the field study findings. This stands as a very practical solution to some of the issues raised by Plowman et al. and Schmidt. Furthermore, a key advantage to proceeding in this fashion is that the continual complaint that designs could have been much better realised if only the designers had studied people’s actual working practices before they started making anything, is met head-on through the very constitution of the team. The approach is not without challenge: the ethnographic findings infuse the entire design process on both a practical and conceptual level – requiring designers, developers, and ethnographers to work closely together despite their different orientations, concerns, and perspectives. The upside of this is that through the process of routine everyday working interactions, designers rapidly come to the ethnomethodological orientation and are able to articulate deep insights in proposed design solutions without the inertia that often stalls projects as different groupings try to understand where they are each “coming from.” Furthermore, ethnographers themselves quickly acquire a sensitivity to the ordinary everyday problematics of design and the “grammar” of design practice, so to speak. The biggest challenge, however, is that the timescales of ethnography, design, and development do not sit easily in sync. Ethnography is often carried out at the start of the

228

Making Work Visible

project whereas development may continue for a number of years after. This means that the movement from studies to abstractions to actualizations can require the kind of resourcing that is difficult to obtain in a commercial enterprise. By adopting this approach we are obliged to engage directly with the question of how to move from the studies’ detailed analytic description to the abstraction required for innovative concept design. This makes us all too familiar with what Ackerman calls the social technical gap (2000), that is, that workplace studies reveal how flexible, nuanced, and contextualised human action is, yet current technology is far from being able to support such flexibility and nuance. Compromises are needed according to what the (research) technology can do today and what those working on it might solve in the near future. Thus, whilst the aim is to develop innovative new technologies, this innovation is naturally constrained by what is technologically feasible. Indeed we believe the creativity of design comes from how this balance between what is desired and what is practical is addressed. Certainly it informed the specific case study we are presenting here.

The Ethnographic Study In order to illustrate our method, we describe the Remote Call Assist (RCA) project that set out to improve the troubleshooting of problems on office copier and print devices. This was a multiyear project, motivated by a desire to reduce the costs of device repair whilst at the same time improving customer service. Our goal here is to outline the principal steps taken and demonstrate not only how ethnography can lead to design, but also how constraints can come into play when one begins to move from ethnographic insight towards the creation of prototypes. There are various means for effecting repair on a broken device including, from the most costly to the cheapest: a site visit by a service engineer; expert support in a call centre; and customers engaging directly with an online troubleshooting knowledge base (TKB). Since many problems can be fixed with minor interventions, in theory it should be possible to move support down the chain away from service visits. The aim of this project was to understand the barriers to adoption and use of the two remote support channels (call centre and online) and to develop innovative technologies to address them. Here we focus on a particular part of the overall project: moving from ethnography to the development of research prototypes. This was situated in the earlier of several, interleaved, phases:

Ethnographically Informed Technology for Remote Help-giving

229

Understanding the Current Situation Data from ethnographic study of the call centre was analysed to understand the work involved to make troubleshooting sessions work. This understanding is central to (re)design as it enables the development of loose requirements for both the expert–customer interaction (by highlighting current problems) and the customer-only interaction (by revealing the sort of support customers require from experts, which the online system lacks). Conceptualising Innovative Solutions Based on the problems and opportunities revealed by understanding current practice, a number of solutions to be applied to various aspects of troubleshooting support were conceived. Rounds of Iterative Prototyping and User Testing Putting some of these conceptions into practice was a phased process, with design starting with the supposedly simplest interventions and continuing through to the more radical. For each intervention the work was iterative, consisting of prototype design and testing. The ultimate aim was to develop technologies which could be handedoff to the business groups and put into actual production and use. The Field Study and Setting Two ethnographers conducted a three-week ethnography of Xerox’s European troubleshooting call centre, which provides telephone support across Europe for customers who have problems with their office devices. The ethnography consisted of two consecutive site visits, one by each ethnographer, of one and two weeks respectively. The call centre consisted of a number of large open-plan offices and the troubleshooters were divided into groups, located in physical areas of the office according to the types of machines and countries that they supported. A troubleshooter’s basic setup consisted of a PC equipped with a call-management system, a phone and wireless headset, and various hard and soft copy materials to support their work. In addition, models of all the photocopiers they supported were located around the office. The study involved observing the troubleshooters at work, sitting next to them as they took calls (often listening in on an additional headset) and shadowing them when they moved about the office.

230

Making Work Visible

As well as taking copious field notes and collecting relevant artifacts, we made audio and video recordings. For legal reasons we were only able to record the troubleshooter’s side of the conversation. However, the customer’s side was captured as closely as possible in the field notes. After the field study many of the calls were transcribed. Analytic Method Our multidisciplinary team consists of people with skills in ethnography, design, computer science, and development. In this particular case, after returning from the field there was intense pressure on the team to produce innovative design concepts. As with all aspects of the work, the analytic approach has to be fitted to need. Whilst in an ideal world one might have weeks or even months to work with the data, in this case the approach revolved around a number of team sessions, with ethnographers describing the setting while being probed by other team members. This resulted in a set of eleven classes of findings about troubleshooting work. These include: collaborative constitution of problems and causes; situating instructions; situating the problem in relation to prior events; embodying the solution; effective searching of the TKB; supplementing the TKB; using the TKB in interaction; conferring with colleagues; ensuring accountability; and termination (see O’Neill et al., 2005a; O’Neill et al., 2005b; Crabtree et al., 2006). We used Suchman’s (2007) notion of human–machine configurations as a means of analyzing and structuring our corpus of rich field data to envision future technologies. This enabled us to uncover the critical dislocations between the actors, technologies, knowledge resources, and their environments that served to inspire technology design (Castellani et al., 2009). The Findings We will take two findings of a quite different order and show how they informed design. These findings relate to what we have termed the physical dislocation between the site of the problem and the site of the problem resolution (ibid). The first finding is of a relatively “scenic” (Button, 2000) character and led to some reasonably obvious design conclusions, though creating and implementing an actual solution was not so straightforward. The second finding carries much richer implications and, whilst the core design was not of a radical nature, the ways in which it was designed to support actual interaction was shot through and through with a detailed understanding of how the actual work of troubleshooting is accomplished.

Ethnographically Informed Technology for Remote Help-giving

231

Finding 1: Accessing the Troubleshooting Resources This finding became immediately obvious when listening to the calls: in more than half the cases the phone was not in the same location as the problem device. Customers, therefore, either had to go to-and-fro between the phone and the device or engage another person to relay the troubleshooters’ instructions; sometimes they flatly refused to troubleshoot. Fixed line phones tend to be private devices, located in customers’ offices, whereas office copiers and printers tend to be located in shared spaces such as corridors and many office-based workers do not have work mobile phones. Customers requiring help fixing the device need to be able to move between the help resources and the device relatively easily – so they can carry out inspections, tests, and instructions to first understand the nature of the problem and then to fix it. Customers are often somewhat unwilling participants in the troubleshooting, so it is important to make the troubleshooting as easy as possible for them.

Finding 2: Embodying the Solution The second set of findings also arises because the troubleshooters are remote from the customers and their devices. Analysis of the ethnographic study revealed the work that was done by both the troubleshooter and the customer to make the troubleshooting session work, given the distributed nature of the interaction. The lack of access to the device shaped the troubleshooting work in two crucial ways – (1) effecting the ability of the troubleshooter to situate the instructions for diagnosing and repairing the device, and (2) giving the troubleshooters cause to embody the solutions so that they could give the customer good instructions. Instructions can be situated for the competency of the user and in the ongoing stream of interaction (such that the right instruction is given at the right time and can be carried out without or with only minimal clarification). Whereas troubleshooters could situate the instructions for the exhibited verbal competency of the users, there were often problems situating the instructions in the ongoing interaction, as shown in Excerpt 12.11. In this case the troubleshooter gives detailed instructions for what the customer should look for and do, describing the different doors using indexical referents (left and right) with additional descriptors to try to avoid confusion (e.g., how the door opens (downwards)) (lines 1–6). Despite this, 1

The troubleshooters turn (TS) is transcribed from the recordings. The Customers turn (C) is not verbatim, rather it comes from the field notes. Actions are depicted by . Not all customer turns are recorded.

232

Making Work Visible

Excerpt 12.1 Troubleshooter’s remove help-giving

1 TS Ok it’s probably saying open the upper left hand side door? Probably one of right there’s two doors there that you open there’s the first door that opens downwards and then there’s a 2 door in front of that which is the hot area of the machine so you 3 don’t touch that area and you just need to check that to see if 4 there’s any paper sticking out that you can actually remove just 5 6 to see if if you know you can remove it there 7 {Customer turn not recorded} 8 TS Yeah course no problem take your time 9 <customer goes away> (long wait) <customer returns> 10 TS Hello 11 C Can only find the big door and the little side door 12 TS Yeah yeah yeah you know the when you slide the finisher away from the machine you can open the upper left hand side door 13 that opens downwards? Then just in front of that there’s 14 another door and that’s where the hot area of the machine is so 15 don’t you don’t touch the roller or anything just jus you’re just 16 17 looking for any paper that you’re able to actually 18 C Can’t see any paper and there’s only one door 19 TS No the the with the first bit you’ve got an upper left-hand side door and a lower left-hand side door now the upper left-hand 20 21 side door has two doors that you can open the lower one doesn’t 22 C I’ll go and check

the customer returns, having failed to find the right door (line 11) leading the troubleshooter to try again (lines 12–17) but the customer counters this with “. . . there’s only one door” (line 18). The troubleshooter again reformulates his description (lines 19–22). This time the customer checks and comes back having found the right door and removed the paper. Most confusions such as this are resolved, but they take extra time and effort by both parties. This work arises because the troubleshooter does not know the customers’ orientation nor what they are doing now, nor the state of the machine, except through what the customer tells him and occasional noises he may hear. An initially strange sight in the call centre was seeing the troubleshooters embody the solution, that is, miming actions whilst they were talking to the customer even though the customer could not see them. The reason for this soon became apparent, a photocopier is a three dimensional object and the interaction with

Ethnographically Informed Technology for Remote Help-giving

233

it is, of course, sequential (you need to open the door, before you can pull out the tray – which might involve releasing a catch, etc.). Therefore to give good instructions to the customers, troubleshooters often mimed the instructions whilst talking to customers – that is they embodied the solution. Other methods for envisioning instructions include going to a similar device model (these were positioned around the call centre) to enact the instructions. Of course, because the customer does not have visual access to these resources, the troubleshooter must translate this back into words for the customer and situate these in the ongoing activity. These methods for embodying the solution are done in the abstract – that is, of necessity, they relate only to a generic machine, not the customer’s specific machine, in its current state, that the customer has to interact with. This in turn makes it difficult to situate instructions and resolve misunderstandings.

Using Ethnographic Findings to Inspire Design In this project we were not simply designing technologies to help the people in the environment we had observed (the call centre) work better, we also had a remit to improve self-troubleshooting using online resources. Thus we needed to use our understanding of the assemblies of people, materials, technologies, and spaces to design reconfigurations of those assemblies – where roles, environment, and technologies can change. The team had an understanding of the features of the work of troubleshooters. As can be seen from the previously listed materials, observations of the work of troubleshooters made it clear that a great deal of the work is characterized by what might be termed “dislocation management.” Ways of working and interacting had therefore evolved that amounted to methods for dealing with the “problems” of managing dislocation. The next stage was to engage in a series of brainstorming sessions to conceive of innovative solutions to these various problems we had observed. This involved bringing in the knowledge of the team about the state of current technology research, as well as the state of current practice. We conceived a series of concepts that might be seen as a continuum from improving the online troubleshooting system to facilitating the call centre interaction. The range of solutions are described in Castellani et al. (2009); we will focus here on the design conceptions that came from the findings outlined previously. As might be expected, initial design solutions are often quite basic and simple. That is what makes solutions tractable. The role of deeper insight frequently relates to how to make simple solutions fit to the richness of actual situated practice. It is an understanding of situated practice that facilitates the elaboration of basic ideas for design.

234

Making Work Visible

Self-troubleshooting: Accessing the Resources Customers need to carry out various acts on the problem device. However, at the moment this is hindered by both phones and computers that might be used to access troubleshooting resources often not being collocated with the device. This observation led to us conceiving a simple design solution for self-troubleshooting – bringing the troubleshooting activity into a single physical (and virtual) space: that of the device to be repaired. To this end, we created a design which provided access to the TKB through the device itself. Devices already provide limited troubleshooting support: typically suggesting operations to be performed or providing warnings, when a problem that can be detected by the device arises. These instructions have the advantage of being activated by a known device state, so do not require search. Presented instructions are triggered by user actions, enabling them to be situated in ongoing user activity. However, this support only covers situations where a one-to-one mapping between a sensed device event and a solution is assumed and is therefore limited to just a few of the many possible problems a user might be experiencing. We used a newly implemented embedded UI on the device to act as a web browser that was capable of displaying a version of the online TKB redesigned for the new interaction parameters. This required adapting the pages for the new screen dimensions, for the touch-screen interaction and for interaction at the device. For example, query typing requires a soft keyboard displayed on the screen and when combined with smaller screen size resulted in a design where query-entry and results-viewing were on separate pages. Typing on a touch screen keyboard tends to be slower for most users than on a conventional keyboard as it is often single-finger input. Further, the interaction now takes place standing at a public device, reducing further the time likely to be devoted to the diagnostic work (Martin et al., 1999). These factors signaled the need to redesign the interface and interaction mechanisms for the on-device display of the TKB pages. Putting the resources on the device constitutes conceptually a big step forward in reducing the physical dislocation between the device and the troubleshooting resources. However, this new assembly introduces new expectations around “what the system knows.” For instance, it creates an expectation that the resource is now integral to the device and thus will know the state of the device in ways that were not possible with the online system. As well as expecting that the system would know its own state (what doors were open, for example), users expected their search results to include only those relevant to the particular configuration of their device. Thus by

Ethnographically Informed Technology for Remote Help-giving

235

physically situating the resources on the device, users expected that these resources would be logically connected. This presented us with an interesting design challenge since the support resources are actually located on a web server remote from the device – they only appear to be situated on the device itself. To partially counter this, we used diagnostic analysis and a rule engine to suggest solutions corresponding to detected problems, so that where problems were accurately detected users did not have to search. Expert Troubleshooting: Embodying the Solution For expert-supported troubleshooting on the phone, the customer mediates any troubleshooter interaction with the device and the troubleshooters’ understanding of what is happening is limited to what they can hear over the phone. As we have seen, this causes extra work for both the troubleshooter and the customer in terms of establishing shared referents, resolving misunderstandings, representing actions to be undertaken, and so on. This is a classic CSCW problem: How can we give access to physical objects at a remote site that by their nature cannot be easily shared digitally? The audio-only channel of the phone is limited, so when thinking of design to support troubleshooting with a remote expert, a central concern becomes discovering just what about the physical object and the problem situation might be shared digitally such that articulation work can be reduced. To solve these problems we are developing a prototype that enables the customer to access the call centre directly through the device itself using Voice Over Internet Protocol (VOIP) and a headset. We use a secure data and audio end-to-end connection to the call centre to transfer data about the device – sensor information, historical records, and so forth – to the troubleshooter. This data is then used by the remote server, in combination with its own stored data for that device – for example, records of other troubleshooting sessions – to build an initial representation of the problem. This includes a visualization of the device in its current state shown on both the troubleshooters’ and the customers’ interfaces. All of these elements build upon understandings of actual situated practices elaborated during discussions between the ethnographers and the designers, in the course of both formal meetings and informal interactions. A good deal of the early interaction between troubleshooters and customers is about making the problem visible and sensor data can feed critically into this. Many machines are used by numerous different staff members and the history of any one device, especially from a technical point of view, is rarely available to the user standing at the machine with a problem. Furthermore, customers phoning in have nearly

236

Making Work Visible

Figure 12.1 Troubleshooter and customer viewpoints

always tried a number of things before making a phone call. Historical and sensor data is also able to reveal some of these recent actions. This reduces the need for troubleshooters to ask customers to do things they have already done, a source of frequent complaint. Another significant part of the opening interaction between troubleshooters and customers is the formulation of a problem. Customers’ opening descriptions are shaped around a set of assumptions about what a troubleshooter might need to hear and the limitations of their own knowledge. Troubleshooters then have to work with the customer to arrive at the kind of problem articulation they actually need. Current state representations provide troubleshooters with another resource for understanding the formulation of a problem and provide customers with the grounds for understanding what the troubleshooter is seeing to be relevant about the state of the device. Figure 12.1 shows an example of the two viewpoints for the troubleshooter with the customer’s view in the bottom right-hand side of the current prototype. When the customer performs actions on the device, they are shown on the shared representation, thus the operator can infer, for example, if the customer is following the instructions correctly. The troubleshooters’ representation includes functionality such as (1) the means to view this representation from different spatial perspectives, to facilitate at-a-glance recognition of problems, (2) control buttons, through which they can interact with the

Ethnographically Informed Technology for Remote Help-giving

237

customer, for instance, to indicate a part of the device such as a door, (3) the means to demonstrate actions using visual indicators, for example, LED lights. This system was designed to help troubleshooters to understand the current problem state and to situate their instructions in the ongoing stream of activity as the remote troubleshooter would be able to “see” what the customer had done more or less as it happened. Reciprocal viewpoints are supported and operators and customers will be able to coordinate and coorient around the representation of the object. Although it may seem to be a relatively simple representation, it is actually able to capture salient indexical information, unlike troubleshooters existing representations, which are unconnected to the ailing device. This means that the “just thisness” of the problem can be explored and revealed. All of these features reflect an appreciation of how troubleshooters currently manage dislocation through their interactions with customers. They build upon existing methods – problem formulation, the uncovering of relevant information beyond the current state of the machine, the establishment of shared referents, instructed action, report and repair, and so on – that became visible to ethnographers during their observations of troubleshooter–customer interaction. Furthermore, the elaboration of design around these practices turned upon not just the ethnographers in the team appreciating their methodical character but the designers as well. The capacity to see beyond the scenic features and recognize what is methodical and orderly about a set of practices is hard won. The great strength of a multidisciplinary team is that the designers themselves rapidly come to look for “order in the plenum” (Sacks, 1963).

Conclusion: Informing Design through Ethnographic Study To conclude, we pose some questions about the constitution of successful design and the practical difficulties of realising design in the context of an industrial setting, difficulties to which we feel our approach is uniquely addressed. What Is Design Success? This is a particularly pertinent question in an industrial research lab and it has ramifications for criteria of assessment and success. It is important to point out that what we present here are design conceptions, that is, as researchers our remit is to create innovative technology conceptions, typically developed to prototype, not product, stage. Although there are parallels with the use of ethnography in software implementation, as presented for example in Viller

238

Making Work Visible

and Sommerville (1999), this is a notable difference. We do however use an iterative design approach, moving through several cycles of mock-ups, validation, and design refinements. The design process itself is one of creativity but also of compromise since it cannot be abstracted from the setting in which it occurs. Designs must be innovative advances of the state of the art, as demonstrated by their patentability. They must also address a real business need. However, the process to adoption is a long and uncertain one with many real world and organisational factors determining progress. Although one measure of success is doubtless about usability combined with usefulness, in our situation it is crucially about demonstrating and communicating the strength of concepts that use innovative technology to address real world issues. Measures of validation in an industrial research lab therefore have to include patentability and business validation, as well as usability tests. All of the designs described here have been patented and we have conducted user tests for some and business validation for others. Business validation includes adoption of the concept by business groups, whilst matters such as look and feel (and hence the usability) can change dramatically during implementation. Challenges and Opportunities Much has been written about the use of ethnography in design, with a distinction being made between ethnography for requirements engineering and ethnography to understand the social world and thus develop the conceptual foundations of CSCW. Whichever approach one takes, the in-depth understandings of the “constitutive practices of how people do what they do, the ‘interactional what’ of their activities” (Button, 2000) are essential. That is, whilst not saying that ethnomethodological ethnography is the only ethnographic approach that could work, to really inform design, any ethnographic approach needs to provide deeper than surface analysis of what happens on site. The case we have presented here demonstrates how basic designs arising from initially scenic descriptions acquire their true viability by moving on from the scenic elements to an understanding of the orderly features of practice. In the end, we do not see the dichotomy between foundational work and the design of technologies for specific settings as being quite as clear cut as it was portrayed by Schmidt (2000). Certainly in our design setting we are designing innovative technologies for specific settings and whilst we use all and any findings that help us to do this, including scenic features. Our ultimate aim in doing ethnography is to outline “how orderliness is accomplished in cooperative endeavors” (Schmidt, 2000). In our case, ethnography informs design on both a very practical and on a conceptual level and this is

Ethnographically Informed Technology for Remote Help-giving

239

further facilitated through team collaboration of designers, developers, and ethnographers. The ethnography does not produce anything resembling direct design requirements. Instead the intense interaction between ethnographers, designers, and developers produces a set of principles to which we try to remain true during the design – but which could likely be addressed in any number of ways. On the practical level, the design is grounded in realworld “problems” and on the conceptual level, the deep understanding of practical action in a setting helps to colour and inspire the design. In this case for example, it was the observed dislocations between actors, technologies, and expertise that were the impetus for designs which could attempt to ameliorate them. However, the particularities of the choices we made in how to develop these understandings into design concepts and then into prototypes were shaped by a mixture of practical constraints (e.g., the limitations of current research technology), opportunity, and creativity. A key point is it takes skill to recognize the best opportunities for innovation and choose which findings to develop, whilst simultaneously honouring the orientations and orderly practices of those to whom the innovation must ultimately be pitched. At the same time design is not only a creative process but also a product of its situated practice. Certain settings can promote creativity, whilst others constrain it. Working in an industrial design lab often means that it is prevailing economic considerations that will serve to promote or constrain innovation in various ways. An important part of the calculus of an integrated team is therefore the working through of this problematic as well. Often the priorities of teams dedicated to design and teams dedicated to studies of real world practice can be divergent under the pressures of expediency. Orientations towards what one thinks one’s business-motivated overseers are going to want to hear you are going to build and towards what one thinks they might want you to go out and study is not at all the same thing. Working as an integrated team means you must ultimately arrive at a univocal position here. One of the most significant values of our integrated approach over the years has been that this univocality has to, by need, become centered upon design growing out of a proper understanding of the real-world character of working practice.

13

Sign of the Times at the Department Store Replacing Paper with Electronic Signs Johannes A. Koomen1

Towards the end of the 1990s, Xerox Palo Alto Research Center (PARC) spun out an invention called Gyricon in the hopes of commercializing it. Gyricon,2 a kind of electronic reusable paper, is a static display consisting of thousands of tiny magnetic balls suspended in a thin layer of oil sandwiched between two layers of a clear substrate containing a thin wire grid (see Figure 13.1). The balls are half white and half black, and their orientation can be changed by a magnetic field induced by applying a small electric current to the wire grid. One of the major benefits of the invention is that, once changed, the balls remain in the same orientation and hence the display maintains the same image without needing continuous power. The new company, named Gyricon Media Inc., was to produce signs in various shapes and sizes, targeting conference venues, hotels, and other public places where information such as event announcements and directions can vary from day to day. Business Services, exploring possibilities for creating solutions and services around the new display technology, approached a major department store chain with the objective to replace the traditional paper signs in the stores with Gyricon signs. The department store chain consisted of a flagship store close to company headquarters, and scores of large-, medium-, and low-volume stores distributed throughout the Eastern United States. Company headquarters, in addition to the usual business staff, also contained well over a hundred Buyers offices, typically staffed by a Buyer and several Buyer Assistants. 1

2

240

Although she never had the chance to go on location, Ann Stimson was an integral part of the Work Practice team for this project; I gratefully acknowledge her significant contributions. Also, Dena Germano and John Faklaris of the Services Group provided much useful input and feedback on our study findings. For more information about the technology, see Electronic Reusable Paper (Gyricon) at PARC: http://www2.parc.com/hsl/projects/gyricon/ and Gyricon LLC in the news: http://www.prnewswire.com/mnr/gyricon/11646/.

Sign of the Times at the Department Store

241

Figure 13.1 Smart paper sign. © Ken Buschner

The initial phase of this project included a feasibility study undertaken by a Lean Six Sigma (LSS) team. This team concentrated primarily on the processes around the creation, production, and distribution of signs, and the various quantities of signs in different departments. Although we assisted in the definition of these processes through interviews of key personnel in the Marketing, Creative, and Printing divisions at the company’s headquarters, our Work Practice team was invited to participate specifically to study the existing practices and processes around the placement of signs on the store floors. To do this we performed three field studies at different department stores, investigating the practices of a special team of Sign Associates, whose entire job involves changing the thousands of signs in each store (tens of

242

Making Work Visible

thousands in some of the larger stores) in response to marketing changes, sales events, seasonal changes in merchandise, and so on.

Methodology The LSS team interviewed various department managers and staff at company headquarters who were involved in the creation and distribution of merchandise signs. We shadowed the LSS team on many of these interviews, videotaping the interactions. This provided us with a lot of useful background information. Since our primary focus was the in-store work practices around the placement of merchandise signs, we made four field visits to the flagship store and two other large stores in the same urban region between October 2003 and February of 2004. We shadowed about a dozen employees responsible for sign placement during their entire shift, some from 6 am to 11 am, some from 5 am to 2 pm. We collected 39 hours of video of our observations and interviews. We analyzed our notes and the videotapes, creating representations of the workflows, (Brun-Cottan and Wall, 1995; Wall and Mosher, 1994) and documenting our findings and recommendations in a PowerPoint presentation, which we delivered to both our business group and the department store’s executive management. To assure better coverage of the work practices and to uncover possible regional or size-related differences in practices, we proposed making field visits to a small volume store and a store in a different part of the country. Unfortunately, the project’s budget and timeline could not support this.

The Life of a Sign A fair amount of activity precedes the creation of a sign, namely the planning of events and the design of department- and event-specific sign templates. Specific signs are electronically created and distributed to the stores, where they are printed and displayed. Paper signs for special events are destroyed after the event; regular paper signs are disposed of at the end of a season. Pre-sign Planning and Design The stores cycle through two major seasons, summer and winter, and the whole sign process follows this cycle. Six months in advance of each season, the Advertising Committee creates an advertising calendar for the season, marking special as well as regular sales events for various departments. Merchandise Managers and Buyers strategize about their specific business

Sign of the Times at the Department Store

243

and determine the merchandise for various sales events, recording their purchasing and pricing strategy in a buyer-specific strategy document. The Merchandise Information Office uses the advertising calendar to create a monthly Price Change Calendar, marking deadlines for price changes and sign creation. The Price Change Calendar is typically created with a two to three month lead time. The Visual Department determines the look and feel of the stores, and approves sign templates created by the Sign Shop. The Sign Shop creates and manages sign templates using a proprietary software system called FSign. The FSign system contains the entire corporate sign library. The Sign Shop makes approved sign templates available via FSign to the Buyer staff for sign creation. Some sign templates may be carried over from previous seasons. Sign Creation The staff at each Buyer’s office uses approved FSign templates to create and store merchandise signs. Using the Buyer strategy document and the Price Change Calendar as guides, a Master Sign File is created for every merchandise item prior to the start of each season. The Master Sign File contains a regular price sign and a standard sale sign for the item. During the season, additional signs are created as needed; new Master Sign Files are created for new merchandise, and promotional signs are created for special sale events. Existing signs may also need to be altered to reflect changes in the Buyer’s strategy document. Each sign has a unique number, which appears at the bottom of the printed sign as a barcode. Prices are copied from the strategy document and pasted into the signs, as well as into the FPrice application and the Up/Down Report. The FPrice system drives the Point-of-Sale terminals where customers buy items; the Up/Down Report documents the dates on which the price of an item changes, and therefore when the signs associated with that item, identified by its number, must be put up or taken down. The staff notifies the Buyer for approval whenever price changes have been made. They also notify the Sign Shop whenever signs have been created or changed. Approval and Distribution When the Sign Shop receives a notification from a Buyer office that signs have been created or modified, the Sign Shop prints a copy of each sign to verify that the sign information fits the template properly and looks good. If they are approved, the signs are placed in a queue for electronic distribution

244

Making Work Visible

to the stores. If they are not approved, they are sent back to the appropriate Buyer staff for correction. Every weekday, starting at 5 pm, the queued signs are sent to each store in a Blast, which updates each store’s local copy of the FSign system. A blast can contain 400 or more signs. During the season a blast typically contains about 60% promotional and 40% Master File signs. It also contains, for each sign, the number of copies to print. This number is preset and is based on sales volume and store size. In addition to the blast, a Sign Blast Report is also sent to each store. This report lists all signs included in the blast, grouped by Family of Business (such as Men’s Apparel). For each sign the report shows the sign number, the heading and subheading, the regular price and the sale price, the date for the sign to be put up, and the date for the sign to be taken down. Print and Display Each store has a dedicated staff for dealing with merchandise signs, consisting of a Sign Manager and one or more Sign Associates. They typically start at 5 or 6 o’clock in the morning, and are expected to complete the work of changing signs in the store by opening time (10 am). They leave by 11 am, except for the manager who leaves at 2 pm. Each store also has a Sign Room, where supplies are kept. The Sign Room is equipped with a PC and a monochrome laser printer. Each morning the manager logs on to the PC, enters the FSign system, selects the most recent blast, and prints it on standard letter size paper. The FSign system prints the appropriate number of copies of each sign in the blast. The Sign team then sorts the printed signs by floor and department, and places them on dedicated sign carts (see Figure 13.2). These carts, typically one per floor, hold storage containers with manila folders for signs, as well as supplies such as sign holders, scissors, and generic sale and percentage-off sign banners. The team then disperses over the store, taking their carts, usually two or three associates per floor, and sets about displaying the signs. Signs are displayed either in letter-sized metal sign holders placed near the merchandise, or cut and folded to fit in small plastic sleeves (“hang tags”) that are attached to individual items with a plastic wire tie. Although cross-trained, the Sign Associates typically work in their “own” areas of the store. However, the associates we observed displayed a strong sense of teamwork and common goals; Sign Associates whose departments are light in terms of sign changes on any given day help out their colleagues in other departments as needed until the whole job is done.

Sign of the Times at the Department Store

245

Figure 13.2 Sign cart

Disposal After signing is completed for the day, Sign Associates organize and replenish the carts. Unused signs for the current event are discarded. At the end of the season, the stores destroy paper signs for seasonal merchandise; the Sign Shop either destroys electronic signs or prepares them for reuse the following season, per Buyer staff request.

Detailed View of In-store Signage Practices The official task for the Sign Associates is to replace the signs associated with certain merchandise prior to store opening time, coinciding with the beginning (sign-up) or the end (sign-down) of a sale event. Of course, almost all sign-up activities of new signs imply a sign-down activity of a corresponding old sign being replaced. The Sign Associates receive the signs to be placed in a blast, and to guide them through their task they also receive the Blast Report, the Up/Down Report, and the Price Grid (a document similar to the Up/Down Report but specific to merchandise in the Home Department). In practice, things are not quite that straight-forward. The Blast Report is not really useful to the Sign Associates because the signs included in a blast are the signs that were just created by a Buyer Office and approved by the

246

Making Work Visible

Sign Shop, and they may or may not be placeable the day they were received in the store. Although many of them end up getting placed the same day, some must not go up until one or more days later; the Sign Associates may place these signs in a storage bin on their cart, or may discard them. So some of the signs to be placed on any given day are in that day’s blast, but some of them were blasted earlier and may be in the storage bin. Although the Up/Down report is the official and seemingly obvious way to steer the Sign Associates through the store replacing signs, it is not a very efficient way of accomplishing the task: going through the signs one by one on the Up/Down report means constantly running through the department to locate the merchandise, because the merchandise location is only coarsely related to the listing order in the Up/Down Report. This leads to a lot of wasted time and effort, which the Sign Associates cannot afford as on many days they may have to replace hundreds or even thousands of signs. Then there is the issue of price consistency and integrity. The Sign Associates team prides itself on making the shopper’s experience as smooth as possible, when it comes to pricing. The price for an item is obviously available in the FPrice Point-Of-Sale (POS) system, which is the price that comes up at the cashier’s register; but it may also be on a sign by the merchandise, or in a newspaper ad, or in a promotional flyer. Ideally, these prices are all the same. In reality, the Sign Associates regularly find discrepancies, and resolving them is part of their job. Since customers should be able to purchase an item for the price found in advertising, the Sign Associates will alert the Sign Shop and/or the appropriate Buyer’s Office if the POS price or the sign price is different than the advertised price. The fact that Sign Associates have their “own” departments greatly helps in this price integrity checking: if you know your merchandise, it is easier to remember the specials, and it is easier to spot “obvious” errors, such as a leather jacket with a sign that has a price of $39.50 instead of $395.00. So the Sign Associates typically start out their day by scanning newspaper ads and promotional flyers to cue themselves for items in their department(s). Then they go through their entire department, checking every single merchandise item. Using a Radio Frequency (RF) Gun, they scan the bar code on each item to obtain the POS price (see Figure 13.3). If the POS price is different than the price on the current sign, the current sign is replaced with a new sign. Hopefully, the new sign is in the current blast or in the storage bin, and the Sign Associate places it in the sign holder. The old sign is either removed from the holder and discarded, or placed in the storage bin, or left in the holder behind the new sign (provided the old sign is the regular price and the new sign is the sale price). If the item is going

Sign of the Times at the Department Store

247

Figure 13.3 Scanning an item

on sale, the Sign Associate may also mark it as a “special” by placing an event banner (“topper”) on top of the sign, such as “One-Day Sale,” “Valentine’s Sale,” and so forth. Some Sign Associates were observed to complete each fixture before going on to the next; others scan several items first, and then return to the cart to collect the needed signs and place them. If a needed sign for an item is not available, or it is but it has the wrong price, the Sign Associate makes a note of it, recording the merchandise name, sign number, regular price, and sale price on a notepad. The Sign Associate also leaves a physical “breadcrumb” on the item by placing the current sign backwards or upside-down in the sign holder, or placing the topper sideways, so that it’s easy to locate the item again once the Sign Associate has obtained the correct sign. After completing a section, an area or several rows, some Sign Associates will notify their manager that they need particular signs, and some will go back to the store’s sign room to access the FSign application themselves and find the appropriate signs (which may require editing to change the price before printing it). Six Ways of Dealing with a Missing Sign We observed six different ways Sign Associates obtained a needed sign that was not in the day’s blast or in a storage bin on the cart.

248

Making Work Visible

1. If the Sign Associate knows the sign number, she (or her manager) can access the FSign system on the PC in the store’s sign room, enter the sign number, and print the sign. 2. If the Sign Associate does not know the sign number, she (or her manager) can access the FSign system on the PC in the store’s sign room, search for the appropriate sign by product type and/or brand, and print the sign. Often a sign exists for the merchandise but has the wrong price, so the Sign Associate modifies the price before printing it. 3. Sign Associates can reprint signs using the RF Gun by reading the barcode from an old sign or from the Up/Down report on the selling floor and clicking the Print button. This is rarely used, however, because of the physical distance between the merchandise and the sign office where the printer is (often on a different floor). 4. When the FSign system was down, one Sign Associate was observed creating signs in MS Word using a template stored on the PC in the store’s sign room. 5. If the item is on sale and the sale price, say $69.99, is an “obvious” percentage of the regular price, say $99.99, Sign Associates sometimes decide to replace a price point sign with a generic percentage off sign, computing the percentage themselves. 6. Sign Associates sometimes decide to replace a product-specific price point sign with a generic price point sign, that is, a sign that shows the price but has no merchandise information. Murphy’s Law Applies to Signing Also During our field visit to one of the department stores we witnessed the following sequence of events. Although not typical, it does illustrate the various challenges encountered by the Sign Associates, and the resourcefulness they bring to their job to complete their mission. The Sign Associate – let’s call her Sally – has set her RF gun in promotional pricing mode since there is a one-day sale event that day. Working in the Home Department, Sally notices an item that does not have a price attached: the hang tag holder is missing. So she scans the item with the RF gun in order to get the needed sign’s number and the item’s original price and the sale price. However, in order to conserve energy (and hence, costs) the department store has reduced lighting in the early hours of the morning before the store opens, which makes it too dark in many places to read the gun’s display. So she walks over to a better lit area and makes a note of the sign information.

Sign of the Times at the Department Store

249

Figure 13.4 Cutting, folding, and displaying sign in hang tag

Sally returns to her cart to look through the storage bins for the needed sign, and fails to find it. She continues working on other sign tasks, but at the next opportunity she tells her manager, Mary, about the needed sign. Mary goes down to the Sign Room, opens the FSign application, selects the sign by number, and notices that the price is wrong. She closes the sign, opens the Modify Price tool, edits the price, saves the modified sign, and prints it. When she returns to the Home Department’s floor she hands the new sign to Sally. Sally takes the sign over to her cart, and finds a plastic hang tag and a “One Day Sale” topper. The sign is printed on letter-size paper so it needs to be cut and folded in order to fit into the hang tag. After rummaging through her cart she eventually concludes that the scissors are missing, and walks over to a nearby sales counter to find one. Sally cuts and folds the sign, and sticks it with the topper into the hang tag (See Figure 13.4). Returning with the completed hang tag to the item, she scans it again in order to verify that the price on the tag is correct. She walks over to the light and finds that the gun has frozen up. She reboots it, walks back to the item, scans it again, walks to the light again and verifies that the price on the tag is indeed the right one. Now she needs to attach the hang tag to the item. The item does not have any convenient protrusions to tie the hang tag to, and is itself too large for the wire tie. So Sally returns to the cart once more to retrieve a second wire tie, connects the two ties, and attaches the hang tag to the item. Job done. Next . . .

250

Making Work Visible

Findings, Challenges and Recommendations We collected a sizable list of specific observations and opportunities from the entire signage process which we shared with our business-services organization and the department store company’s management. We provide here our most significant findings. Challenges of the Current Signage Practices Pre-sign Planning and Design Personnel issues and time requirements of the current practices and processes limit the frequency of price changes. As one store manager told us, “The frequency of price changes is one of the beauties of retailing – reacting to the needs of the business quickly.” Sign Creation From the staff at the Buyer’s Offices and several other offices we heard that numerous format restrictions, line limitations, message placement requirements, and so forth, unduly constrain the design of signs. The Buyer staff performs a lot of reentry of data or copy-and-paste because the many systems they have to work with are not integrated. They often encounter computer crashes caused by opening multiple programs simultaneously. The Buyer staff finds that dealing with the Sign Shop is inconsistent. Sometimes they call when there are issues, and sometimes they do not. A miscommunication required the Buyer staff to recreate a hundred signs. The Buyers also noted that separate signs are required for Puerto Rico stores. However, they are not able to differentiate between any other geographical areas for sale purposes. The Blast From our observations and interviews of the Sign Associates we found that there is a disconnect between the signs needed for a day’s event and the sign blast: *

*

Not every sign that is blasted today is needed today – the extras get filed on the cart for later use, or discarded (making them missing in the future). Not every sign that is needed today is blasted today – some signs are in the storage bins on the cart (from a previous blast or the season’s start).

Sign of the Times at the Department Store *

*

251

Some signs are missing because the appropriate Buyer’s Office did not create them. The number of copies to print for each sign is hardwired in a blast; the number is specific to each store, but is not sensitive to that store’s inventory – printing the signs in a blast may give more or fewer signs than are needed.

The Environment Signers have only one printer available per store, located in the sign room, leading to contention and loss of time. To conserve energy and reduce costs, the escalators are turned off while the store is closed, that is, during most of the working hours of the Sign Associates. The consequence is that, in order to get to the sign room, the Sign Associates have to either walk up and down several floors worth of escalators, or take the elevator. This makes a missing sign a major issue. Also, lighting is reduced while the store is closed, making many areas too dark for signing, especially reading the RF gun display. Signs are sometimes located in places where they are difficult to access. We observed one Sign Associate spend 15 minutes hunting down a stepladder so she could change a sign on top of an eight foot display case. The Supplies There is only one sign cart on most floors which the Sign Associates on a floor must share, causing them to spend time going back and forth to the cart. They also encounter occasional shortages of supplies for signing – working RF Guns and sign holders were mentioned. The Signs Signs are printed on standard 8.5” × 11” paper, but many signs are placed in much smaller plastic sign tags attached to merchandise. This leads to time consuming cutting and folding to make the sign fit the tag. The quality of the signs is not very high (monochrome, single font). This is exacerbated by a “used” look from repeated handling as many signs (regular price and common sale price) are reused. The Scope For some sales events, the number of signs that require changing is so enormous (literally thousands in large stores), that it is impossible to

252

Making Work Visible

accomplish the task in the normal Sign Associates’ working hours, between 5 am and 10 am, when the store opens. In those cases the Sign Associates manager may decide to start changing signs the day before, or continue changing signs well into store opening times, sometimes as late as 12 or 1 pm. The first option, where an item looks like it is on sale but it is not yet, leads to consumer confusion and price overrides at the registers; the second option, where an item does not look like it is on sale even though it is, may also lead to consumer confusion and loss of sales. The Job When an RF gun fails or when the FPrice system is down, Sign Associates must rely totally on the Price Grid or Up/Down Reports, resulting in significant signing delays. Sign Associates are not always notified of new merchandise on the floor. Too many items on one display can look messy. The Home Department is predominately “no mark” (i.e., most items don’t have price stickers on them) and needs signs for everything. Customers often must go to a register to find a price. When merchandise is moved by other store personnel, signs are often moved incorrectly. Some manufacturers place a suggested retail price on their items which can be different from the company’s regular price. Improving Signage Practices In order to improve the job of the Sign Associates, in terms of quality, ease and efficiency, we made the following recommendations: * *

* *

*

Do not use blasts for printing. Allow easy printing of all signs within an area, department or Family of Business that must be changed on a given day. Provide access to the FSign system (PC’s and printers) on all floors. Add a new Print-from-Gun feature (“print the sign for this item”), that is, determine and print a sign based on an item’s Universal Product Code (UPC), rather than based on a sign number. Provided wireless infrastructure is available, add a printer to each cart, replacing paper sign storage bins on the cart. Printing on demand and on location is much more efficient than printing, handling, and filing extra copies, and searching files for a specific copy later.

We also sketched out how electronic signs could be integrated in the company’s current sign practices, and indicated some benefits and drawbacks of using electronic signs.

Sign of the Times at the Department Store

253

Conclusion As we encounter so often in Work Practice studies, simple jobs turn out to be far more involved and valuable than meets the eye of upper management. The work of the Sign Associates encompasses significantly more than just changing signs. Although many of the complexities and challenges of their jobs stem from limitations imposed by their environment (lack of light, supplies, sign holder shapes, remoteness of printer, etc.) and would be eliminated by switching to an electronic signage system, they provide important additional services that are very valuable to the corporation. They are instrumental in maintaining consistency in pricing between newspaper ads, sales leaflets, in-store signage, and point-of-sale terminals. They keep an eye on inventory, and introduce or remove signs as appropriate. They catch and fix occasional errors in pricing because of their knowledge of the merchandise. They resolve problems or conflicts using their extensive experience. And they use their sense of esthetics as well as “common” sense in deciding how and where to place signs to maximize value to the customer. There were a number of technical and financial reasons, such as limited image quality and unit cost, for the company to make the decision not to go forward (yet) with the introduction of electric paper signs, but the results of our work practice study significantly contributed to this decision. Because our study uncovered and exposed the hidden aspects of the Sign Associates’ job, management realized how much value these employees were bringing to the company above and beyond their official task of changing paper signs.

Part V

Learning and Knowledge Sharing

14

Communal Knowledge Sharing The Eureka Story Jack Whalen and Daniel G. Bobrow

It is widely recognized that an organization’s most valuable knowledge, its essential intellectual capital, is not limited to information contained in official document repositories and databases – scientific formulae, “hard” research data, computer code, codified procedures, financial figures, customer records, and so forth – but also includes the mainly undocumented ideas, insights, and practical know-how of its members (Nonaka and Takeuchi 1995; Stewart 1997; Davenport and Prusak 1997; Brown and Duguid 2000; Wenger, McDermott and Snyder 2002; Whalen and Whalen, this volume). This more tacit or informal knowledge is deeply rooted in the experiences of individuals and the culture of their work communities. It commonly originates as practical solutions – through everyday inventions and discoveries – to the exigencies of work, and thus serves as the critical resource for ordinary work practice. Much of this knowledge often remains embodied in practice. Crucial steps in a new practice and fresh solutions to recurrent conundrum are commonly shared through conversations and stories among small circles of colleagues and work groups, with members filling in the blanks from their own experience. These instructions and stories are expressed in the local community vernacular. The challenge for organizations is to somehow convert this valuable but largely local knowledge into forms that other members of the organization can understand and, perhaps most important, act upon. This chapter tells the story of an effort in one organization at encouraging inventiveness and capturing new ideas generated by people working on their company’s front lines, and at using technology to then share the best of this knowledge beyond the local work group. The story details our experiences over more than fifteen years with the design, development, deployment, and evaluation of the Eureka system in Xerox Corporation. Eureka is being used in Xerox to support the customer 257

258

Making Work Visible

service engineers (CSEs) who repair the copiers and printers installed at customer sites. Several design iterations took it from an experiment initiated by researchers at Xerox’s Palo Alto Research Center (PARC) to measure the value of codified field experience, to a system that is now deployed worldwide. By focusing on communities, and how they share knowledge in ordinary practice, we developed a set of questions and a methodology that we hope will enable others to undertake similar communal knowledgesharing efforts. However, deploying any communal-knowledge system necessarily involves pushing changes within a corporate culture and dealing with that culture’s often antagonistic “immune system response”; understanding the Eureka experience, as well as the problems facing all knowledge systems that have to be deployed in the real world, requires recognition of these challenges. Although our analysis recounts the story of Eureka in Xerox, it has a highly general theme: large organizations face certain knowledge problems when established doctrine – well-known policies, procedures, beliefs – prove inadequate or even erroneous as guides to effective action, and traditional organizational methods for generating new knowledge and managing its distribution plainly won’t suffice. In this kind of not uncommon situation, supporting the creativity and inventiveness of those members of the organization closest to the problem, warranting the insights, and then sharing those new solutions widely can make all the difference between inspired success and abject failure. This is exactly the sort of inventiveness and sharing that Eureka was designed to enable.

Breaking the Frame Helping the Xerox Field Service Force Xerox at one time had nearly 25,000 technicians worldwide to help ensure that the machines they sell to their customers are performing as expected; today the number is closer to 10,000, but the nature of their work has remained essentially the same: as Orr (1996) has pointed out, this work involves a triangular relationship among the technician, customer, and the machine. On many service calls, the machine needs to be repaired and/or adjusted; on some, it is the customer who needs adjustment – in their expectations, procedures, and/or knowledge of the machine. In the early 1980s, with the decreased availability of technicians trained by the armed services to debug complex equipment, Xerox decided to make it possible to use less-skilled service people. They did this by moving away from

Communal Knowledge Sharing

259

documentation and training that simply described the principles of operation of a product, leaving it to the technicians to determine the appropriate repairs. They moved toward what was called “directive” repair and adjustment procedures. These are documented instructions in the form of a decision tree. Each decision step was of the form “do the following setup and test; make the following measurement (or observation); if the result is A, do X; else do Y.” The intuition embodied in this form of documentation is that technicians need only be trained how to correctly use the documentation to diagnose any machine failure. Our group at PARC had a background in Artificial Intelligence, and in modeling electromechanical systems. In particular, we had expertise in building programs that perform efficient diagnoses of machine faults given an abnormal symptom and the ability to obtain observations/measurements from the machine (de Kleer and Williams, 1987). As a test of their technology, we decided to build a model of one complex module of a particular photocopier, and demonstrate how a program could provide guidance for a technician in its diagnosis and repair. The hypothesis was that if we were successful, the model-based expert system running on a laptop carried by technicians in the field could replace the documentation and support an optimized work process for isolating faults. Our group succeeded in building Rapper (Bell et al., 1991), a modelbased expert system that used a model of the recirculating document handler to provide optimized guidance in isolation of faults in that module. The model was constructed with the help of Xerox engineers and technicians; it captured the behavior of all the faults that could be found using the standard documentation. We showed it to some technicians who said, “That’s amazing.” “Would it really be useful if we had a complete model for the machine?” we asked proudly. “Not really – though it is amazing – rather like a bear dancing. It is surprising to see it do it at all.” Our group probed further for the issues that underlay the technicians’ negative response. The issues turned out to be rooted in the everyday exigencies of work for technicians and the practices they had developed in response. First, small optimizations of the time required to isolate any known fault were not worth much. Only a small portion of an average twohour service call was actually devoted to diagnosis. Second, technicians usually knew the procedures for the common faults and so required no guidance. The thorniest issues, though, were more surprising. For many products (those produced by Xerox’s Japanese partners) there were no full descriptions of their operation. Additionally, the diagnostic documents were produced from experience in inserting faults into the machines in a

260

Making Work Visible

laboratory and then recording the symptoms. But the hardest problems in the field for all machines were not those covered by the documentation – they were new problems. There are a number of reasons why new, unexpected problems occur in the field rather than in laboratory tests. Xerographic machines work differently in particular environments – for instance, extremes of temperature, humidity, and dirt. As machines age, components develop new failure modes; and when a machine comes on the market, there may be new components with unexpected interactions. With vibrations and other slow acting processes, faults develop that are intermittent, and these are the hardest to diagnose. These difficult cases caused most of the long calls, resulting in average call time increases. Accordingly, we decided to observe what technicians actually did in their day-to-day practice. We started with technicians in the United States, accompanying them on their service calls. Most of the time, technicians would approach a machine, talk to the customer, and know exactly what to do to return the machine to good working order. Once in awhile, they ran into a problem that they hadn’t seen before and for which there was no answer in the documentation. They would try to solve these problems based on their knowledge of the machine. This often worked, but sometimes they were stuck. They might call a buddy for ideas, using their radios, or turn to the experts – former technicians now serving as field engineers – who were part of the escalation process. When unusual problems were solved, they would often tell the stories about these successes at meetings with their coworkers. The stories, now in the possession of the community, could then be used in similar gatherings and further modified or elaborated (Orr, 1996; Brown and Duguid, 1991). This practice pointed to the importance of noncanonical knowledge generated and shared within the service community. It suggested to us that we basically could stand Artificial Intelligence on its head, with the work community itself becoming a “living expert system” and with ideas flowing upward from the people actually engaged in work on the organization’s frontlines (cf. Doubler, 1994; quoted in Ambrose, 1997). The Colombus Experiment In 1995, a member of our group at PARC, a French national, began to spend time riding with French technicians to understand if their practices were similar to those in the United States. According to local doctrine, quality service meant uniform service, and uniform service meant following

Communal Knowledge Sharing

261

the instructions in the manual. Technicians would show this visitor their pristine service manuals. When it became known that he was truly a researcher, however, and interested in what they actually did in the field, they showed him their “second set of books” – marked up with their annotations on the recommended processes, and with clever solutions they had discovered (Bell et al., 1997). Many technicians, particularly those with the most experience, also had cheat sheets that captured their invented solutions to undocumented problems. Technicians who were starting on a new machine often asked more experienced technicians for copies of these cheat sheets. In a series of workshops in France, we asked technicians whether they thought they had knowledge that was valuable to share beyond their workgroup. They were not sure, though they shared some stories about difficult “problem machines” and how they repaired them. The responses of others in the workshop were comments like, “Knowing that, I could have saved five hours last week.” Our team asked the groups whether they would be willing to share these solutions more widely. Would a group that shared its hard-won knowledge lose their performance advantage in benchmark comparisons to other groups? And would it be worth the time and effort to document this local knowledge, so it could become portable, could travel beyond the confines of the local work group? Our team believed that this knowledge could have significant value, and received the backing of the French service organization – including management and the “tigers,” which were expert field engineers and played a key role in the escalation process – to try an experiment. This required three things: an initial knowledge base of tips; an easy-to-use vehicle for distributing this knowledge so technicians would turn to it; and an experimental design that would let us conduct a valid test. We elected to focus on a single product, the Xerox 5090, a high-speed complex copier that had been in the field for a couple of years. Having the tigers edit and validate tips that had been volunteered at the workshops, adding more tips that the tigers themselves used, developed the initial case base. There were between 100 and 200 tips in all, structured in terms of Symptom, Cause, Test, and Action. A machine-presentable version of the standard 5090 documentation was also made available. The vehicle for delivery was a standard laptop, which was just starting to be deployed in the United States but not scheduled for deployment in Europe for at least another year. Engineers on our team wrote Colombus (the French spelling of Columbus), a software package designed for ease of use by the technicians. When the laptop was turned on, the application started

262

Making Work Visible

automatically, so that Windows was completely hidden; they only had to learn this simpler, dedicated software. A single search by presenting symptom (either the numeric fault code indicated on the 5090 console, or a descriptive term like “copy quality”) would bring up, on an integrated dashboard interface, access to any material concerning this symptom, both from the tips database and the standard documentation. Our team took a prototype of this software to the tigers in France. There followed an intensive week of discussion and testing with the tigers, with quick implementation of suggested improvements and changes (usually in overnight hacking sessions). This not only improved the functionality of Colombus; it transformed Colombus from PARC’s idea to their tool. This experience became the basis for the design methodology throughout the Eureka project: codesign everything with the user community, making changes whenever necessary on a rapid and recurrent rather than cyclic basis in response to feedback. The experimental design for Colombus tried to account for the diversity in technicians who service the 5090 (5090 calls per month, years of experience, locale, etc.). Forty technicians were chosen to participate in the experiment. They were given laptops and three hours of training in using the software. Another forty constituted a control group, matched carefully for call frequency, experience, and locale (many were from the same workgroups). All 5090 calls made by both groups were tracked using the standard Xerox metrics for service. These included the cost of replacement parts put in the machine; average time per call; number of unscheduled maintenance (UM) calls (in contrast to scheduled preventive maintenance calls); percentage of time the customer called back for service within a short time after a UM call; and broken calls – the number of times a technician had to leave a job with the machine not yet repaired (for example, to get a part that was not available). During the test, the informal response was extremely positive. For example, technicians who were not in the experimental group would ask to borrow the laptop over lunch to help them with a difficult problem they were encountering. Although this was encouraging, the metrics after two months were startling. For the same, two-month time period, the experimental group had about 10% lower parts cost and 10% lower average service time than did the control group, without differing significantly in the other service metrics. However, this was not a long enough test to convince Xerox’s Worldwide Customer Services (WCS) – the organization responsible for service strategy and technology – that field knowledge was valuable, was worth investing

Communal Knowledge Sharing

263

resources to develop a Colombus-like system that could be used by the entire service force. WCS still believed that the best service would be obtained by having experts – primarily design engineers – create good documentation, supplemented by a commercial expert diagnostic system. They also believed issues like safety and legal liability would become a problem if anyone other than these experts helped define how field service was to be done. Thus, even when we extended the test to six months, with continued better performance by the Colombus group versus the control group (5% additional improvement in parts and labor, and a reduction in broken calls), WCS was not persuaded to change their investment plans. The results were convincing enough to our PARC team, though, and – more important – to Xerox France management. One field engineer put it this way: “This is the first time people have truly paid attention to the field, to our knowledge.” This led us to search for a way to extend the use of the tip knowledge base to all French technicians. French Minitel Eureka In extending the use of technician-invented solutions to the entire French service force, we faced two problems: the technical means for distribution and access that would fit into and support technicians’ work practice, and the social process to ensure the database’s value over time. The latter required us to work with the technicians to understand what would promote ongoing participation of a significant part of the work force and ensure continual updating of the knowledge base from experience. We could not use laptops to solve the distribution and access problem, and money for service investment was limited. In addition, in 1994 communication through phone lines or even the Internet was too expensive. A suggested stopgap was a printed tips booklet. We deemed it a poor alternative: it would make existing information available, but not help at all to make this an ongoing and growing resource. The French Minitel system was chosen as the key to the distribution problem. Minitel was a nationally deployed system of the French telephone company (PTT). It consisted of a small keyboard that connected to the phone and to a local display monitor (initially, a television). PTT distributed these keyboards free of charge after realizing they could quickly recoup their cost by printing fewer telephone books and fielding fewer calls to information. Xerox France technicians already used a Minitel-based service for call management – picking up new calls from customers entered by the work support center and closing out their calls when finished.

264

Making Work Visible

With the medium chosen, how were the messages to be generated? Our team worked with several different communities that were part of the service force to create a design for the process: the CSEs, the tigers, and the technical support hot-line specialists who both answered customer complaints and were first in the escalation process when a CSE had a tough problem. The hot-line specialists could have viewed this as an attempt to downsize their organization; but job cuts had already taken place, and they looked upon this as a way to potentially ease their workload. The tigers could have viewed this as “stealing” their knowledge, but they felt there were more than enough new problems. They also felt it was advantageous to get new solutions to tough problems quickly out to the field. The CSEs liked the idea that their hard-won field knowledge could travel beyond their own work group, but they worried about four issues: If they submitted a tip, would it disappear into a black hole? Would they get credit? How would they know they could trust each tip? And how would they get the right tip at the right time? In meetings with all the different community members, people came up with solutions to each of these issues. Quality: To ensure their quality, each tip would be warranted by a validator known for their expertise on the particular product line. In the Xerox France organization, there were already people designated as specialists for each family of products in every district or “Customer Business Unit” (CBU). The tigers would oversee the process to provide for a wider view. Bottlenecks: When a new tip was submitted on the Minitel, a message would be sent to the relevant group of validators, thus ensuring that there wasn’t a single point of failure. One of these should pick up the new tip within a few days of its submission. Rather than just pass/fail, the validator could converse with the submitter to ensure that the tip both captured the appropriate information, and was written clearly enough. The CSE would get a chance to edit the tip, and improve it, learning in the process. Incentives: We asked the community if they thought management should pay for each tip submitted. Not at all! As one tiger put it, “This would make them focus on counting the number of tips created, rather than on improving the quality of the database.” Their suggestion was to include the submitter’s name on each tip. This would act as positive reinforcement for good tips, and a negative one for flawed, waste-of-time ideas. The validator’s name and dates of submission and

Communal Knowledge Sharing

265

acceptance would also appear on each tip, providing open feedback for how well the process is working. Integration with work practice: With the tip database on the Minitel, new screens were added to the call handling. These allowed CSEs, when they decided they might be faced with a difficult call, to search the database based on key symptoms taken from the call record. They could also get directly into the database from a customer’s site if they could get access to the local Minitel. The organization of the tips was simplified to Problem, Cause, Solution. Entering new tips, and commenting and voting on the success of existing tips would usually be done from the office or at home in the evening. Implementation and deployment: Since implementing this system was a change from the plans that WCS had for the ultimate solution, they declined to finance this countrywide experiment. In partnership, PARC and Xerox France paid for the implementation of this Minitel based system. The software was ready in four months, but the key issue was how should the field be encouraged to use it. A champion from the French tiger group and a member of our group went around the country talking with each group of CSEs about service problems and how they could use this system to improve themselves and others. They met with over 60 product leaders, and helped train all 1300 French technicians. Participation was carefully tracked, both in terms of the number of times that the database was referred to, and the number of new tips entered. There were strong differences among workgroups: in cities of the same size, use could be high or quite low. Revisiting these regions, providing some training through examples of use, and reintroducing the purpose of the system helped encourage broad participation. This strategy, can be described as “hands-on participatory implementation,” which was a radical departure from the top-down, cascade model for program implementation commonly used in Xerox. Experience with use: The Minitel system opened with databases for three products. By the end of the first year, CSEs had opened over forty databases encompassing products from convenience copiers to highend printers. Also, more than one new tip was being added to the database each day. Over 77% of the tips were being validated within five working days, and 80–90% of the submitted tips were accepted. Participation was extraordinary. Over 20% of the CSEs had

266

Making Work Visible submitted a validated tip, and CSEs were consulting the tip database on the average two or more times a week.

So what were the technicians getting out of these tip documents? What did they think it was important to share? The tip content included diagnostic information, since that was obviously crucial, but there was much more. Some examples: *

Diagnosing unusual, costly failures

Bimetallic corrosion builds up on A and causes intermittent failures that seem to be B. Replacing B makes the problem seem to go away because A is moved in installation. First clean A, and later replace by new gold-plated AA, available as Part #1234. *

Workarounds

Paper curl in a dry environment causes excessive jams on baffle Q. Putting Mylar tape from tool kit on edge will ease problem. *

Easing the job

To make it easier to adjust M, paint Whiteout on the back wall near M. *

Comments on documentation

Before replacing the sensor P as the manual says in Repair Procedure R, inspect reflector Y because it might be out of alignment, causing this symptom.

Does this quite varied content make a difference to the company? Comparing France to the rest of Europe, it went from being just an average or below average service performer to being a benchmark performer. The French service metrics were soon better than the European average by 5–20%, depending on the product. This unexpected “experiment” was made possible by the fact that laptops were just being rolled out to Europe around that time, and no other European country had the communication infrastructure in place to adopt a wide-scale Eureka process. On a more qualitative basis, we saw three different kinds of effects on technician practice for France Eureka. In preparation for a call, using Eureka was helpful in ensuring that a part likely to be causing failure could be picked up before going to the customer site; this reduced broken calls. On site, Eureka accelerated and improved the diagnoses that CSEs made for the hard problems and reduced the number of times a call had to be escalated, thus reducing the load on the technical support hot line for recurrent calls about redundant problems. It also significantly reduced the

Communal Knowledge Sharing

267

learning curve for technicians on new products that are introduced into the field. Spreading Eureka In June of 1996, we decided to bring Eureka to another community, and to intersect more directly with the laptop rollout. At that time, some 6000 laptops had been deployed to Xerox CSEs, including 1200 in Canada to the entire service force comparable in size to France. In addition, we were able to team up with a tiger from the Dorval Technical Support Center, near Montreal, who was keen about the Eureka concept and became a local champion for its deployment. The challenge was to adapt what was learned from France to the Canadian service environment. We confronted an initial set of critical technical issues: *

*

*

The laptop was not widely used. While the corporation was committed to the computer as a platform to dispense technical information and manage work process in the field, technicians had long depended on their traditional skills and practices and most were skeptical about the need for this technology. CD-ROMs or floppies were the primary means of distributing information to the laptop resulting in sporadic and slow distribution. Separate applications were used for call management (dispatching and tracking all customer service requests), for the now electronically presented documentation (referred to as “e-doc”), and for parts inventory and ordering management. Thus, there was no easy way to make a connection between these independent applications and searching the Eureka tip knowledge base.

How we dealt with these issues provides additional evidence for how technical problems necessarily interact with the social and community context. We couldn’t directly solve the laptop acceptance problem. But we hoped that Eureka would prove to be valuable enough that technicians would be motivated to use the computer in their work. To achieve this, our local champion from Dorval Technical Support hit upon the idea of taking existing technical information databases, largely created by field engineers, that had been distributed in paper form to the field and converting them to Eureka tip format, to our Problem, Cause, Solution framework. This information was already valued by the field but hard to access or even keep track of in paper form. (Over time, this has resulted in Eureka becoming a digital portal to the field for all technical service information.)

268

Making Work Visible

To address the distribution problem, we decided to build a server-client system, with a thick client on the laptop that could be easily reconciled with the community database on the server. This would afford quick and continual updating of the knowledge base, every time a technician was able to dial into the server. To resolve the issue of what kind of communication and server to use, we once again built off common community practice. The common communication infrastructure for technicians at that time was a telephone connection to a bulletin board service (BBS). Some technicians used the BBS on a regular basis for discussing problems and sharing ideas, and we saw that this familiarity could work to our advantage as a platform for knowledge sharing with Eureka. There already was support for phone lines to the BBS, and technicians knew the protocol for gaining access. Still, this platform had no direct support for a Eureka process workflow. So we had to build programs to support this in a rather baroque programming environment. The BBS had a further disadvantage compared to Minitel that allowed users to easily add databases for new products, change the list of validators for any product, and directly manage the system. The BBS required centralized administration and a more complex social process which sometimes engendered delays and hence frustration in the community. Taken as a whole, though, the merits of using a system familiar to the community far outweighed the disadvantages. For accessing the knowledge base on the laptop by content, technician work practice dictated that our search engine had to be extremely fast, as well as easy to learn and use. A software engineer in Xerox’s Printing Systems Group (PSG), Bob Cheslow, had designed a program that met all these requirements, named SearchLite, evolving it over a long period of time in response to community feedback while working with the technical support group for PSG machines. His machine service history and knowledge was closely compatible with our Eureka design principles, so he joined us in designing the system architecture for Eureka. The integration of service applications on the laptop, so that they could be searched and managed concurrently with Eureka, was technically possible. However, the laptop software and documentation was distributed and maintained by a central organization. Eureka was just an experiment still operating on the periphery of this organization. At this point, then, it would have to remain another separate application. This peripheral status also meant that our Eureka champion for Canada had to play this role while simultaneously performing his main job as a tiger, handling escalations from

Communal Knowledge Sharing

269

and solving problems called in by technicians in the field. Validators were also, in a sense, volunteers who took the job on without being relieved of their ordinary duties. We also had to address community process issues with regard to adapting the Eureka tip authoring, submission, and validation methods developed in France to the Canadian context. *

*

*

Product specialists did all validation in France, with field engineering overseeing the process. Would this same division of responsibilities work well in Canada? Validation had turned out to be such an important aspect of the system’s success and value in France that managing this process for each different community wanting a Eureka system was an essential task. The French had rejected any financial incentives for tip authoring. In Canada, however, there was a financial incentive program already in place for submitting service suggestions. Should this same system be applied to Eureka? Because French technicians were using Minitel, they always had the most recent information in the knowledge base. The Canadian process would require technicians to explicitly download the latest information to update their client knowledge base. How often would they want or need to do this to make the system effective?

Because the organizational structures were similar in Canada and France, with product specialists in each CBU, it was natural to make those Canadian product specialists the validators as well. Incentive structures were another matter. Canadian service management did not want to change or give up the financial incentive program that they believed was contributing to significant improvement in service performance. Consequently, technicians received the same small financial reward for tips as for any other service suggestion; later, compensation was given only for a validated tip. Updating the knowledge base proved problematic. It required connecting to the BBS, but not all technicians had experience using it and a significant number found the process cumbersome. Moreover, the fact that Eureka was a separate application from call management, requiring dialing in to a completely different server (with a different phone number), created further obstacles to frequent, easy use. As a result, a large number of technicians – roughly, 40% – rarely or never updated their knowledge base. A social process was used to try to improve the situation, with the Canadian champion making visits to each CBU to encourage updating and provide training.

270

Making Work Visible

Even more complex than updating the knowledge base, was upgrading the software with changes. It required distributing floppy discs to everyone in the field, and hoping that they were able to use them in a timely manner. This created so many problems that a system for upgrading by downloading the software components from the BBS was eventually implemented. Finally, Eureka was no longer only an experiment initiated by PARC researchers and with few management expectations. Even the Minitel implementation, although it involved all French technicians, was still experimental in character. In Canada, Eureka was now an official, managementsponsored program, with certain expectations for service performance improvement and with some financial support from one of Xerox’s major business divisions. But management had never dealt with a program where the norm was that requirements emerged from user experience with an initial prototype, and where rapid prototyping with a consistent pilot group was conducted before any large-scale deployment. They expected programs where requirements were first “locked in,” with traditional methods for interviewing stakeholders to collect them, where large-scale deployment followed quickly on the heels of limited pilot testing, and where widely spaced, scheduled releases were used for any changes. Given these beliefs, managers tried to set expectations for when they wanted things done independent of our process for rapid prototyping and debugging with extensive community involvement. This clash of design and deployment methods – one expecting people to adapt to new tools as given to them, the other expecting a coevolution of people’s work practice and the tools they were using to support and enhance that practice – had negative results. When schedules could not be kept, the field was disappointed and higher-level managers lost confidence in the ability of the Eureka team to deliver; the team was discouraged. Despite these conflicts, Eureka was successfully launched for twenty products in only six months (starting in early 1997). For another four months, the Canadian champion trained the product specialists, with the specialists then training CSEs; a training video on CD-ROM was also created and distributed reducing the need for this direct training. Confronting the Organizational Challenge: Eureka Moves to the United States While Eureka had proven successful in France and Canada, these two countries each had a relatively small service force and geographic area. The United States was an entirely different situation: in 1997 there were

Communal Knowledge Sharing

271

close to 10,000 technicians spread over a huge area. Most important, the organizational dynamics were quite different in the United States: much more bureaucratic and hierarchical, due largely to its size and thus complexity. The U.S. rollout was gated by the deployment of laptops to all the CSEs. Since the United States and Canada shared a common infrastructure, there were no new technical issues. All the issues revolved around adapting the process to a differently shaped organization, and to the scale of the service force. Canada had around 100 validators; if the same percentage were to validate tips in the United States, 800 would be needed. It is easy to create a community of 100, where people can get to know each other and the skills they bring to a task. The same is not true when scaled by a factor of eight. Accordingly, U.S. service management decided that validation would take place locally, in each CBU, with local groups selecting a validator for each product family (the United States does not have product specialists associated with each CBU). As with Canada, the validators would take on the task without any change in the rest of their workload. The initial foray in the United States began in 1997 and involved a pilot program in several locations. The pilot only took hold, however, where there were local champions in the service force, as was the case with both France and Canada. Beginning in June 1998, Eureka CD-ROMs were distributed via the mail to field managers, who were then expected to distribute them to technicians in their work groups. The CD included a Computer-Based Training module, but no hands-on training or direct engagement with technicians around the program was planned. This strategy was designed for mass distribution of software or documentation, and cannot be uniformly effective with sociotechnical systems like Eureka. In places where people knew about the system and became champions, or where we engaged the local group it quickly gained traction and was widely used; in other places, it became one of twelve programs that had to be somehow implemented over the next quarter and adoption was correspondingly slow. An alternative, “participatory deployment” strategy for an organization with 10,000 technicians like the United States would have started with the pilot champions, the technicians and managers who were most knowledgeable about Eureka, going to a few other locations in the U.S. service community and talking about their experiences and ideas. Because these people were peers of technicians at those locations, they would be trusted. This would have created more local champions and knowledgeable users, who could then have gone to still more locations with a similar sharing method; a direct, hands-on engagement around Eureka would have spread across the entire country.

272

Making Work Visible

The upfront cost in time and travel for participatory deployment is greater than for the mass distribution, “spray and pray” strategy. But we believe this cost would have been recouped fairly quickly because larger numbers of technicians would start to use the program effectively, resulting in a shorter learning curve and better performance. The results from France and Canada support this argument. The main difficulty with getting management in the United States to embrace this idea was the lack of understanding of what combining the social with the technical actually required. Nevertheless, as has been the case with each service community, U.S. technicians, once they learned about the system, enthusiastically accepted Eureka. One technician’s remarks are typical: “In all my years in Xerox, the two best things ever given to us are the radios and Eureka.” In fact, although the original plan was to complete rollout in the United States before moving to any Xerox organizations in Europe, Latin America, or Asia, demand from technicians in these countries for Eureka was so intense that the corporation had to begin distributing the system worldwide. This was so even in places where technicians could not update their knowledge base in the standard way of connecting to the BBS, because of cost and technology limitations; here, technicians used floppy discs to share and update tips. One story that nicely captures the value of sharing the knowledge worldwide is featured in the 1999 Xerox Annual Report. A technician from Montreal traced a chronic problem with a customer’s high-speed color copier to a fifty-cent fuse holder and authored a Eureka tip. A technician in Brazil encountered a similar problem with the same copier – a problem so severe the customer wanted the machine replaced – discovered this tip during a test run of Eureka, saving Xerox the $40,000 replacement cost. In 1999, U.S. technicians authored approximately 2000 tips. Results were promising. There were over 9,000 “solves” using Eureka in the United States and Canada in the fourth quarter of 1998 alone. The knowledge base on the digital portal Eureka included over 30,000 records in all: tips, Service Bulletins, and other technical information documents. All the records were authored in English, but there was a language translation function for selected phrases and terms. Tips from the United States, Canada, and several European countries made up about 20% of the total and were the most rapidly growing portion of the knowledge base. By the first quarter of 2001, the size of the database had grown considerably as the number of countries using Eureka increased. There were approximately 50,000 technician-authored tips and over 300,000 records in all. Moreover, the number of problems solved using Eureka had increased to nearly 200,000 annually, by 2001, 300,000 solves.

Communal Knowledge Sharing

273

Having reviewed the story of how Eureka was designed and deployed in three different service communities in its first years of operation, focusing on the adaptation to their different organizational and technical environments, we now recount our efforts at that time to answer three important questions about its everyday use, through careful field studies: How did users respond to their experiences with Eureka? How did they adapt it to their work practice? What were the barriers to more effective use?

Field Studies of Eureka in Practice After Eureka had been in the field in the United States for several months, a member of our research group spent some twenty days over the course of three months traveling with and talking to technicians in San Francisco, concentrating on one work group in particular. We also visited four different CBUs around the country to try to understand the user’s experience of this new tool. We did this through group and individual interviews, intending to frame the discussion by asking technicians if Eureka was a tool worth using, and if so, how they used it and how it could be improved. Once they discovered that we were from the team that had designed and helped launch Eureka, they volunteered remarks like: Best reason we have for having a laptop. I use it on probably 50% of the calls that you go to where don’t walk in the door and immediately go “Well, this sensor’s broken” or something like [that]. Anything where something doesn’t immediately jump out at you, it’s the first thing I turn to.

We also had specialists tell us that they tried to encourage everyone to use it when they have an unsolvable problem, as the following remarks indicate. Most of the time before I get to a site, I try and figure out some directions to go. I look around in Eureka and see what’s out there so that when I get there I know what I’m gonna do. If we’re going to an escalation, we pull it up before we get there, if it’s not something that we’re familiar with, something that may be new. So it’s something we’ve gotten in the habit of doing, so we don’t even think about it. It’s very key and very important that way. A lot of times you’ll read something and you’ll think, “Man, I never would of thought of that.” Now maybe it doesn’t actually work that time, but now you’re thinking about a lot of things that will work that wouldn’t have ever occurred to you. You know there’s another train of thought, another set of eyes.

This was not surprising, since when first designed, Eureka was conceived primarily as a “tool of (near) last resort,” when routine fixes failed to resolve

274

Making Work Visible

a problem – particularly the intermittent ones that are so difficult to diagnose – and past experience didn’t point to an answer or line of attack. Many technicians certainly used Eureka that way, whether with machines they are working on or as a source for suggestions to be given over the radio when colleagues are asking for help. But we discovered that technicians use Eureka in several other interesting ways. Most striking was that many used it as a tool of first rather than last resort. For example, one technician who works on high-volume copiers told us how he used Eureka in combination with the product documentation. Before I go on a call I just like to see what picture I’m looking at as far as some [possible] fixes [in Eureka]. Then I drop into the documentation, if I felt that there wasn’t anything [in Eureka] that jogged my memory. And then I’m on my way. Because keeping that footprint of what some of the fixes were and then just going down through [the repair procedures in the documentation], it just accelerates things.

Thus, even before seeing the machine, this technician tries to develop several solid leads regarding the source of the problem, the likely repair procedures necessary, what parts were needed, and so forth. Many other technicians reported similar patterns of Eureka and documentation use: I check Eureka before I go to a site, that way when I get there I already know what parts to bring in. The customer is always impressed when you show up and already have an idea on how to fix the machine, and have the part with you. They just love it. It also saves me a lot of time. Whether it’s a whole trip back to get a part, and then go back to the customer, or it’s just a trip back to the car, I can use any spare time I can get. [Eureka isn’t] so much an end as it’s a beginning. Someone will call up over the radio with a fault code, and I can look it up in Eureka and can start scrolling through common causes. Actually it’s faster for me to find it in Eureka than it is to go in and fire up the documentation CD and go for the repair procedures [in there]. And not only that, getting into the book is just the book, you know, where here [in Eureka], if I bring it up and get a clue, not only am I gonna get a clue what that jam is, but also the potential area of fixes.

This technician also reported that he felt Eureka was useful even on those occasions where the tip didn’t provide the precise solution because it allowed the team to rule out certain sources of trouble, thus narrowing the search for the correct one.

Communal Knowledge Sharing

275

Finally, technicians also use Eureka as an informal learning tool. One technician who services mid-volume machines told us that browsing through the tips allowed him to see what has worked for others. And that’s basically what I do. Whenever I download [new Eureka data] I like to look through it. See what guys are doing. I go looking through the tips and bulletins. It just sort of teaches me a lot.

Essentially, by casually reading the tips and service bulletins, divorced from any actual repair situation, this technician was able to use Eureka as an instructor who offers a new set of lessons each week. This technician went on to note that attending to how others found workarounds has helped him develop his own skills. That is, beyond the actual content of the tips, he broadened his own way of working, his own knowledge. But what about the barriers to effective use? The laptop computer required to operate Eureka was both a resource – since it afforded rapid entry and sharing of tips throughout the service world – and a problem. The problem was partly the laptop’s technical limitations, particularly its short battery life and difficulty finding an available electrical outlet within reach at customer sites, the time required to boot it up, the instability of the Windows operating system, and the difficulty finding the free phone line that was then the only way to connect (dial in) to the Xerox server and thus update the database when in the field (remember at this time broadband connectivity to the internet and wifi was virtually unknown). In addition to these technical barriers, however, many technicians simply mistrusted, were unfamiliar with, and resented computers, and so did not use the laptop except when absolutely necessary. Some told us they did not use the laptop because they felt it added time and work to their daily routine. These technicians explained that when laptops were first introduced, they actually added workload by forcing technicians to do call handling on the computer, a task they previously accomplished by phoning in to a Xerox work support center. That is, it was done manually, through a human relationship and using a dependable device, the telephone. Two years passed, they claimed, before time-saving software was introduced to make the call-handling application on the computer more efficient and useable. Thus, while the laptops certainly had the potential to make technicians more effective in the field, at least some CSEs felt they made their lives more difficult. Our regular field observations confirmed these reports. We found we could travel with CSEs for several days and see the laptop used only once or twice a day, and only for call handling, with some work team members never using it. Many teams had developed workarounds to the computerized

276

Making Work Visible

call-handling system, with one member of the group logging on to the system, usually by connecting from a Xerox parts drop and almost never from a customer site, checking the queue and taking care of all the dispatch assignments for her coworkers. Here, the radio served as a way to keep each other updated on the status of the call queue for the group, and to communicate about dispatch assignments. At that time, many technicians had still not integrated the laptop into their everyday work practice. This can be contrasted to their experiences with the radio, which had become an indispensable tool for communication and collaboration with coworkers and thus an integral part of the technician’s work life. Radios, however, don’t have to be booted up; they are always on, providing a persistent connection to the work community. Radios also don’t lose battery power in less than three hours, and they don’t crash on a regular basis. Indeed, they don’t have a noticeable and difficult to manage operating system, offering instead a simple, fairly intuitive interface that borrows extensively from another, very familiar device, the telephone. Another barrier technicians identified was that as they were going through a work process, they had to enter information several times, and had to do independent searches in the Eureka knowledge base and the edoc. Too often the division between the tools that technicians use reflect their separate institutional development rather than the needs of those for whom they are designed. E-doc was completely separate from Eureka because each was developed separately. This integration of all the tools on the PWS was the biggest “ask” received from the field that came out of those feedback meetings. As one CSE noted: I want a dance partner that, when I out on the floor, we’re out there moving as one. I don’t want her stepping on my toes, and I don’t want to be banging into her, knocking her over.

The most striking feature of this comment is not that CSEs dance, but that they want a single partner when they do. It seemed plain to us that integrating e-doc and Eureka would enhance the tool for those who now used it regularly, and could draw in users who were afraid of the laptop. Updating the Eureka knowledge base had been a problem for CSEs since we launched the client-server system in Canada. As we noted previously, putting Eureka on the BBS separated the connecting and updating from the more routine connecting to the call-management system. As one CSE put it, “Is there any way to move Eureka out of BBS and into a system we routinely use on a daily basis?”

Communal Knowledge Sharing

277

In sum, we learned that technicians not only were using Eureka in creative ways, they were regularly thinking about making it more effective for their work. This is exactly the kind of inventiveness from the field that Eureka was meant to capture. It stands as further evidence of the pervasive importance of working with the users to make the system fit their needs – to artfully integrate the technology with their enhanced practice.

Eureka II – and Eureka Today Even before we did our field studies, we recognized many of the issues that were facing the technicians in using the laptop, and the opportunities new technologies were presenting, in particular the advent of cost-effective communication on the Internet. For reasons of space, we cannot go into detail about the next implementation of Eureka – named “Eureka II” – that was rolled out worldwide late in 1999 and throughout 2000. But we can mention a number of features in terms of the problems they solved for the work practice of technicians. Eureka II and the accompanying laptop suite of tools for technicians were developed by a multiorganizational team that included personnel from PARC and Xerox’s Worldwide Customer Services, with additional programming skills obtained through contracts with software houses. To bring together everything that technicians needed to do when connected, Xerox deployed a global-service network with multiple servers. As technicians logged on to the call-management system to get their next service call, the Eureka web server was able to download in the background any updates to the knowledge base. This same mechanism updated the e-doc, and if necessary updated the software they were using on the laptop. All the information sources became accessible through a single search mechanism based on searchlite. So when technicians had a problem, they could see “hit-list” references to tips, and multiple places in the formal documentation where they might find helpful information. In addition, annotations could be made on already existing documentation. Such “post-its” could be kept in a private knowledge base, or if desired submitted as a tip. When validated and shared, it would appear not only in the hit list directly, but as a link on the page where the annotation was made. The success of Eureka as a knowledge-sharing framework inspired other work communities in Xerox to adopt this practice. For example, software analysts started using Eureka. We also moved from the bulletin board system that had many features, both administrative and functional, that were unnecessary for Eureka, and created a much simpler web server

278

Making Work Visible

that would both hold the knowledge base, and support the simple workflow from authoring to validating to subscription management to downloading. Simplification and stripping out the unnecessary were goals we hoped to achieve. Eureka was so successful that the system was now an official, “mainline” program in the company. This meant that requirements from many places poured in, and we were constantly trying to balance our belief that simpler was better with corporate managers’ beliefs that if Eureka was the answer they wanted to be the one to generate the question. Most notable of these was a service manager’s intuition that the best way to simplify the training for technicians was to embed the Eureka application in the standard Internet browser (e.g., Internet Explorer). We tried to explain that this would complicate the implementation significantly because the software would then be dependent on exactly which version of the browser, which operating system, and which service packs were loaded on each machine. The response was that this was not as important as simplicity of training. We explained that knowing how to browse only minimally overlapped with knowing how to use the functionality of Eureka for searching, authoring, and so forth. But as manager of this major program with years of experience working with the field, he insisted that the tradeoff would be worth it. The program subsequently made that choice; unfortunately, as we carried out the Eureka II release, we found that there were far more complications than imagined, and the delay was significant. The issue is not a manager’s mistake, but rather how decisions are made in the standard software development process, in contrast to the bricolage approach with which we had started. In addition, because the project now had the attention of top management, schedules were sometimes based on managers desires to achieve certain goals by certain times, rather than on the work that was necessary to achieve the desired state as indicated by the implementers. We understood the pressures that these managers felt, nonetheless this often led to internal battles, and then to the appearance of slipped schedules as managers pronouncements could not be realized. It made us very aware of the difference between singing in the spotlight and singing in the shower. In the years since Eureka II was deployed, however, that spotlight eventually shifted back to the technician community, which is where it belongs. PARC certainly played a leadership role in the conceptualization and design of the Eureka system at the start and through its early years, but as our story makes plain, the very idea of sharing new solutions to vexing technical problems was invented by the CSE work community – it was a

Communal Knowledge Sharing

279

basic feature of the everyday practice in their local groups. And it was fundamentally different in character than using the service manual RAPs (for recent and detailed comparison between practices using the RAPs versus the Eureka tips, see Vinkhuyzen and Whalen 2007; Yamauchi, Whalen, and Bobrow 2003). Having observed this practice in the field, PARC collaborated with the community to develop technology that could dramatically enhance this practice, making it possible to share ideas all around the world. But from the beginning, the technician community managed the actual tip sharing process, with peer validation as a key element. And it was not too long before community leaders took over the technical maintenance and future development of the system from PARC, which was always our goal. Accordingly, Eureka is community owned and maintained in a way that few if any corporate knowledge systems are today. We firmly believe this is what has made – and continues to make – Eureka so remarkably and perhaps even uniquely successful. Over the past eight or so years, under the leadership of CSEs, Eureka has of course continued to evolve in design and grow in size and participation – and in value. By 2005, for example, the number of solves documented worldwide was nearly 800,000. Technicians across every part of the globe where Xerox products are installed now use the system, and it is estimated that this pervasive use saves the company close to 20 millions dollars each year in service costs. With respect to size in particular, there are over 100,000 tips registered in the database. This has led to new concerns. For instance, how do you maintain a database of this size and, more important, how do you deal with the inevitable redundancies and complexities for a system of this scale? Although the tips are still organized by product family, and so the size of any single family’s database remains manageable, searching for and finding the “right” knowledge amongst all this information – recall that e-docs and all other technical information was integrated into the system in Eureka II – is a challenge. Nevertheless, CSEs have found ways to deal with these challenges. For example, to handle apparent redundancies, technicians have elected to keep most of the tips in question in the database, finding that whatever the similarities, each tip makes its own distinct contribution in some fashion. Moreover, each tip represents the efforts of a community member , and they are reluctant to delete the product of that work. There are some efforts being made at combining information, though, preserving the recognition earned by all contributing parties. This is more evidence of the complete intertwining of the technical with the social with Eureka, the knowledge system and the community that creates and culturally “owns” that knowledge.

280

Making Work Visible

Finally, it is important to recognize that in this and other ways, Eureka foreshadowed – and was a model for (whether publicly acknowledged or not) – the emergence of what today is called “social media”: Internet- and web-based technologies that allow the creation and exchange of usergenerated content, supporting the democratization of knowledge and information (e.g., Kaplan and Haenlein, 2010). The web sites based on such technologies and the basic human need for social interaction include those that promote social networking (e.g., Facebook, LinkedIn, MySpace); collaborative knowledge creation (Wikipedia); social news (Digg); opinion sharing (Yelp); and community Q&A (Yahoo! Answers). Businesses are just beginning to figure out how to take advantage of these “from-thebottom-up” technologies and the social interaction and massive networking they enable amongst their employees and between their employees and customers (Li and Bernoff, 2008; Shirky, 2008). At least in terms of internal corporate networking and user content creation, Xerox and PARC were in a strong sense leading the way. It remains to be seen whether and how future versions of Eureka might incorporate Web 2.0 technology and the communal practices it can support.

Organizational Barriers to Change At the end of the day, the story of Eureka is a tale of how the development and deployment of a system for sharing knowledge created on the frontlines became a vehicle for organizational change. However, our story also reveals that this change was not without conflict and challenges. These are precisely the kinds of messy details that are rarely included in the writings on knowledge management or organizational learning. It is worthwhile to expand a bit on these details and their larger meaning for knowledge sharing in practice. In the initial stages of the project, during the experiment in France, few people in Xerox management believed that there was much value in what the technicians learned on their own in the field. Additionally, they could not see how a tip system from the field was much different than prior suggestion systems, all of which were highly centralized and controlled. And although technician tips quickly proved valuable, people in different organizations in the company felt that it was more important to get centrally produced documentation in front of the technicians than support them in creating new knowledge. This new way of doing business made them nervous; in a sense, the corporation’s protective “immune system” against threats to a well-entrenched way of doing business was activated.

Communal Knowledge Sharing

281

With these obstacles, getting support in the form of organizational resources for the project proved difficult. Time and again, as the story of Eureka makes clear, we ended up relying on local champions who somehow managed to cobble enough resources together to get the job done. Moreover, in the initial stages, we had to sometimes operate like a guerilla group because opposition to the idea of a bottom-up knowledge system based on the value of field know-how was enough to kill the project if we openly challenged such deeply entrenched convictions; our first experiment in France was conducted there in part because it was out of sight of the central Xerox organization. After France, it was only by convincing one product manager to give us the funds to try the Canadian experiment that we were able to finally get data that would convince the nonbelievers. (Ironically, the person who led the French effort, was later given an award from the Worldwide Customer Service organization for launching Eureka that read “despite the resistance of WCS.”) Speaking of convincing data, we recognized at the start that informal response to the collection of tips and the users informal assessment of their value would not be accepted by the service organization. We knew we had to show hard, “bottom-line” data from believable experiments. But comparative studies are more difficult if people are clamoring to get a feature. Then everyone improves, but there are no comparative metrics. In some ways, resistances in the organization to making a wide-scale change like Eureka worldwide in just one step allowed us the opportunity to gather better measures of the kind they would accept. We believe that some of the more important measures are the responses of the field to being listened to and the potential for changing how we do business. For example, we could put out a product with only minimal diagnostic documentation (getting the new product to market more rapidly) and then use the field force to help understand where and how it needs improvement – that is, to construct the diagnostic documentation in the field. Once Eureka did begin to receive official recognition and support from CS (they eventually assigned one of their best people to champion it), as the performance data and, more than anything, the enthusiasm of the technicians themselves overcame resistance, and deployment in the U.S. was approved, the project ran into a very different sort of problem: Eureka became a major corporate program. But why was this problematic? In France and Canada, since we were conducting guerilla experiments, we could involve the users more in the decisions of how to adapt Eureka to local needs and practices. Moving to the central WCS organization, however, engendered a change of philosophy. As Worldwide Customer Service,

282

Making Work Visible

they felt they had to design a uniform, worldwide solution. The power was for the most part wrested from the users’ hands. Related to this, management was familiar with distributing corporate programs in a hierarchical (“cascading”) fashion. Eureka works much better when peers mentor each other on the system’s uses. The French and Canadian experiences demonstrated the value of participatory deployment. Unfortunately, as we have explained, this was not how the rollout in the United States, by far the largest Xerox organization, was done. While the U.S. deployment was eventually very successful, the problems that developed because technicians weren’t closely involved in the process (especially those technicians who had the most experience with Eureka and could have been an invaluable resource for their coworkers in other parts of the country) hampered the project’s achievements over the first year and more. This can be explained in large part by recognizing that WCS, and Xerox Corporation more generally, has emphasized cost savings in field service, and has viewed the service organization as a cost center. As a result, significant investment in service is discouraged, unless it is matched at the time of investment with equivalent cost reductions (instead of seeing investment as a front-end expense that will have a back-end, long-term payoff). This had consequences for Eureka deployment, and for how the program now operates. When the service organization was asked to change how it viewed the importance of field knowledge, the field technicians who were charged with contributing to the new knowledge base and maintaining its quality were not given enough time to learn and engage in the change activities, because this would have required a significant investment in time (and thus money). As a result, there was not enough time to reflect on and practice the new way. Although the president of Xerox said that Eureka was a key program, the organization did not want (or were not able) to make the necessary investment for building the kind of process infrastructure – such as training resources and time – to make it more successful. More important, with respect to program operation, technicians were expected to author tips and validators were asked to provide rapid turnaround and validation of submitted tips – a critical role in the process – without giving any of them relief from their current workloads. Finally, we have to ask; did Xerox become a better Learning Organization as a result of the Eureka project? One way to answer the question would be to point to the impact on field-service performance, and to the degree to which most technicians use the program on a regular basis in many different ways, including especially learning new ideas and approaches to machine repair. The service organization, taken as a whole, has also been transformed

Communal Knowledge Sharing

283

to some degree by the bottom-up Eureka approach, which has had an impact on this particular organization’s operational philosophy. Still, although these are certainly significant achievements, after almost fifteen years, the corporation – the corporate “organization of organizations” – has not yet taken full advantage of what a knowledge-sharing program like Eureka makes possible. By this we mean that the sharing of knowledge between organizational units in the corporation, such as those responsible for field service, training, documentation writing, engineering, and manufacturing, has not been affected by Eureka. This is despite the following. *

*

*

*

At least some of the information contained in the Eureka knowledge base could be incorporated into the documentation, the nowcompletely digital service manuals. As a key part of the diagnostic process in the field (and, more broadly, continuous learning in the field), Eureka use should be included in all service training. Manufacturing needs immediate feedback from the field about the performance of new machine models, and while they receive a good deal of data about machine performance during the first (and most critical) ninety days of a new model being introduced, they have many problems with the usefulness of this data and believe that Eureka could provide a vehicle for collecting much more useful information, directly from those in the best position to assess a new model’s operation, the technicians. Engineering, which is responsible for designing new machines, could benefit from direct experience from the field as a resource for determining the value of or problems with various features and functions, as well as operational design more generally.

The current Eureka process, which is dedicated to technicians authoring tips for fellow technicians – and not, then, for documentation authors, engineers, quality control experts in manufacturing, or trainers – obviously cannot address all these areas or solve the interorganizational knowledge transfer problem. It does point the way, however, toward what a solution could be, and to the need for making use of knowledge collected on the corporation’s frontlines (no one in the corporation spends as much time with Xerox customers or knows as much about Xerox machines as technicians) throughout the corporation. That would help Xerox become a true Learning Organization. At the same time, what we could perhaps describe as the “spirit of Eureka,” which is the belief in the value of bottom-up knowledge sharing,

284

Making Work Visible

of the importance of drawing directly on the experience and creativity of frontline employees, has had some interesting effects on the corporation as a whole. The most interesting of these is the number of requests to our group at PARC from different parts of the corporation as well as from other corporations for helping to create a “Eureka-like” knowledge system for their organization or operational unit. In the past eight years, we have developed systems to be used by the Xerox sales organization to share knowledge about “customer solutions” (special configurations of machines and services that help customers solve important business problems), by Telecom Italia customer support centers to help answer customer questions and requests for service, and in collaboration with General Motors Research Laboratories, by manufacturing engineers in assembly plant body shops to track “build” problems and solutions across shifts and between different plants. It may well be that not only the systems themselves but the belief in and spread of this “spirit” becomes the most important legacy of Eureka in Xerox and elsewhere.

15

Designing Document Solutions for Airline Maintenance Advisories Patricia Wall and Johannes A. Koomen

Introduction At the turn of the 21st century, Xerox Global Services (XGS) was looking to build on Xerox’s traditional strengths in product development and solutions sales by providing business process consulting and outsourcing services for large enterprise customers. This would expand the value Xerox could bring to its large customer base by helping customers reduce operating costs and improve processes. XGS was deploying conventional approaches such as Lean Six Sigma (LSS) as well as developing a set of assessment tools to help guide and define their consulting practices. One area of interest to XGS was the ethnographic-based Work Practice approach, already in use in Xerox Research for several years. Prior to the existence of XGS, extensive studies of field service technician work (Orr, 1996, 2006; Bobrow and Whalen, 2002, Whalen and Bobrow, this volume) illustrated the value of applying work practice methods to understand technical work to inform the design of technology solutions intended to support that work. XGS recognized that the qualitative Work Practice approach was an ideal complement to the quantitative nature of LSS studies. Although the facts and figures resulting from a LSS study are necessary to demonstrate and justify the need for new or changed processes and technologies, they do not sufficiently support all complex consulting engagements. In many cases, a deep understanding of the social and cultural aspects of a work process is necessary to define not only what needs to change, but how best to make those changes so that they fit in the social context of the work and are adopted by the workers with minimal disruption. In 2003, the work practice team in the Xerox Research group was asked to support an XGS project with a major US airline. The airline wanted to improve the issuance of Maintenance Advisories, which the company’s maintenance engineers used to communicate new or revised procedures and best 285

286

Making Work Visible

practices to 3500 aircraft mechanics and inspectors worldwide.1 The authoring, approval, and distribution process for these advisories was manual and paper intensive. The underlying motivation for the project was the airline’s desire to ensure that its entire staff of mechanics, inspectors, and their supervisors read and acknowledged new maintenance advisories within twenty days of publication. Full compliance was not only company policy; it was a Federal Aviation Administration (FAA) requirement. Anything short of 100% compliance put the company at risk for FAA fines. In collaboration with XGS and the airline, the work practice team was asked to comprehend current maintenance advisory practices and issues, particularly with respect to the read and acknowledgement steps of the process, and to propose solutions to address observed issues and ultimately improve compliance.

Scope of the Study In discussions with project sponsors, the maintenance process was selected for study because internal audits indicated that advisories were not consistently being read within twenty days of publication. The process was labor and paper intensive, a potential opportunity for improvement by making the documents and reporting tools available online. The airline had maintenance manuals online and recently started to make the maintenance advisories available online. They were in a transition phase; although the advisories were online, the processes around their creation and use were not. (The transition from paper to digital practice is noted elsewhere in this volume. See Tolmie; Watts-Perotti et al.; Martin and Wall; Koomen for a variety of perspectives on this topic.) The project sponsors felt the issues with compliance could be solved by looking at the last stage of the process, where maintenance crews read and acknowledged the advisories. Some of the sites had better compliance than others and the processes for tracking compliance varied across the sites. The fieldwork team requested a more comprehensive view of the advisories, including where they came from and how they were made available to maintenance personnel. Initial explorations resulted in structuring the study to encompass all aspects of the advisory process. Figure 15.1 illustrates an overview of the 1

See Gouws et al., 2006, for an overview of studies, or the lack of, on the nature of maintenance work; See Reason and Hobbs, 2003, for a summary of a study of airline maintenance errors.

Designing Document Solutions for Airline Maintenance Advisories

287

Figure 15.1 Overview of the key steps in the maintenance advisory process

process, including creation of advisories, approval, distribution, read and sign off (acknowledgement), and follow-up reporting/audits. As it turned out, there were no formal descriptions of the entire advisory process, although some organizations created standard operating procedures, primarily checklists, to capture steps in the production of advisories.

Methodology The team used an ethnographic approach to understand current practices related to maintenance advisories. This included open-ended interviews and observations of study participants in the context of their work environment and activities. Interactions were videotaped for later analysis. Fieldwork was conducted by two teams, each consisting of an ethnographer (from Research) and a consultant (from XGS). Study Sites It was decided to study five airline locations representing a range of issues and practices associated with maintenance advisories. The sites included: headquarters where advisories were created and distributed; a “heavy” maintenance location where aircraft undergo significant routine maintenance; and three airline hubs where rigorous daily maintenance activities are carried out on outbound aircraft. All locations, except headquarters, had maintenance technicians and inspectors who were required to read and acknowledge advisories. Each maintenance location instituted different processes for managing and tracking the read and acknowledgement process.

288

Making Work Visible

The Essential Field Guide A maintenance expert currently working at headquarters was identified as our “guide” for the five sites during the course of the study. Since we required access to maintenance areas not normally accessible to the public, our guide negotiated access at each of the sites. It proved extremely helpful that he came up through the ranks of the maintenance organization as he had relationships with each of the sites and knew many members of the maintenance staff personally. Our guide provided an overview of the maintenance advisory process which provided an initial roadmap and structure for the fieldwork. One of the challenges of conducting a work practice study as part of a services consulting study is that consulting engagements tend to be brief relative to work practice studies conducted in research. This study was limited to two weeks in the field with customers, which included all five sites. Logistically, it was helpful that three of the sites (headquarters, one hub, and the heavy maintenance site) were located in one city. In addition, since the maintenance crews worked in three shifts, it was possible to conduct fieldwork over two shifts on some of the days. Airline Headquarters The study was initiated at airline headquarters, where technical publishing operations took place. We interviewed three members of senior management (also project sponsors) to obtain their view of the advisory process and to identify challenges they were facing. This also provided the opportunity to observe advisory creation and distribution activities underway by the technical publishing staff, including two authors, an editor, an IT support person, and publisher. Maintenance Sites The team studied maintenance managers, crew chiefs, mechanics, and inspectors in each of three airline maintenance sites. Maintenance managers were responsible for ensuring and tracking compliance for the maintenance crew at a given airport. Nine managers participated in the study. Crew chiefs were responsible for ensuring that the technicians working on the aircraft had time to read and acknowledge current advisories. Nine crew chiefs participated in the study. Mechanics (also referred to as technicians) and inspectors were responsible for performing aircraft maintenance and ensuring that proper procedures were followed. Twenty-four mechanics/inspectors were

Designing Document Solutions for Airline Maintenance Advisories

289

interviewed and observed in maintenance bays or break rooms where computers were available for reading maintenance advisories. Sign-off binders were also located in these areas. Analysis At the conclusion of the first week of fieldwork we completed interviews and observations at all but one fieldsite. With teams conducting interviews and observations across four sites we collected more than forty eight hours of videotapes during the first week. The team completed a preliminary round of data analysis prior to the second week of fieldwork. The preliminary analysis involved the review of videotapes, transcription of selected tapes or segments of tapes, and the creation of several illustrations summarizing current practices, issues, and opportunities. This analysis took about four weeks and was shared, verified, and refined with representatives from the fieldsites. This provided some flexibility, so the team could study the fifth site and use the remaining time in the second week to engage the participants in workshops to discuss potential solutions and implications for current work practices.

Findings: Observations of Current Advisory Practices To get a first hand view of the Advisory practices, the team conducted interviews and observations of people involved at each stage of the workflow. Our observations served as the basis for documenting current practices as well as barriers to compliance, which provided the foundation for developing new solutions. We gained insights about how the process was supposed to work, where and why it didn’t always work, and what people were doing to compensate for problems they encountered. Following are key observations about each phase of the Advisory workflow. Authoring Advisories were authored by engineers who monitored maintenance notices published by various aircraft and engine manufacturers. A draft advisory was written using a word processing program, but might be handwritten or a combination of electronic and paper documents containing illustrations or photos. The draft was reviewed with a manager, revised as necessary and then sent to Technical Publishing for editing and approval. The draft was scanned and stored by the Technical Publishing department as part of the document history.

290

Making Work Visible

Approval The Approval workflow was initiated when a draft advisory was received in the Technical Publishing department. Since drafts came in a variety of formats and states of completion, an editor assigned the advisory a tracking number and put the draft advisory into a Word template. If there were illustrations, this could involve scanning to convert them to an electronic format and/or adjusting the images to fit into the template. The Word version was then sent back to the author for review and revision. Once reviewed by the author, the approval cycle was initiated. Technical Publishing printed the advisory for approval by the original author, their department manager, the Managing Director of Quality Assurance, VP Technical Services, and the VP of Aircraft Maintenance. A routing slip with the approver names and a place for each signature was attached to the printed copy. The advisory was circulated to each approver in turn. Any of the reviewers could suggest changes. Once the approval cycle was completed, the advisory was put aside for a batch release of advisories. Each advisory release was accompanied by an Advisory Index that listed new, current, and recently expired advisories. Distribution Distribution of the advisories was the responsibility of Technical Publishing staff that created a PDF of the advisories, stored them in a central location and sent an email to another group in Technical Publishing who prepared the files for insertion in the online system (referred to as eDocs). This step involved adding the files to a database and a web site. When the advisories were available online, an email was sent to an “Advisory Notification” group list, announcing the new release to all maintenance managers and supervisors. A set of the advisories were also printed and inserted in a master binder in the company’s library at headquarters. Advisories used to be copied and mailed to maintenance stations around the world. Providing electronic access to advisories in the maintenance stations was intended to reduce time and distribution costs while ensuring all stations had access to the advisories immediately following publication. Read and Acknowledgement of Advisories This process was different for managers/supervisors than for technicians. Once managers and supervisors received email notification announcing new

Designing Document Solutions for Airline Maintenance Advisories

291

releases, they could go into the eDocs system from their PC’s, read the advisories, and reply to the email indicating they had read the new advisories. The email was kept as a record of acknowledgement by the person tracking compliance for the site. The process for technicians was somewhat different. Once the maintenance supervisors at the airports were notified of a new advisory release, they would announce the release at daily crew meetings. This was necessary because the technicians did not have company email accounts. Each maintenance bay had a few PC’s available providing technicians’ access to online technical manuals and maintenance advisories. The PC’s were located either in break areas or small rooms off the main maintenance area. Once the releases were announced, the crews had twenty days to read the advisories and then sign their name in a three-ring binder with acknowledgement sheets (ordered by advisory name/number). Supervisors were responsible for tracking technician compliance. Binders containing technician signatures were checked by the supervisors periodically to ensure that their crews read and signed off each advisory. This is where the most time consuming and tedious practices were observed. The supervisors spent a significant amount of time (up to three hours per week) checking the binders, page by page, and matching crew signatures against each advisory sign off sheet. If any technician was found to be delinquent or in danger of delinquency, the supervisor had to rearrange maintenance work to make time for the technician to read the advisories. Some delinquencies were due to legitimate reasons, for instance, a technician was on vacation or in training. The legitimate excuses had to be noted, so there were no consequences if those delinquencies were exposed in an audit.

Audit Follow-up and verification of compliance was done with internal audits as well as external FAA audits. Notification of an impending audit was not required. When there was notification, a master list of advisories was printed out in preparation for the audit. The person responsible for audit support (we saw managers and administrative support in this role) collected all the binders containing the lists of technician acknowledgments so they were available for review by the auditors. Auditors reviewed the binders or relied on a summary of compliance status provided by the maintenance station manager. Any discrepancies in compliance, that is, advisories unacknowledged past the twenty day limit, were documented.

292

Making Work Visible

Figure 15.2 Workflow in transition from paper to digital

In Transition from Paper to Digital Prior to advisories being available online, they were printed and kept in a folder near the sign-off binders, so the technicians could read new advisories and then sign the acknowledgement. There were two issues with printed advisories. One was that they might not be the most current version, so the technicians might read an out-of-date advisory. The other issue was that the advisories tended to move around the maintenance bay, or even between bays, so technicians might not be able to read the advisories even if they had some time available. Making the advisories available online alleviated these problems. However, the second issue continued to apply to the acknowledgement binders. Even if a technician was able to read the advisories, he might not be able to sign the acknowledgement because the sign-off binder was nowhere to be found! Figure 15.2 illustrates the mix of paper and digital aspects of the observed workflow. Incorporating the acknowledgment in an online system by capturing technician’s digital acknowledgements would alleviate issues with missing binders.

Noncompliance Undercurrents Throughout the study a recurring theme arose about maintenance staff motivation when it came to reading advisories. Some managers raised a question about how seriously maintenance staff took their responsibility with respect to advisories. Was noncompliance an indication that the advisories were not being taken seriously? Were the technicians being resistant or was the process making compliance difficult?

Designing Document Solutions for Airline Maintenance Advisories

293

Figure 15.3 Aircraft in a heavy maintenance facility, where aircraft go through several weeks of maintenance

The work practice approach proved to be particularly effective in exploring these undercurrents; in particular the perspective of the technicians, which on the surface might be interpreted as resistance to compliance. Interviews with technicians revealed that they took their role as maintenance technicians very seriously. There was a sense of immediacy and attention to detail in the work, with an ultimate concern for passenger safety. The technicians were performing all kinds of maintenance activities, on a variety of aircraft, in busy airport environments with tight schedules and turnaround times. The participants all voiced the importance of keeping current with advisories. However, on a busy day (or a string of busy days) there may not be time to leave the aircraft maintenance area, log onto a PC to read the most current batch of advisories, and locate the binder to sign off on them. Technicians were also in agreement that they should be knowledgeable about all aircraft models in the fleet, even if a particular model did not regularly land at their airport. (At the time of the study, the policy requiring maintenance staff to read all advisories, even those for aircraft not typically scheduled to land at their airport was under review.) One technician mentioned that they needed to be prepared to perform maintenance on any model in the fleet because “you never know when an aircraft that may not normally land at this airport, for example a Triple 7 (Boeing 777) might be rerouted to your airport due to weather. You don’t want to have to fly a maintenance crew in to maintain it.” To address potential delinquencies, supervisors would schedule time in the technicians’ day for them to catch up on advisories. Figure 15.3 illustrates a heavy maintenance facility, where a complete maintenance may take several weeks. The technicians’ primary focus is on the aircraft, which is removed from the areas where PC’s are provided for reading advisories, the documents that may provide critical knowledge to successfully perform maintenance work.

294

Making Work Visible

The opportunity for a technology solution was to make the process as easy and accessible as possible for technicians, supervisors, and their management. For technicians, this might mean that when they do take the time to read advisories, the system makes it possible to acknowledge advisories as they are read, rather than hunting around the maintenance bay for sign-off binders. For supervisors, this might mean providing assistance by automatically tracking crew compliance and sending reminders when compliance deadlines are approaching. For managers, this might mean automatic generation of compliance reports and automatic reminders to supervisors if compliance is at risk.

Translating Field Observations to Solution Design The work practice study uncovered several issues in the advisory workflow, suggesting opportunities for improving the process while building on practices that are already working. While the initial request was that the study focus on the read and acknowledgement part of the process, taking a more comprehensive view by looking upstream at where the advisories originated revealed opportunities to streamline the entire process. Making the Work Visible Using videotapes and transcripts of interviews and observations, we created illustrations representing the key features of the observed work practices for each part of the advisory workflow. These highlighted activities in the workflow that could benefit from redesign and the introduction of technology. As the team explored potential workflow and technology alternatives, these were mapped onto the current workflows to get a preliminary view of the impact of the proposed solution. The representations were shared with the project sponsors to discuss and refine proposed solutions prior to the start of implementation. Figure 15.4 illustrates the current Read and Acknowledgement workflow and how it would change with the proposed solution. The proposed solution would eliminate the sign-off binders, automate email notifications, and enable electronic acknowledgement for reading the advisories. Similar illustrations were created for each phase of the advisory workflow. Solution Design Based on the work-practice assessment, the team codesigned a solution with the client to address the key issues that impacted those involved in the

Designing Document Solutions for Airline Maintenance Advisories

295

Figure 15.4 Representation of current and proposed practices for the Read and Signoff portion of the workflow

Advisory workflow at every level (headquarters, management, supervisors, and maintenance crews). Requirements were based on the problem areas identified in the work practice study, primarily focused on streamlining the manually intensive, time-consuming aspects of the work. Having a team member with computer science expertise facilitated the translation of observed problems into descriptions of technology capabilities that would address the problems. This translation took the form of a preliminary system architecture consisting of software components, definition of smart document templates,

296

Making Work Visible

and proposed workflows. This served as the primary specification documentation for the system design. Smart Documents To simplify authoring, the solution used the new “smart documents” technology in the forthcoming Microsoft Office 2003 suite. XML-based templates standardized the Maintenance Advisory format and guided authors through the authoring process by prepopulating certain sections and providing authors with standardized sets of choices in other sections, such as to what aircraft model or power plant type the advisory was applicable. When the authors submitted a completed draft advisory it was stored in a repository, and a workflow database was updated accordingly. The system was integrated with the airline’s email infrastructure, and the author’s supervisor was notified by email of the new draft advisory. The supervisor could then approve the advisory as submitted, or provide comments and suggestions for revision which would be returned to the author. Once approved by the supervisor, the advisory was then automatically routed to designated approvers in the Quality Assurance, Maintenance, and Technical Services departments simultaneously. Each approver could approve or reject the new advisory electronically, and independently of the others. Upon receipt of all required approvals, the system generated an email automatically to all maintenance managers and supervisors in the field. Because the technicians did not have individual email accounts, announcements of new advisory releases were still made by supervisors at daily crew meetings. However, the 3-ring binders used to record acknowledgements were eliminated. An electronic acknowledgement feature enabled technicians to “sign-off” on advisories as they read them online. The new solution also automatically notified supervisors of the approaching 20-day read/acknowledgement deadline. Reminders were generated if the deadline approached and compliance was less than 100%. In addition to “smart document” templates, the system utilized a central repository, an SQL database for workflow status tracking, and Excel templates and macros for reporting. The system was integrated with the airline’s internal web servers for browsing, reading, and acknowledgement of active advisories. As the solution was implemented, the work practice team continued to work closely with the implementation team to help translate field findings into the solution. Some of the interactions were around technical implementation of the solution, for example, should it be possible to access expired advisories? Others were oriented to technician’s interactions with

Designing Document Solutions for Airline Maintenance Advisories

297

the interface, for example, how to indicate which advisories are pending, and which have already been read. Participant Feedback The team received feedback as the solution was put in use. The manager of advisory tracking and compliance at one airport, a study participant who had developed one of the most comprehensive compliance tracking systems we observed, had the following feedback regarding the solution: “Truthfully, this program has eased up numerous hours of tracking. On time compliance rates are up drastically. The flow of information is continuous, fast and reliable. All of which the airline industry has needed . . . Thank you for the help and listening to our needs and wishes from ALL our technicians.” Business Impact From a business perspective, the impact of the solution resulted in significant savings, on the order of $750,000 annually. This included savings from reduced costs to create, approve, and distribute the advisories as well as elimination of nonvalue add activities associated with compliance checking and accessing and acknowledging the advisories. The metrics to determine savings came primarily from the work practice study as it uncovered practices that involved time-consuming manual activities as well as participant-suggested ways to improve the process. The solution provided the underlying technology and improved workflow to achieve compliance targets.

Reflections and Conclusions This project was a collaborative effort between the Xerox Research and the Xerox Global Services intended to explore the benefits of blending traditional consulting approaches with a work practice approach. The consultants regularly conducted structured interviews and workshops with stakeholders to map workflows and capture issues, using this to guide solutions design. The work practice approach provided an in-depth view into a range of practices, including a cultural dimension, opportunities for improvement, and insights about how to support change in this context. This project was among the first of several collaborations between Xerox Global Services and Research. Other chapters in this volume (Sprague et al., Chapter 7; Koomen, Chapter 13, Plurkowski et al., Chapter 19) describe some of the other collaborations.

298

Making Work Visible

The brief time in the field that comes with consulting projects was not without its challenges to the work practice approach, which encourages ongoing observations and interactions in the field over time. Videotaping the interactions helped capture the details of our interactions in the field. But then again, it was a challenge to analyze and summarize results and recommendations in the scheduled timeframe. The consultants brought a conventional process improvement and business value orientation to the project, based on the Lean Six Sigma methodology. Work practices provided a substantive complement to this. Key to the study, and typical of a work practice approach, is that all the interviews and observations took place in-context. That meant following the people into offices, maintenance bays, and break rooms, wherever the work took place. This approach provided insights about key issues with current practice and suggestions for ways to support change within this organizational culture. The insights obtained using the ethnographic approach enabled the consultants to articulate the value and impact of the proposed solution in customer-relevant terms. Work practice representations helped the XGS implementation team and the customer envision proposed changes and their implications from a sociotechnical perspective. Subsequent feedback from the participants indicated that the solution addressed all the key issues identified by study participants and ultimately met the needs of the maintenance staff and management. The collaborative team formed between consultants and researchers resulted in a real-time cross-training experience. Although the researchers participated in previous XGS projects, this project had significant time constraints and high visibility in the customer site. The work practice study, solution design and deployment in the customer site had to be complete in less than four months. At the same time, the researchers were coaching the consultants in field methods and data analysis techniques. This resulted in the transfer of knowledge about work practice methods to the consultants. The experience affected the researchers as well, as they developed a keener appreciation for the accelerated pace, pressures, and constraints that impact consulting practices. This experience has been put to use as the Research organization, at the request and sponsorship of XGS, has implemented a work practice training and mentoring program for consultants so they can continue to apply appropriate aspects of work practice methods in their consulting engagements. (See Plurkowski et al., Chapter 19, this volume for a description of this program.)

16

Transforming Information System Design Enabling Users to Design Yutaka Yamauchi

The IT industry used to be highly profitable with proprietary technologies. Over the years, however, as computing technologies have become open standard and modularized and largely commoditized through severe competition, the profits declined. Vendors sought profits from software services. However, large-scale software development, particularly custom software for a single client, is risky and costly. In this environment, IT vendors have been struggling to find ways to reduce risks, improve productivity, and secure profits. Faced with these challenges, a large IT vendor and PARC started an ethnographic project to understand ways to become more competitive. In the beginning, the IT vendor asked us to study upstream activities in which engineers interact with their customers to develop specifications of the system. The relationships among the different players are summarized in Figure 16.1. There was recognition within the client organization – and in the industry – that this upstream phase was the source of most problems, which would often manifest themselves only in the later phases. For instance, when requirements are not defined precisely, the resulting system deviates from what users wanted and a major rework is needed. Rework simply adds unexpected work to project members’ already tightly planned schedule. Rework causes more rework and worsens the situation. Members have to work until midnight or all night and over the weekend to get the work done. The schedule slips. Members burn out. This was a familiar story with the IT vendor. All upstream work is difficult, because each client’s situation is always unique. And whereas some clients have a clear idea of what the system should be like, many others have little idea. The system needs to fit in the client’s specific environment. Therefore, few standard processes and rules exist to guide the upstream work – only some abstract guidelines. Without 299

300

Making Work Visible

Figure 16.1 Players in the project

the proper understanding of how upstream work is done, managers often try to contain the problem rather than to solve it (Vinkhuyzen and Ikeya, this volume). New standard processes and measures have been introduced to help understand the complexity of upstream work. For example, new check points have been added to the process so that managers can intervene when they see a problem. A third party quality inspection team can be brought in to ensure quality. Yet these efforts do not necessarily solve the real problem. Project members are required to provide detailed reports frequently. This can provide peace of mind for the managers, but project members face an even greater workload. Managers also have a tendency to attribute success or failure in upstream work to members’ personality. One manager said, for instance, that he came to the conclusion that skills for this work were innate and not teachable. The organization needed a better way to understand upstream work. The fieldwork revealed a myth held by many members of the IT vendor – that the IT vendor’s engineers could somehow determine the system’s design based on requirements the client’s users describe. The problem was two fold. First, as prior studies have repeatedly shown, users could not easily specify their requirements (Greenbaum and Kyng, 1991). If engineers accepted what users described and implemented it exactly, the resulting system would be too costly and complex. Second, a greater problem was that as members of the client organization that is paying the IT vendor, users were in the dominating position vis-à-vis the engineers. Users dictated the design and engineers had to listen – a reverse situation of participatory design where users are often sequestered from design decision-making (Greenbaum and Kyng, 1991; Mumford, 1983; Suchman, this volume). Engineers had little leverage to alter users’ demands. This power dynamic combined with the users’ unclear requirements made it almost impossible for engineers to determine a design. Further fieldwork revealed that a few users could actually learn to design a system within a short period of time. They explore various combinations of detailed system features and come to propose a design that engineers could not come up with. Once users started contributing to the design, many of the problems were overcome. However, this user learning did not happen

Transforming Information System Design

301

easily. This chapter describes the process by which users begin to learn how to design. By analyzing the processes by which users were learning how to design, we could build a new model of engineers’ work practices. First, I describe three examples of users’ learning in detail. Then I discuss the methods and tools we developed for this new model and how we subsequently designed and delivered a training program to engineers.

Field Studies Challenges in Users’ Designing Designing requires a special orientation, or what is often called design attitude (Boland and Collopy, 2004) or designerly ways of knowing (Cross, 2006), which is different from the orientation that users are typically familiar with (Simon, 1968). There is abundant discussion in the design education literature on the difficulty of teaching how to design (Cross, 2006; Schön, 1987). This is largely because the skills for design are not well defined and no systematic teaching method exists. This difficulty is more marked for system users than for design students because users do not have a strong intent to learn and become designers. Users of information systems such as accounting, production planning, and logistics systems are used for routine operations in which they use the same screen repeatedly to enter and obtain data. Creating something new is not part of their everyday work. Designers are said to have a solution-oriented attitude as opposed to problem-oriented attitude (Cross, 2006; Lawson, 2006). They use solutions to understand problems and often transform these problems. Of course, designers do not begin without understanding the problem. They try to understand the problem as much as they can but design does not come naturally from that understanding. Because design is a practice of form giving, or creating something new, there is no single best solution to the problem. Typically users are familiar with identifying and analyzing problems rather than designing an artifact. For instance, accountants, users of an accounting system, are responsible for understanding an accounting problem by means of detailed analysis. They cannot start designing something, for instance, different accounting processes, without a clear understanding first. Similarly, this design attitude requires a perspective that considers the design of the whole while examining all the detailed parts (Brooks, 1995). This is because a design problem is “figurally complex”– a change in one part changes the whole design (Schön, 1990). The whole requires some type of guiding vision that consistently dictates the design of its parts. If people

302

Making Work Visible

take a problem-oriented approach, they typically design a solution to address each problem. Consequently, the resulting design will consist of various heterogeneous solutions each of which solves a different problem. This type of design becomes complicated and large, and therefore difficult for users to understand, employ, manage, and maintain. Designers are good at deriving such a guiding vision early on and try to solve a number of problems under that framework (Darke, 1979). Typically users of an information system accomplished their own parts of the work, defined as parts of the larger organizational routines or standard operating procedures. It is difficult for them to think of the whole, beyond their immediate work processes. This is compounded by the fact that they are so familiar with their current work practices that they cannot easily think in alternative ways. Envisioning the whole requires revisiting implicit assumptions and reframing problems in a new light; users invested in their own practices find this difficult to do (Carlile, 2002). Ethnography helped us understand all these issues that users face in learning how to design an information system. In one project, we conducted longitudinal studies of requirements development and basic design (Yamauchi, 2009). In this project, the IT vendors’ engineers worked with the client in a service industry to develop a new accounting system. All the users that participated in design were accountants. The requirements development and basic design phase lasted for three months. My colleague and I started observations in the fifth week but could review audio recordings of previous meetings. We then followed many meetings between users and engineers in which they developed the high-level design of the system and some meetings among engineers only in preparation for the meetings with users. We observed forty two meetings between the users and the engineers and audio recorded fifty seven hours of data. I also observed twenty one meetings among engineers only and recorded twenty one hours of data. We also obtained copies of documents used. We then analyzed these audio recordings and documents in detail. The data showed that users typically have no interest in designing. Users do not want to be designers and after the project is over would go back to their own organization and continue their work as a user. On top of that, users, like everybody else, are typically busy with their own work and see design as extra work that does not add any direct value to their own work (Grudin, 1993). A typical user says to engineers: “I will explain to you what we want; please build the system to realize it.” Because users belong to the client organization that hired the IT vendor, most users think of design as the vendor’s job in exchange for payment. The payer–payee relationship

Transforming Information System Design

303

tilts the power balance to the users. Engineers of the IT vendor could not easily teach users how to design when users showed no interest. Inevitably, engineers could not easily realize what users wanted and proposed alternative designs that required users to alter their work practices. Users typically refused such design proposals and insisted on their original requirements. For instance, users said, “We can’t do without [the feature] . . . No way. Please add it,” and “It’s a matter of principle . . . at least we need [the features].” Engineers had little leverage to overthrow these persistent demands. Engineers had to give in and agreed to add these features.

Case 1: Dominating Requests for Breakdown Remarks Fields in Accounting Slips Users demanded a particular design based on their understanding of the problem. In determining design this way, the whole was not touched on, only the parts. For instance, users demanded a free-text remark field for each of the breakdowns in accounting slips. The system that they were going to use did not have this field, only one remark field for the entire slip. As users insisted, engineers decided to include the breakdown remarks fields. This is a design that was derived directly from the understanding of the problem, a problemoriented approach. The reason why the new system did not have breakdown remarks was that the new system was based on real-time accounting where a number of slips are sent to the accounting depart frequently. In the old system, users had to prepare paper slips and therefore tended to accumulate items on one slip. Therefore, one slip contained many breakdowns, requiring remarks for each. In the new system, most accounting slips contained one or two breakdowns. The point is not that the need for breakdown remarks is unwarranted. Rather, the point is that users and engineers did not discuss the concept of the new system – that is, the whole – in determining the design. Faced with persistent demand, engineers could not discuss this in a productive manner. Unfortunately, many of the system features are designed in this manner. Whereas it is encouraging that users participate in design discussion and have certain power to dictate design, the users take a problem-oriented, piecemeal, local design approach as opposed to one that is solution oriented and holistic.

Case 2: Working around Problems with the Accounting Slip Key In the process of working with the engineers, users learned more about design. For instance, the users had used a unique identification number

304

Making Work Visible

called “key” to locate accounting slips. If they needed to refer to a slip, they could type the key in the system. Engineers and users soon found that the new system did not have the same identification number. Initially they thought the journal entry number provided by the new system was closest to the key. After some more discussion a few weeks later, they realized that the journal entry numbers are issued only after the data are posted in the general ledger; this was too late because users wanted to use the key to deal with slips well before posting to the ledger. Users started to explore other possible ways to realize the key using some of the numbers the new system offered. One user representative was particularly active in determining the design, learning details of the system features, and proposing various different combinations of the features. The users and engineers arrived at the design of using the process number of the new system in combination with some other workarounds. This example shows how users can be pulled into design activities. Users were actively exploring alternative designs by examining details of the system features; a significant achievement given that they had no idea about system features in the beginning. Yet, we realized that they were still taking the problem-oriented approach, trying to work around the fixed problem. The new system is designed in such a way that users do not even have to think about the numbers; they can locate slips by means of various search criteria such as employee name, date, slip type, and so on. What became clear is that users had been using the key because of the character-based terminal of the existing mainframe system. That is, users did not use a mouse to click on a slip but typed in a code to bring up a certain slip. Yet, the new system would be equipped with an advanced graphical user interface in which users could simply click. For advanced users like accountants, who memorize the codes, it is faster to type in code rather than to click. But the new system was supposed to be used by all the employees whereas the old system was used only in the accounting department. Furthermore, the existing system had a small display on which only a limited amount of information could be displayed. Therefore, detailed information (particularly lists and tables) was printed on paper. A key was needed to link the information on paper to information in the system. The new system was based on the concept of working without paper. This accounting slip key case shows how difficult it is for users to come out of their familiar world and adopt the solution-focused, holistic approach. Users were actively designing by considering possible combinations of system configurations. Nonetheless, the resulting design, that is, use of the process number with some workarounds, was a direct solution to a

Transforming Information System Design

305

fixed understanding of the problem. There was little holistic vision that guided their design.

Case 3: Achieving Holistic Design for Real-time Accounting Users in the accounting department wanted to print the table of all the slip data (journal table sheet) at certain intervals so that they could quickly go through the slips to check for incorrect information. The new system could not print this table. The users, however, repeatedly made clear that this table was necessary. The engineers were unable to convince the users otherwise. Three weeks later, however, the user representative suddenly said that they had changed their minds after some internal discussions, and now they were willing to drop this requirement. Instead she said they wanted to print out another list: “Actually now we are thinking of changing the policy, we are considering giving up the journal table sheets. Then, we had the cover sheet, right? We wonder how much we can tune up the cover sheet.” Although users had been verifying and modifying slips before posting them to the general ledger, they conceded and suggested that they could check the slips after the entries were posted to the ledger, modifying any incorrect data at that point. A minor modification would enable the system to print the cover sheet. Further, the suggested “policy” change was crucial for the realization of real-time accounting. Because the junior accountant gave up the before-the-fact verification, the data is processed in real-time. The accountant presented not only her grasp of the technical solution but also the high-level policy. The parts of the design had a coherent whole. She had learned about the cover sheet before and tried to use them for a different purpose. She took a solution-focused approach by using the cover sheet to redefine the whole work process, thereby redefining the problem and dropping the need for the journal table sheet. With these cases, we could see the details of the learning process. This gave us insight to why the problem arose – lack of users’ learning – and how we could go about solving it – how to facilitate users’ learning. We then studied three other projects in varied depth and refined the idea even further. For instance, in one project, engineers set up a packaged system with sample data and let users play with it. Then, they asked users to write down their understanding of the workflow. Although the users initially complained about some missing features that they had been using in the old system, they gradually learned to design alternative processes in three months. Eventually, users learned so much that they were able to consider technical details. For instance, one user suggested that if they used a “virtual

306

Making Work Visible

warehouse,” managing inventory data would be much simpler and less costly. Typically, users can only think about a physical warehouse; the virtual warehouse is not a familiar concept for them. On the other hand, we have seen that in many other projects users had difficulty learning to design; the typical scenario was engineers designed for passive, receptive users who simply agreed to the design the engineers proposed. For example, in the project we saw above, the other subteams were not as successful in guiding users’ learning as the user representative. When I shared the idea of users’ learning to design with the engineers, many of them were disturbed, because they thought design was their job. They all agreed that users’ participation was necessary, but users designing their own system was seen as either impossible or a bad idea. These findings convinced us that users can – and should – learn to design their own system even in a short period of time. Inspired partly by von Hippel’s (2005) notion of user innovation, we told our client that in ten years users would be able to design their own systems with various innovative tools. Vendors’ business would lose its raison-d’être just as photography development services largely disappeared when users started to take photographs with digital cameras and print them with inkjet printers. In fact, we heard actual examples of user design in enterprise system development. One IT vendor offered packaged solutions that did not require IT professionals for customization and installation. Users could customize the system by changing parameters intuitively. Another vendor provided clients with a screen design tool that automatically produced script code that processed data, checked for errors, and interacted with the database. Our task now was to put user designing into engineering practice.

Designing a Solution After the fieldwork revealed the detailed understanding of how users learned to design, we started to explore ways to facilitate this learning. Our client, that is, the IT vendor, requested that we create a method that could be used anywhere in the organization. This method had to be usable within the engineers’ current environment – we could not propose any of the heavy participatory design techniques such as future workshops, cooperative prototyping, and design games (e.g., Bodker et al., 2004; Schuler and Namioka, 1993). We opted not to propose a completely new set of documents and tools, and instead refined the ones they used. The method consisted of practical techniques and tangible tools. In order to transform engineers’ everyday practices, we determined that it was necessary to engender a shift in perspective. To do this, we developed

307

Transforming Information System Design Table 16.1 Examples of juxtaposition Juxtaposition Nurturing documents

Concept bugs

A concept familiar to engineers

A concept for alternative practice

Creating documents: Engineers create documents for users to describe the world objectively. Users then read the documents and agree to them. Bugs: Engineers are concerned about defects in code or specification.

Nurturing users: Users learn to design by creating their own documents and develop the documents into their own operation manual. Holistic concept: Even if there is no defect, design can be inconsistent with the holistic concept of the system.

a practical means we called “juxtaposition.” We juxtaposed that which engineers were implicitly familiar with – what the new practices were going to be, as shown in Table 16.1. For instance, system engineers, employing a traditional waterfall methodology to develop a large-scale system in a large team, used design documents intensively. Their implicit model was to create perfect documents that defined everything in detail so that developers in the downstream could understand them in a self-evident fashion. It is not surprising that this model did not work. In fact, engineers themselves acknowledged the difficulty in doing this although they still treated it as an ideal goal to strive for. We then juxtaposed the familiar documents with an unfamiliar concept that users themselves create those documents: “Nurturing documents” instead of “creating documents.” This juxtaposition referred to the idea that engineers needed to facilitate users’ learning and in particular to guide users in their own creation of documents instead of creating documents for them. Then, users would continue developing these documents as they would learn to design and eventually turn them into their own operation manual. The engineers were steadfast about the idea that they “create documents” to represent the world perfectly. We turned this implicit model into a different one by juxtaposing an unfamiliar concept of “nurturing.” Another juxtaposition was “concept bugs.” The notion of bugs was familiar to engineers. Yet, they treated bugs only in the sense of defects in the code or in specifications that were not yet turned into code. As we saw previously, we found fundamental inconsistency in concepts before design is specified; that is to say, parts without whole. For instance, the use of key was not consistent with the new system’s advanced graphical user interface. The new accounting system was to be distributed throughout the organization

308

Making Work Visible

whereas the old system was only used by the accounting department. Accountants were used to the key and could represent much information in its digits. Yet, users of the new system would not be accountants nor would they be familiar with the numbering scheme. Similarly, journal table sheets were not consistent with the overall real-time accounting concept. Before, the fact checking meant batch processing of the data. Even if the design is defined correctly without bugs, the design can be inconsistent with the holistic concept of the system. To bring this idea forward, we used the familiar notion of “bug” and juxtaposed it with an unfamiliar notion of “concept.” The conceptual integrity achieved through holistic design was made clear through this juxtaposition. These juxtapositions were instrumental to highlighting the implicit assumptions practitioners make and to transforming them in a new light. The difficulty in transforming practices lies in the fact that practices are deeply ingrained. Juxtapositions often disturbed practitioners. For instance, some engineers became defensive by the notion of concept bugs, reiterating that “bug” was a special technical term; whereas the point was to reorient these engineers to think that these concept bugs were bugs in that they do not contribute to or even get in the way of the customers’ ultimate goals. It was in some cases not easy to get the point across, but the emotional disturbances indicated that the engineers’ implicit assumptions were being brought into examination. Practical Techniques Throughout the fieldwork, we observed a number of practical techniques that engineers used. For instance, when engineers had a hard time encouraging user representatives to take part in the design work, they instigated a competition among users by visualizing and listing every user’s progress. As users could see other users making more contributions, they felt compelled to catch up. In this way, engineers did not have to annoy users by telling them what they should do repeatedly. Another technique was to designate an owner for each requirement. When a user insists on a requirement, the user becomes its owner. The owner was always made explicit, even in meetings where the customer’s high-level managers participated. Another practical technique was to nurture documents. One tool was a simple alternation of a document format such that users could participate in the creation of documents. One variation was a questionnaire embedded within documents. Users who participate in meetings are supposed to fill in the questionnaire while reading the documents. Another variant was

Transforming Information System Design

309

documents that had blank areas; users were supposed to fill in the blanks and thereby create a smaller part of the design. In one case we observed, an engineer asked user representatives to draw a screen image that they wanted for monitoring production processes in a factory. The engineers did this because they were not clear which variables users wanted at which timing and in what format. If the user representatives could have drawn what they wanted, it would have been clearer. Yet, when asked to draw a picture from scratch, users did not even know where to start. This led us to develop this idea of documents with blanks. We developed various document formats for screen layouts, workflow diagrams, and decision-making trees. We emphasized the practical nature of these techniques as opposed to some abstract theory. These kinds of practical techniques were considered important for the engineers because they felt worn out trying to lead unwilling users to participate more. Simply saying that engineers should encourage users to participate more would put the engineers on the defense without any resources. Tangible Tools Tools were developed in order to put these practical techniques into action. Similar to Muller (1991), we chose to design paper-based tools rather than computer-based ones for several reasons. Engineers use paper documents intensively. Typically meetings are structured by documents. Paper-based tools create interactive media for facilitating users’ learning. We also had to limit ourselves to simple tools in order to prove the concept. In fact, a simple, web-based prototype was developed using Adobe Flex for this purpose. This was a general purpose tool on which multiple users could easily draw workflow diagrams, create screen layouts and specify decisionmaking rules. Although simple in design, this tool was still too complicated for users, as they typically did not have access to computers in meetings, let alone a reliable network. In these design sessions, mostly paper documents and one computer connected to a projector were used. The paper tool consists of the base sheet and stickers to put on the base sheet. Users could place the stickers and write on the sheet. Initially this tool was developed for the creation of workflow. Users could easily design workflow by putting the stickers on the base sheet. Constraints were already inscribed into the base sheet and users could only design some areas of workflow. We created stickers of standard notions of workflow diagrams. We invented a two-layer sticker as shown in Figure 16.2; the first layer contained the detachable adhesive characteristic of sticky notes. This way,

310

Making Work Visible

Figure 16.2 Design of the sticker

users could easily move the stickers around until they determined the design. The second layer had a permanent adhesive. Once users determine the design, they commit to the design by pasting the stickers permanently. This simulated a ritual of “making a commitment” to their own design because they could not physically detach and reposition the pasted stickers. With this tool, there was psychological pressure for users to finalize the design and feel responsible for the design, as opposed to reading documents engineers created and agreeing to them. Users were compelled to think through the details. We then created screen design tools in the same way. We created similar tools for planning the project. Users often had no idea what they would have to do during the course of the system development project. We created stickers listing necessary steps or work breakdown structures for the basic design phase. The idea was to have users and engineers create the project plan together so that users could also feel ownership of the process rather than feeling as if they were being told what to do. Another tool was to create a project staffing model. Again the purpose was to help users understand who should participate in the project and how much. This tool embodied various constraints so that the client could select appropriate user representatives, write down their name, and specify the roles they would play.

Training Program We chose to implement an organization wide training program for deploying the methodology broadly in the organization. The training session was about two hours. We presented the new model of work practice, juxtapositions, tangible tools, and practical techniques as well as various episodes that

Transforming Information System Design

311

we collected from fieldwork. We designed an interactive training so that participants could reflect on their projects and ponder how to make use of the methodology. We organized exercises so that they could create their own paper-based tools in the particular context of their projects. In some sessions, we used the paper-based tools in the exercise so that participants could create workflow diagrams of the project. Interactive discussion led to various new methods and tools and improvements to existing ones. We planned four workshops for the pilot, but added four more by request; 101 people participated in all. One session was with the information system departments of several customers. Overall, the workshops were received well. A number of participants said that they would use the methods and tools in their projects. In the first three sessions, when we were still trying to find a good way to organize the workshop, 83.3% of the participants said that they would recommend the workshop to others. After the fourth session 89.4% recommended it. These evaluations were higher than we had expected given that many participants said that the training was not for everyone. Furthermore, although we started this effort in a bottom up manner, a business group decided to adopt the training workshop as its official process. A manager in this business group even took over as instructor for the training.

Conclusion Ethnography was key to this study in two ways. First, ethnography enabled us to understand the problem in a concrete, clear manner. Upstream activities are difficult to grasp and managers tried to introduce several external constraints to the work to achieve better visibility and control. But, this only contained problems; it did not solve them. Ethnography revealed not only the importance of users’ learning to design but also the concrete processes by which users learn. Second, ethnography was crucial for the development of the methods and tools for new work practices. Whereas there are a number of methodologies that provide guidance to manage system design, most remain highly abstract. Ethnography enabled us to create a domain- and site-specific methodology grounded in what is important to the practitioners.

17

Rethinking How Projects Are Managed Meeting Communication across the Organizational Hierarchy Erik Vinkhuyzen and Nozomi Ikeya

In the summer of 2003 a team of fieldworkers set out to observe the development of a large IT system for a manufacturing company. The company had recently been spun off from a much larger organization, and whereas it was still using the IT systems of the organization it had been part of, the license for those systems would soon expire. The IT vendor had been selected to build an entirely new enterprise information system. It was a big project so in order to circumscribe the observational work the fieldworkers focused on the team responsible for the sales subsystem. From the start there was trouble. The sales department of the manufacturing organization had dispatched a small group to work with the IT vendor to establish the requirements for the system and they insisted that the new system must support their current work flow, even though the software package their management had chosen in the early negotiations of the project was expressly designed for a different workflow. So the sales department’s representatives made numerous requests to have the software package altered in fundamental ways. Furthermore, the sales subsystem team from the IT vendor was relatively inexperienced. Their leader had been selected because of his success in a previous project, but he was familiar neither with typical sales work processes nor with the software package. The junior person knew about the software package, but not sufficiently to answer the customer’s many detailed questions. Another junior person had only two years experience and was of little help. To make matters worse, a few weeks into the project the team leader was given additional responsibilities; he had to oversee another part of the system for which no package had been selected yet. He spent every day meeting with different members of the manufacturing customer, and when he came back to his own office, he had to meet with other IT people to talk about the integration of various packages into a unified system. He had no time to review the documents his junior staff had created for the customer, let alone time to work on his own tasks. Overwhelmed, he began staying overnight at the office. 312

Rethinking How Projects Are Managed

313

Yet despite the long hours, his team was not making much progress and the customer’s sales department representatives were getting increasingly frustrated with the quality of the project work. Ultimately, the trouble in the sales subsystem caused the entire project to be delayed. What surprised us most was how few people inside our client – the IT vendor – had been aware of the trouble with the sales team until it was too late. To us it had been obvious that the team had struggled and that the meetings with the customer had not gone well. And we knew that the project manager had been wary of the sales team’s lack of experience from the outset of the project, and that he had been aware of the problems. Curiously, regular communication meetings – so-called Project Assessment (PA) meetings in which the project manager meets with the senior managers to discuss the project’s progress – had been set up; they met every few weeks. The PA meetings had been especially created by the organization to do “risk management,” that is, to raise awareness about potential problems in projects and to attack them as early as possible. However, in our case no measures had been taken by the senior managers until it was much too late. We had not attended the PA meetings, but started to wonder, what goes on in these PA meetings? Why had the project’s troubles gone undetected for so long? In this chapter we report on a research project we undertook to address these questions.

Background and Method The software business group of the large IT company we studied is organized by industry, but the work is conducted in “projects.” Its employees work with clients on contracts that last from a few months to many years. There are a large number of projects to manage. Higher-level managers are responsible for divisions with many subordinates. Project managers (hereafter PMs) oversee the work as it is being done in the field with subordinates from both their own organization and subsidiaries. In some divisions, especially those with a large number of projects, even system engineers (SEs) with only a few years of experience often manage many subcontractor employees inside projects. The project teams often work at or nearby the customer’s offices, so that the project members are away from their own organization most of the time. They do not have an office at their own company; only senior managers have office space. Once we had decided to make the PA meetings the focus of our investigation, we approached the head of a division we already had a relationship with and he agreed to let us attend his PA meetings. We were allowed to observe and videotape the proceedings of fifteen PA meetings over a period

314

Making Work Visible

of several months. We transcribed the audio and analyzed the transcripts of these meetings. Throughout, we also conducted many interviews with the PA meeting participants in which we would often show them brief clips of video from the meetings in which they had participated. We audio recorded and transcribed these interviews, as well. Additionally, we talked with many other members from different divisions of the organization, asking them their opinion of the PA meetings to gauge whether our findings were common in other parts of the organization. The culmination of our work was a brief video report to a large group of managers. Their response was quite favorable; clearly we had uncovered something fundamental about their organization that they realized needed to be altered.

Observable Features of the Meeting In most projects, PA meetings are held every month; the PA meetings are mandatory for projects above a certain size. The participants to the PA meeting are usually the leadership of the project (one to three members), the sales person(s) responsible for the account, the PM’s direct report, senior managers (both from the sales and system engineering organization), and a note taker. When projects exceed a certain size, or have particularly large problems, members from the project assessment organization – a group dedicated to investigate risky projects that reports directly to the president – attend as well. The meetings can be quite small and intimate, but in larger projects, and especially when there is trouble, they are often large with many stakeholders present. Like any meeting, the PA meeting begins after the people have arrived and taken a seat and the premeeting chitchat ends. Documents are distributed by the project team. Then, the PM begins by listing the documents that all participants have received to ensure that everybody has a complete set. The meeting officially starts when the PM goes through these documents in detail. The senior managers (SMs) listen and whenever they have questions or concerns, they interrupt the PM’s reporting. The PA meeting has a modified turn taking system, characteristic of institutional talk (Heritage and Drew, 1992). When the project team does its reporting, turn taking (Schegloff and Jefferson, 1974) is suspended. However, whenever the reporting is interrupted by a question or comment and a discussion ensues, ordinary turn-taking resumes. The reporting can only recommence after the discussion is closed, which is usually achieved by one of the SMs saying “okay.” During the discussion there are frequent occasions when the SMs tell the PM what he ought to do. Whenever the SMs advise the PM, the note

315

Rethinking How Projects Are Managed Table 17.1 Excerpt from the List of Action Items No. Action Items 1

2

3

Status

Person in Date of charge execution

Reconsider where to put the Sales side is negotiating Fukui with the customer. machine with security in –>PIF will be transferred the project room around 11/25 when the network is set up –>Server will be transferred to PIF on 11/11 and 11/25 Recheck what kind of In progress Mukai measures are taken for risks with regards to the system infrastructure ...

10/24

taker writes down what the SM suggested in a list of action items (see Table 17.1), and at the end of the meeting the note taker reviews all the action items he wrote down, giving the SMs a chance to make any corrections or to add any items the note taker may have missed. This list of action items is in some regards the meeting’s most tangible outcome. The PA meeting is part of a series of such meetings and the PM prepares for a PA meeting by looking at the documents from the previous meeting and updating them. SMs often compare these updated documents with the ones from the last meeting to see how they have been altered. Typically, the meeting is closed after the next meeting has been scheduled. The meeting participants leave, although it is common for informal side conversations to take place at the end of the meeting.

Project Managers Report that Everything Is Under Control The PM’s report is an important part of the PA meeting as the PM is the most knowledgeable about the project. Therefore how the PM presents the project’s progress since the last meeting, the project’s status, and the project’s problems, sets the framework for the subsequent discussion. The following excerpt is representative of the beginning of the PA meeting. The PM starts the meeting by reporting on the action items from the previous meeting (Excerpt 17.1). What is noticeable is how the PM presents the progress on the action items in such a way as to demonstrate that the project is under control. Progress has

316

Making Work Visible

Excerpt 17.1 Project Manager reports on action items

01 PM: 02 03 04 05 06 07 08 09 10 11 12 13 14 15

I now want to start by explaining the PA meeting documents. First, as we had fourteen items in the List of Action Items from the previous meetings, I would like to start with the status of the items. First of all, about the investigation of putting machines in the project room for security purposes point of view, Fukui has done it, and we decided that we will transfer all the machines to the site during the period of November 1 to 25, and we are proceeding according to the schedule. As for the second item, on infrastructure related risks, we were supposed to do it by the next meeting (inaudible) and Mukai has been in the process of executing this, is making progress everyday, this is to be continued . . .

been made in several ways: people are assigned to do each task, they are proceeding in accordance with the schedule, and they are executing planned procedures. With regards to the first action item, the PM reports that a decision has been made to transfer the machines, the time frame for this task has been fixed and the only remaining thing to do is to execute according to the plan. By presenting all these steps that have been taken, the PM demonstrates his control over a concern the SMs had expressed in the previous PA meeting. The status of the second action item is shown as “in progress” in the document (see Table 17.1). The PM emphasizes that the person assigned, Mukai, is “making progress everyday (line 13–15).” However, note that they were supposed to finish “by the next meeting” (line 12–13), that is to say, this current meeting. Instead of stating that they are behind schedule, the PM stresses that steady progress has been made, that is, that there is nothing to worry about and by inference, that things are under control.

Senior Managers Scrutinize the Project Manger’s Report While the PM skillfully organizes the report to demonstrate that progress is being made on the project, SMs scrutinize the PM’s report. In Excerpt 17.2, the SMs asked for specifics and details regarding the cost estimates for additional work the customer requested in order to determine what the project team may have overlooked or neglected to do.

Rethinking How Projects Are Managed

317

Excerpt 17.2 Senior Manager scrutinizes project’s progress

01 SM: 02 03 PM: 04 05 06 07 08 09 10 11 SM1: 12 13 PM: 14 15 16 SM1: 16 17 18

So, have you presented it ((the cost estimate)) yet? Well, regarding the items we have not presented cost estimates which are probably not included in the list I mentioned earlier are, well, the linkage with Ultimate, sorry, I am going to bring this (to the customer) as well, as they seem to want to start it from July next year, so, in fact, the linkage with Package Antilope is yet to be determined How much were the estimates you have presented so far? Well, about 60 to 70 million ((yen)) ((pause)) ((other talk)) . . . In a nutshell, regardless of the money negotiations, is the work already going forward? Or will you start the work after you get the (contract)?

After learning that the PM has started to negotiate over additional payment for additional work, the SM asks pointed questions to find out what the current status is of the payment negotiation. How much money is involved? and especially, has the additional work already begun? When they find out that the work has already started, the SM (SM3) and a sales person (SLS) become concerned and ask the PM when he will be able to secure agreement from the customer for the payment of this additional work, as shown in Excerpt 17.3.

Excerpt 17.3 Senior Manager questions payment negotiations

01 SM3: 02 SLS: 03 04 PM: 05 06 SLS:

Then when will it be decided? If it is possible, well, before the coming meeting with people in charge. On Friday this week, I will meet with the customer and check it then. If possible, we should make it clear ((inaudible))

318 07 SM3: 08 PM: 09 SM3: 10 PM: 11 SM3: 12 13 PM: 14

Making Work Visible Friday? Er, on Thursday. Thursday. The second. On the second, you do that? (Next week). You will get agreement from the customer? In the management meeting. We will confirm it on that occasion.

Pushed by SM3, the PM and the sales person specify the date they will talk to the customer and get an agreement about the additional cost. Through their pointed questioning (lines 01, 11–12) the SM ensures that timely action is taken and holds the project team accountable for finalizing the negotiations with the customer. During this interchange, the note taker writes down the following action item: “present a cost estimate for Ultimate as soon as possible.” Excerpts 17.1–3 are exemplars of PM and SM practices during the PA meetings we observed. Whereas the PMs presented their projects as “under control,” the SMs listened skeptically to the PM’s reports and asked pointed questions to determine the actual status of some of the work and did so quite aggressively. When they felt something had been mishandled, SMs would, at times, get quite angry with the PM. SMs acted with great authority and would tell the project team unequivocally what needed to be done, and the note taker would capture what they said in the action items. We began to feel that the PMs’ practice of presenting the project as “under control” was at least in part aimed at avoiding harsh comments from senior management. One PM confirmed our impression: “The PA meeting is not an occasion where we should raise all problems we have. This is an occasion where we state what we perceive the issues are and how we would like to deal with them.”

So whenever PMs presented problems they always did so with candidate solutions. SMs could either approve the proposed actions or suggest alternatives, but in either case, the PM dictated how the discussion was framed. Another way PMs tried to keep control over the discussion was by omitting difficult issues entirely. As one PM commented: “As it is a reporting session, we hide what we do not want to tell them.”

Rethinking How Projects Are Managed

319

The interactional dynamic of the meeting, then, was one of two opposing parties. On the one side was the project team, playing defense – presenting the project as under control. On the other side the SMs played offense by trying to uncover what hidden problems there were, and what the project team neglected to tell them. The SMs’ desire to uncover hidden trouble is understandable, but in many cases the advice the SMs give is not perceived to be of great value to the project team. As one PM told us: “One wonders whether it [PA meeting] is an occasion when the project manager can really ask for advice. The senior managers hold the PA meeting just because they expect that it is less likely that a project will fail as long as they have PA meetings, but in fact, they may not be able to provide good advice. And, if that repeatedly happens, the project manager may just give formulaic reports.”

The organization’s original intention of introducing this project assessment meeting was for both SMs and the PM to discuss the project openly in order to tackle problems as early as possible. So it was ironic that neither side was satisfied with how the meeting was organized, and felt that they were unable to discuss the project openly in the meeting. What we found in the PA meetings was an interactional dynamic in which the project team was unwilling to present problems openly while SMs were aggressively trying to uncover problems. As we had conducted other fieldwork in the organization, we could see that the PA meeting was a reflection of the organization’s hierarchical culture, in which younger members were expected to obey orders, and rather than challenge SMs’ views openly, they would avoid discussing certain problems altogether to avoid SMs’ “unnecessary anger.”

Addressing the Meeting Design Problem We presented our findings to PMs and SMs alike; this was a delicate matter as we had to hold up a mirror to their PA meeting practices as they really were. Both stakeholders agreed with our analysis, and they supported our redesign of their current PA meeting process to try and improve its outcomes. The question was, how to change the organization’s culture? We felt that rather than take the culture head on, the most effective way would be to implement simple, concrete measures. We were convinced that one of the factors that contributed to the dynamic described above was the very way in

320

Making Work Visible

which the PA meeting was organized and executed. Whereas the PA meeting had an analyzable interactional structure, the structure had not been deliberately designed to achieve certain goals. The only discernable structure was the PM’s report interrupted by SMs’ scrutinizing questions. With the implicit meeting structure revealed, we could use it to redesign it to deliberately address its shortcomings. Based on our findings and the literature on effective meetings (Means & Adams, 2005; Kayser, 1995; Schwarz, 2002) we came up with four concrete interventions to improve the PA meeting. 1. Advice Segment The first thing we wanted to create was an officially sanctioned time inside the PA meeting in which the project team could ask the SMs for advice. It would be a discussion in which they could present problems without sanction and, especially, without candidate solutions. 2. An Official Agenda In order to ensure that the advice segment became a distinct and permanent part of the PA meeting, it had to be more structured and time had to be more carefully managed. We advocated that the PA meetings have an officially sanctioned agenda with time allotments as well as desired outcomes for every activity, in order to make the discussion more productive. 3. Premeeting Activities In order to implement the agenda successfully, we realized that it could not be created unilaterally. We had noticed that SMs were often dissatisfied with the documents PMs brought to the PA meeting, so we designed a pre-PA meeting process in which the PM would send the SM the agenda as well as the status report documents at least a day ahead of the PA meeting. This gave SMs the opportunity to ask for alterations and additional information. Furthermore, we asked them to discuss the agenda before the meeting. Thus both parties would be involved in organizing the meeting ahead of time as a way to ensure that the agenda was sanctioned and could be adhered to. After the SM and the PM reach consensus about the agenda, the PM shares the agenda with other stakeholders and encouraged them to contribute. We hoped this would encourage other participants to bring their topics to the meeting as well.

Rethinking How Projects Are Managed

321

4. Meeting Facilitation Of course the whole plan could easily fall apart unless the meeting itself was managed with some discipline. Therefore we suggested a neutral facilitator be present who could watch the clock and mediate if the discussions ran too far afield. We asked that the agenda be followed and the time allotment for each item be honored.

Implementation and Evaluation When we presented these suggestions to the members of the organization, they were quite intrigued, but they asked us to facilitate the meetings ourselves. We reluctantly agreed and worked closely with the SMs and PMs to prepare the agendas. When we were constructing the agenda with the PMs, some of the younger PMs found it difficult to put topics under the new advice segment. It was only when we repeatedly explained that the SMs had agreed that no solutions for problems were necessary here that they agreed to put some items in this part of the agenda. A more experienced PM, however, liked the idea very much and readily used it to raise a concern he had for which he desired the SMs’ involvement. These experiences only further reinforced our finding that it was very difficult for PMs to present problems without solutions; they were keenly aware that the PA meeting was not just a discussion about the project, but also a time when their managers would assess their competence. Of course, this did not change after our intervention, but the presence of an officially sanctioned advice segment of the meeting allowed the PM to raise issues that he might otherwise evade. The pre-PA meeting was generally found to be quite helpful by both PMs and SMs. According to one PM: “There is often a gap between what the senior manager wants to know and what the project manager reports, and that is problematic, because they end up talking cross-purpose. . . . I used to get struck in the gap each time I went to the meeting. So, if the senior manager can show me before the PA meeting [what he wants], then I can adjust my report accordingly.”

During the research phase of this project this PM had told us that PA meetings were unhelpful and something he just had to endure, so we considered this comment to be a significant improvement. We also suggested that all the items on the agenda had time allocations and this too was welcomed by the PMs as they now had an officially

322

Making Work Visible

sanctioned resource they could mobilize when the discussion went off on less helpful tangents. Of course, this worked even better when one of us played the facilitator’s role. After all, we had no stake in the content of the discussion and could jump in at any time to ensure that the agenda was being implemented; something much harder to do for someone involved in the discussion. After we had facilitated a PA meeting, the PM commented: “It was good to have somebody controlling the topics of discussion. It is hard to keep time. Usually system engineers continue to discuss until they come to a conclusion once they start discussing.”

The creation of an agenda was also welcomed by the experienced PM, as he felt that SMs had been discussing topics that were of little interest to him for a long time, preventing him from getting his own concerns discussed. He used the agenda to his advantage by referring to it regularly and keeping the discussion focused. All in all, our interventions were simple but quite successful. In interviews following the implementation of the PA meeting redesign, both SMs and PMs conveyed that the PA meetings were more productive as the discussion was better focused. The changes to the meeting turned an obligatory monthly meeting that was often antagonistic and that many PMs felt they just had to suffer through, into a meeting that both parties considered beneficial. Despite the simplicity of the intervention, it actually helped the meeting participants reconsider an ingrained cultural practice. We do not know the extent to which these cultural patterns changed permanently, as our project ended shortly after the experimental implementation. We only worked with a small subset of SMs and PMs and the degree to which they were willing and able to spread the practice by themselves is questionable. Our success in facilitating the PA meeting was due at least in part to our neutral status at the table. In a hierarchical organization such as that of our clients, having the change come from within the organization would have been more difficult, as members would naturally adopt the very communication patterns we were trying to disrupt.

Summary and Conclusion Our ethnographic study of PA meetings and the analysis of the video data showed that the meeting was not accomplishing its primary goal, namely to

Rethinking How Projects Are Managed

323

create open communication to talk about IT projects risks. The way the meeting was originally structured and implemented made it a struggle between two opposing parties, the SMs on offense, and the project team on defense. As a result, PMs considered the meeting of little value and something they had to suffer through. We held up a mirror to the participants of the meeting by presenting them with bits of video that showed the ways in which the meeting was ineffective in achieving its risk management aims. Both parties agreed to let us change the meeting by introducing a structure that made it more collaborative. We facilitated some of these meetings ourselves. Both parties were happy with the results as they felt the discussion was more productive and addressed issues that the PMs considered important. Our intervention worked primarily because our ethnographic research had sensitized us to the organization’s culture. We knew we had to work with both the SMs and the PMs, and when it came to implementation, it was helpful that we were a neutral third party. The intervention did not change practice radically, but through some very deliberate changes to the structure of the PA meeting, we were able to make significant progress in changing the organization’s pattern of communication in ways that many had heretofore believed impossible. Instead of asking the participants to change their attitude directly, which would have been difficult and probably would have resulted in resistance, we made structural changes to the PA meeting itself that caused subsequent changes in the participants’ behavior (for similar work, see also Ikeya et al., 2010).

Part VI

Competency Transfer

18

Fujitsu Learned Ethnography from PARC Establishing the Social Science Center Koji Kishimoto with a Preface by Jack Whalen

Preface This chapter differs from all others in this volume in that it describes a three-year PARC project with Fujitsu – named here as the “PARC Knowledge Management Project” – from the point of view of our partner organization. Authored by the Fujitsu leader on the project, Koji Kishimoto, it provides a detailed account of the difficult situation Fujitsu faced in their system development division, and why they sought the help of PARC ethnographers. Kishimoto also gives us a frank assessment of what Fujitsu hoped to get from PARC – essentially, a “silver bullet” in the form of readyto-use techniques and tools that would produce immediate and dramatic improvement in their business. Moreover, as Kishimoto makes clear, Fujitsu system engineers were used to exhaustively plan out all their project work in advance, and so our ideas about exploration and discovery, where you first need to understand the true nature of the problem and then gradually begin to mark out a strategy to attack it, appeared rather strange. But soon enough we both found ways to work together effectively, to learn from each other. And as the chapter makes plain, at the end of the day we were quite successful. One of the most important accomplishments of the project, the creation of a Social Science Center inside Fujitsu’s system development organization – surely a first in the long history of software development – was achieved not by Fujitsu copying PARC’s ways of doing applied ethnography but rather by adapting our principles to their own operation, to their own organization’s culture. Nothing could have made us happier.

The Social Science Center Fujitsu’s Social Science Center (SSC) is using ethnographic fieldwork as a service to our customers, in order to make visible our customers’ current 327

328

Making Work Visible

business processes – those actually used in the field. By starting with the facts rendered visible through the fieldwork, we can then help the customer innovate their business. At the same time, SSC is offering fieldwork education inside the company to train our own engineers in these methods. Looking back, Fujitsu1 has been providing IT-related services to its customers for more than 40 years, with its in-house training focused almost solely on technical education. This is the very first time in Fujitsu’s history to officially introduce knowledge based on cultural anthropology or social science, such as ethnography or fieldwork. Naturally, the move attracted a lot of attention from external audiences as well. Why was it possible to realize this kind of initiative despite Fujitsu’s company culture which is highly technology-oriented with strong resistance to different kind of approaches such as social science? In fact, it was made possible through collaboration with PARC, which some say had a “black ship effect” on Fujitsu. This “Black ship effect” refers to Commodore Perry’s four black ships that demanded opening of Japan in late 19th century. The visit is seen by many Japanese as a shocking but necessary wake-up call that outsiders alone can effectively make. Let us describe what kind of effect the introduction of ethnography by PARC had on Fujitsu and how we developed it into business. Inauguration of PARC-Knowledge Management Project One of the things that served as the background of collaboration with PARC beginning in 2004 was the fact that Fujitsu was having a hard time in those days from reduced profitability and quality in its core business of IT system development. The other problem that plagued Fujitsu at that time was the low morale among Fujitsu system engineers (SEs), who played the central role in that key business area. In an attempt to improve the situation, the first thing Fujitsu did on the business front was to enhance auditing on contracts and upstream processes in order to nip in the bud as much as possible what were likely to end up being unprofitable projects. We also made a structural reform by way of integrating sales departments, which tended to put sales figures first, and SE departments, which were inclined to prioritize quality. The aim was to have 1

Fujitsu is a leading provider of IT-based business solutions for the global marketplace. With approximately 186,000 employees supporting customers in more than 70 countries, Fujitsu is the world’s fourth-largest IT services provider and No. 1 in Japan. Headquartered in Tokyo, consolidated revenues is US$47 billion for the fiscal year ending March 31, 2009.

Fujitsu Learned Ethnography from PARC

329

them work together in the same group to raise quality awareness on the part of sales departments so that profitability and quality could be balanced. (The effectiveness of these measures was vindicated by the ensuing recovery in the performance of system development business.) PARC’s ethnography was also expected to help with the improvement of SE’s morale. SE’s work involved so much frustration and cooped-up feeling that people mockingly called it a “3K” job. In this case, 3K stands for kitsui (difficult), kitanai (dirty) and kaerenai (can’t go home) in Japanese. People said this because: SE’s work is “difficult” as it involves a lot of coordination among different things; SE’s tend to be “dirty” since they frequently miss taking baths because they often have to work all night; and SEs “can’t go home” as they are always too busy. To tackle the situation, we did not want the kind of advice we typically get from critics but wanted to identify fundamental issues from the field’s perspective. Based on that, we also wanted to build a mechanism to share and utilize knowledge in order to help improve the lack of information sharing among SEs developing IT systems, as well as to boost their effort to improve quality. We thought that would generate active and creative communication, eventually leading to an upbeat workplace atmosphere. In consideration of the importance of such knowledge management, PARC-KM (knowledge management) project was kicked off in August, 2004, comprising of PARC ethnographers, Fujitsu Laboratory researchers, and Fujitsu SEs. At the time of the project inauguration, the term ethnography sounded highly attractive to Fujitsu SEs who had been single-mindedly pursuing IT and were working under the difficult business situation with a sense of being in a deadlock. SEs expected the new approach to produce various kinds of ready-to-use techniques and tools or even IT infrastructure that could be used immediately in their day-to-day work. However, as a matter of course, there was no such silver bullet. PARC members had to start with the effort to understand what Fujitsu’s SEs were, which meant they had to take time to do ethnography to understand the actual situation by visiting various workplaces where SEs were at work. The expectation gap grew between PARC and Fujitsu, which was after immediate results and ready-to-use weapons for the actual business warfare.

Ethnography Process Learned from PARC: Changes of SEs in the Field With the passage of time, it dawned on both Fujitsu and PARC that there was 180° difference in their stances as shown in Table 18.1 (Maegawa et al.,

330

Making Work Visible

Table 18.1 Difference in stances between PARC ethnographers and Fujitsu SEs PARC ethnographers

Fujitsu SEs

Social science centered

Engineering technology centered Focused on quantitative analysis Projects should be precisely scheduled Output should be defined in advance

Focused on qualitative analysis Projects should be open-ended, without setting milestones in advance Do not predefine output in advance

2009 is the original source of this data): with this much difference, the collaboration between PARC and Fujitsu might very well have collapsed. Still, thanks to the understanding of the then president and then executive vice president of Fujitsu, the project somehow managed to continue. Meanwhile, despite strong criticisms from working-level Fujitsu employees, PARC ethnographers remained proactive and kept plunging into the work of Fujitsu SEs. Gradually, their effort based on firm belief and findings gained from tenacious observation in the field began affecting Fujitsu SEs and positive comments about PARC ethnographers started being heard from SEs in the field.

Three Ethnographic Processes What we learned from ethnography brought to us by PARC were basically three processes, which were: “field observation,” “reflection,” and “codesign.”

Field Observation PARC ethnographers’ field observation did not really produce a breathtaking discovery of some new facts. Actually, the key was the fact that they made many least surprising but quite important findings. These were the kind of things that seemed quite a matter of course when pointed out but were in effect difficult for people in the field to notice unless they were told. PARC ethnographers saw things from a third-party perspective with pure insight unfettered by ties Fujitsu employees found hard to break and made deep analysis on the observed facts. For example, based on field observation results, PARC ethnographers pointed out that participants from the field level did not report problems,

Fujitsu Learned Ethnography from PARC

331

which could not be solved on the field level alone, because they knew that if they reported a problem they would be requested to report the solution to the problem as well. This is an example of an issue that can’t be identified as a problem by people inside Fujitsu. It is even less possible for them to escalate it to higher management. However, thanks to persistent endeavors by PARC ethnographers, including video recording of the actual meetings to add visual appeal, there arose discussion in Fujitsu regarding the need for a forum of constructive discussion where working-level people facing difficulty in the field could candidly consult higher managers.

Reflection When a system development project is going on, various problems arise but in many cases project members somehow manage to tide them over with stopgap measures. What is usually on their mind at such times is, “Let us formulate a fundamental measure when this project is done in order to make sure we can handle things in an organized manner the next time around.” But in reality, they cannot afford to do that as they are usually put into the next project as soon as the one at hand is over and become busy wading through immediate difficulties with stopgap measures just as they had been doing before. This cycle is repeated on and on. We would like to cite one example of “reflection” effort introduced by PARC members. In a project with multiple sub-groups, there were a lot of troubles related to interfaces among different subsystems. Every time trouble occurred, one of the sub-groups implemented some stopgap measure. But most of the problems were minor ones that could be prevented if they had good communication in advance. PARC members presented the actual status of interface-related problems as one of the results of the field observation and convinced people in the field that it was the fact. Then they led every member to think about the background of the problems through reflection during the morning meeting that was attended by all the members. This effectively guided the project members to take spontaneous actions, which flowered in the form of communication during the morning meeting that led to the prevention of such problems.

Codesign PARC did not simply observe the field and prompt individuals to reflect themselves based on the facts obtained through their field observation – PARC also generated a specific and creative forum of “codesign.”

332

Making Work Visible

For example, in the case of a certain project, there were different strategies in different departments and they somehow managed to come to terms among stakeholders with different axes to grind. In order to share this delicate fact learned through their observation, PARC ethnographers provided a forum in the form of “codesign” session. PARC ethnographers brought together stakeholders with conflicting interests, reported facts they observed through field observation, and served as the facilitator during discussion where various opinions were heard. By doing so, PARC members coordinated a forum for stakeholders to face each other fully to have constructive discussion on issues that concerned several different departments, although such issues are typically left unaddressed and remain vague. One participant in the codesign session self-deprecatingly said, “I feel embarrassed about the fact that to discuss this important issue we had to be told to do so by a third party and we even had them organize this forum for us,” and greatly thanked the contribution made by PARC members.

Social Science Center If we compare IT industry’s approach to seek immediate result to Western medicine, PARC’s ethnography-based approach can be compared to Oriental medicine. Western medicine sees a disease locally, analyzes, and pursues the cause of the disease whereas Oriental medicine tries to understand the disease not just locally but in relation to the patient’s mental and environmental background as well. Western medicine can be seen as an approach to “cut off the bad part.” In contrast Oriental medicine can be seen as an approach to “heal from inside.” Ethnography introduced by PARC showed an effect like that of Oriental medicine’s. It was very interesting in the sense that a Japanese company was prescribed an Oriental herbal medicine by a U.S. research institute. Incidentally, while the term ethnography is often used to mean “field observation” – including the interpretation of what has been observed – and writing qualitative analysis as the output from the observation, the ethnography Fujitsu learned from PARC and interpreted in our context includes the whole set of processes from “field observation” to “reflection” to “codesign.” Since it goes further on to “reflection” and “codesign” without stopping after summarizing field observation results as a research paper, it becomes easier for people in the actual field of business to accept because what is offered to them comes in the form of solutions. In order to hand down and continue this effective ethnography in a proper manner, Fujitsu created a new organization called Social Science

Fujitsu Learned Ethnography from PARC

333

Center (SSC) mainly comprising former PARC-KM project members. The organization is intended to serve to continuously revitalize organizations inside Fujitsu. In recent days, SSC is adding various flavors to the basic processes of “field observation,” “reflection,” and “codesign.” For example, we have a service called “organization monitoring,” which we keep providing inside Fujitsu group. It is a service to use IT to make visible the status of communication among the members of a target organization and continuously measure the organization’s ability to change itself against a certain scale, with the resulting trend represented in the form of a graph. As the effect of ethnography started getting recognized, it was decided to start considering the possibility of using this approach not only for internal organizational improvements/changes but for a service for customers as well. More specifically, we began exploring the possibility of doing field observation on customers’ work processes and business to render them visible, and leading the visible facts to various improvements and changes. In order to realize that, while retaining the strength of ethnography, which is to capture facts as they are, it was also necessary to come up with Fujitsu’s original way of doing ethnography based on the assumption that Fujitsu’s stance described in Table 18.1 was shared with our customers.

Business Ethnography Ethnography has come to be taken up in business magazines in Japan, too, over the past couple of years. One of the articles that heralded the trend was found in a feature on ethnography in January 2008 in the magazine Nikkei Electronics. The theme for the feature was “designing the users’ ‘experience.’” In that article, in comparison with other research methods such as interviews and questionnaires, the strength of ethnography was described as “an approach that makes it possible to grasp requirements customers are unaware of,” while its weaknesses described as “timeconsuming,” “dependent on individual skills” and “lacking in quantitative analysis.” The condition for us to be able to offer the service to external customers was to improve these three weaknesses, making it possible to schedule the activity as well as to make the output predefinable (Kishimoto et al., 2008, 2009). First, to solve the problem of ethnography being “time-consuming,” instead of using the usual approach of single-person ethnography, we tried to reduce time consumption by having a team of people do ethnography, as was the case with PARC’s ethnography. In order to do that, we defined each

334

Making Work Visible

activity step in detail so that ethnography can be done in a controlled manner without too much variability even when multiple people do it. Activity steps we have defined encompasses everything from selling the idea to an internal audience as well as having advance coordination with the people concerned as part of preparation ahead of time, how to introduce customers, field note format, how to write field notes, and how to sum up the result as a group, and so forth. With regard to the countermeasure for the problem of being “dependent on individual skills,” we made sure every member clearly distinguished between facts and personal findings/interpretations in writing field notes. Furthermore, during group discussion we made it a point of sorting things out starting with facts. By honing these skills through practice, SSC members’ competence to write down facts and make findings dramatically grew. By now they have reached a level that they can write more than 100 pages of field notes based on a one day observation regardless of the type of work or industry observed. Of course, it is not only about the quantity but their field notes are so high in quality that they can give a very vivid idea of what it was like in the field to those who could not go to the site. Thanks to this, people concerned can have a simulated experience of the field and feel as if they had been there. Then the whole group can produce a high-quality interpretation based on the accounts of facts and findings that are perfectly satisfactory both in terms of quality and quantity. To address the problem of ethnography “lacking in quantitative analysis,” a tool has been developed so that quantitative analysis such as work categorization can be done automatically based on the time and name of the work captured in field notes. “Business Ethnography” service comes as a package of vividly visualized facts as described above, qualitative analysis based on them, and quantitative analysis illustrated by visuals such as graphs. The service was already applied to customer work sites at more than 40 companies and SSC has carried out training with more than 400 Fujitsu employees. Through these experiences, we are now developing an increasingly satisfactory array of basic schedule, output template, and so forth, that can be presented to customers in advance. In addition, as a result of analysis done on more than 10,000 findings based on facts, SSC has developed an analysis database called “Fieldwork Map,” which is used to verify the completeness of individual fieldwork results. Furthermore, the results of field observation are classified into common human activity categories, such as walking around, PC work, telephone call, and so forth, to make it possible to perform quantitative benchmarking among different companies.

Fujitsu Learned Ethnography from PARC

335

Conclusion It is true it took some time, but ethnography we have learned from PARC earned recognition for its importance inside Fujitsu. Subsequently it has been evolving in terms of business context, through SSC, as a service unique to Fujitsu. SSC is now working both on internal and external fronts. Inside Fujitsu, we are making steady efforts to keep improving our work places, for example, through “organization monitoring.” As we learned in our experience with PARC ethnographers, there is no short cut and we have to keep working on ourselves. On the external front, we often receive voices of surprise from our customers, such as “I was surprised to learn that Fujitsu, which is an IT vendor, has this kind of service,” or “so, Fujitsu is not simply selling hardware and software.” Meanwhile, we also get comments like, “we can be convinced because it is backed by graphic depiction of facts,” or “the homespun image of this service is just like Fujitsu.” As you can see in these comments, ethnography is certainly changing work styles of both Fujitsu and its customers. While our organizations keep continuous efforts to improve themselves, Business Ethnography is evolving in way to open up a new frontier. At one company, Fujitsu is engaged in a discussion with the customer on a concrete level not as a mere vendor but as a partner that knows business issues in the field. In addition, collaboration based on Business Ethnography has started between Fujitsu and the information systems department of one company. As can be seen in these examples, strategic partnership between Fujitsu and its customers is now being formed. The seed sown by PARC ethnographers has taken root and is now growing steadily on our soil.

19

The Work Practice Center of Excellence Luke Plurkowski, Margaret H. Szymanski, Patricia Wall, and Johannes A. Koomen

In late 2005, a Xerox business group, Creative and Technical Communication (CTC), began to sell its services and solutions to clients outside of Xerox. This transition to the external market would require consulting skills, yet the subject matter experts inside of CTC had very little, if any, experience working with clients other than Xerox. Additionally, because CTC would be a new player in the external market, it needed a distinct way to differentiate itself from its older, more experienced competitors in order to gain traction and increase revenue. Tom Hurysz, Vice President of CTC, recognized the need to invest in work practice training for his consultants and to establish a center of excellence in his organization that could support their external marketing efforts. Historically, work practice researchers had helped CTC (see Sprague et al., this volume) and other business units with various customerfacing projects; in these engagements, subject-matter experts would work alongside work practice analysts. So Hurysz was familiar with the value of work practice study; and a recent rental car client engagement confirmed the value that work practice could bring (Sprague et al., this volume). Hurysz was quick to champion a work practice study training effort in his organization, stating, “I want work practice study to be a part of the consulting methodology because I think it reveals issues and opportunities that a normal process engineering method is not going to get at” (personal communication, 2008). The goal was to integrate the work practice competency with their existing competency in Lean Six Sigma methods, in all four of their core offerings that address content use and management across the product lifecycle: Workflow and Content Management, Authoring, eLearning, and Translation (see Figure 19.1). Leveraging expertise from across the organization, the Work Practice Center of Excellence (WPCE) project was a collaborative effort across the three Xerox research centers with work practice competencies – Palo Alto 336

The Work Practice Center of Excellence

337

Figure 19.1 Creative and Technical Communication’s offerings

Research Center, Xerox Research Center Webster, and Xerox Research Center Europe. PARC, in particular, had some recent experience establishing a corporate competency center as part of a multiyear research project for Fujitsu (see Kishimoto and Whalen, this volume). The goal was to develop a self-sustaining work practice training program within CTC in three years. The objectives for the first year were: (1) to understand the skills and needs of the target audience, CTC employees, (2) to gather and synthesize all of the instructional materials developed by work practice researchers in Xerox over the years, (3) to assemble a reasonable structure for the training program and modify the curriculum as needed to fit this structure and the target audience, and (4) to pilot some low-level training before year-end. Year two was designed to be the pilot of the entire training regiment, and throughout the year, we revised the curriculum and delivery as needed. In the third year, we launched the competency by certifying a critical mass of work practice analysts as well as a certified work practice trainer who could sustain the center of excellence in our absence.

Developing and Delivering the Training The project team met in early 2007 at a two-day workshop to discuss several important topics, including an overarching vision for the training program. Over the next couple of months, the team assembled a repository for collecting existing ethnographic training materials (Button et al., 2003;

338

Making Work Visible

Ikeya et al., 2007). Our goal for the curriculum design was to clearly convey how to do work practice analysis. This is why the project team made case studies and applied learning in the field the foundation of its training regiment. In April and May 2007, PARC project members initiated the codesign phase of the curriculum by touring CTC and interviewing its employees to gain a better understanding of the target audience, what the work in each of the four core service areas was like, and how to shape the training in order to best support their work. During these two visits, the project team engaged closely with the dozen CTC employees selected for the first round of training activities. This study enabled us to build a relationship and get buy-in from CTC participants, and, most importantly, to get their input on how to structure the training. It also gave them a sense of ownership, knowing they influenced how the training was to be designed and delivered to them. We found that while the CTC consultants lacked formal training in ethnographic field methods, they would often adopt a client-centered perspective in their work, even if they were not referring to it as such. Additionally, we learned that several employees had been previously trained in Six Sigma methods, so we took this into account in designing our training. We shaped our qualitative methodology to complement Six Sigma and ensure they were not missing out on the truly valuable “nuggets of practical variation” that Six Sigma often overlooks in its goal to standardize processes. The first training was piloted in June 2007 to a large cohort of CTC employees and management from CTC and other Xerox divisions. This “Introductory Training” was delivered over two consecutive half-days, and included three hours of classroom instruction, a two-hour field exercise, a three-hour group presentation exercise, and a brief concluding quiz and course evaluation. From this cohort, the first group of Work Practice Apprentices was selected to continue training in the fall. This was also the first exposure to work practice analysis for several of the managers present. The training enabled the participants to gain valuable insight through the field exercise and group presentations. As one manager commented: “Once I got it, it was like I could put on my work practice glasses and see the world differently.” At this point in the project, an interesting divergence took place. Originally, the project team had specified two levels of training for CTC: an “introductory” course that all employees could attend to understand what work practice was, even if they themselves were not tasked with work practice analysis in their job role, and an “apprentice” training level where a subset of CTC consultants would be trained more in-depth in order to

The Work Practice Center of Excellence

339

Table 19.1 Levels of work practice competence 1. Awareness: “I can recognize an opportunity for WP value.” 2. Introductory: “I can identify where to position WP as part of a solution.” 3. Apprentice: “I am training in WP methods.” – Classroom Training (Level 1): “I am learning WP methods.” – Field Training (Level 2): “I am practicing WP skills in the field.” 4. Xerox Certification: – Certified Work Practice Analyst: “I can effectively use work practice methods independently and can contribute to a WP project team.” – Expert Certification: • Certified Work Practice Project Manager: “I can design and lead a WP study effort.” • Certified Work Practice Trainer: “I can train others in Xerox in WP (levels 1–4).”

actually utilize work practice methods in future external client engagements (see Table 19.1). The first apprentice training was slated for October 2007. The divergence came when the project team realized CTC would have trouble selling this new competency unless sales people could be trained in what it was, what the benefits were, and when it was appropriate to use. The intractable problem here was that Xerox’s sales team was spread across the United States, and so it would be impossible to train them in groups using the “introductory” level training that had just been piloted. The team devised a third level of training, which they deemed “awareness” level work practice training. This course was designed to be quite brief (ninety minutes) in order to fit into the extremely busy schedules of Xerox sales people. Additionally, it was developed with the idea that it could be delivered remotely over conference call (which occurred throughout late 2008 and early 2009), and eventually transformed into an e-learning module that could be deployed easily over the web. Its content was strictly geared to the needs of sales people, and was piloted in-person in Palo Alto, California in July 2007 to a group of five sales people who could provide feedback about its design and delivery. Once again, we involved the target audience in codesigning the solution, a cornerstone of Xerox’s work practice tradition; this was instrumental in getting this level of training tuned quickly and accurately (see Whalen and Whalen, this volume). The team moved forward with piloting “apprentice” level training to three selected CTC employees in October 2007. This course lasted four days, but would later be expanded to five full days in the second delivery. It included alternating sessions of classroom instruction with direct field experience in a

340

Making Work Visible

Xerox print shop embedded inside of a client’s location. Over the course of the week, the apprentices were trained in the work practice methods described in the introductory training, building upon experiential learning through a mock-work practice study in a real customer environment. The apprentices left this week with a deep understanding of work practice methods, and some understanding of how to apply them in real world situations. To further bolster the apprentices’ skills, the project team decided to add field mentorship as the critical step in the certification process. This would allow apprentice candidates the ability to participate in authentic work practice studies with real deliverables and get the hands-on experience they need to master the work practice skills they had learned in the classroom. Additionally, it would be a way for the training mentors to see their apprentices demonstrate their mastery of each skill deemed critical for Work Practice Analyst certification As a result of a field mentoring exercise in February 2008, the training curriculum structure would diverge once more, as the team became aware of another gap in their training hierarchy. Who in CTC would lead work practice studies once they began selling this service? And who would continue to train progressive generations of Work Practice Analysts in CTC once the researchers moved on from this research and training engagement? The project team rallied around an idea to develop advanced certification levels above the previous pinnacle of the program, the Work Practice Analyst certification level. Now, certified Work Practice Analysts could be selected for two “advanced certifications,” based upon their unique skill sets in the areas of project management and mentoring/instruction. To receive advanced certification as either a “Work Practice Project Manager” or “Work Practice Trainer,” a candidate would have to complete minimal additional coursework (over two days), and participate in additional field experiences with apprentices-in-training. Project Manager candidates work closely with the project leads of work practice studies that apprentices participate in as part of their field mentorship regiment. Skills such as how to scope a prospective work practice study for a client service contract, how to deploy resources effectively for high-quality fieldwork, and how to manage a team of Work Practice Analysts in the field are acquired by working closely with the lead on an actual project. Similarly, trainer candidates shadow existing project team members as they continue to deliver training and mentor apprentices in the field. Important skills for trainers are acquired through these experiences, such as how to effectively teach the curriculum to new trainees, how to engage productively with apprentices in the field during mentorship, and how to iterate on the curriculum in the future to ensure its continued effectiveness.

The Work Practice Center of Excellence

341

Lessons Learned Optimal class size has been constantly adjusted throughout the three-year project for all levels of training. We found that awareness training delivered to sales people can be of any reasonable size for a conference call, since the curriculum is shaped to be an overview of work practice and its value for potential customers. With the conversion to an e-learning course, individualized consumption is possible and the problem is largely solved (although there is no built in feedback channel to an instructor for questions or deeper enrichment). For the introductory training, we first piloted it to about twenty five employees and managers, at about an equal split. We found in our second workshop that having a 4-to-1 ratio of employees to managers works considerably better, and that capping this course to about twenty is more manageable and the field exercise is much more potent for the attendees. In this smaller class, CTC employees seemed much more engaged, less inhibited when speaking to the group about their experiences in the field, and overall were more successful in learning the core principals and reinforcing them with others. To pilot the first apprentice training workshop, we accepted only three CTC consultants in order to maintain a 1-to-1 ratio between trainers and apprentices and to devote all of our attention to our apprentices when engaged in the field experience. For our second workshop, we accepted nine apprentices and had four instructors. This proved to be taxing, not only on the part of the instructors who were stretched thinly with groups of two (or three) apprentices each in the field, but also for the apprentices themselves as they received less time devoted to presenting their work and having their questions addressed. In the future, class size will be limited to a 1-to-1 ratio, which will be much more reasonable for CTC to self-sustain with their certified trainers. Additionally, it is detrimental to the apprentices during field mentoring if their classroom cohort is too large, as only a few apprentices are able to be taken into the field with a trainer at any given time, so the skills they have learned will begin to atrophy before they are able to apply them. Another challenge the project team continues to struggle with is how best to impart the skills needed to conduct qualitative data analysis, synthesis, and reporting. Unlike quantitative methods, which are based upon mathematical principles that are easily transplanted from context to context, qualitative analytic skills are impossible to impart through a textbook or a cursory classroom lecture. Work practice analysis truly requires that an analytic perspective and the associated qualitative analytic skills be derived

342

Making Work Visible

through numerous hours and experiences with real data. There are no shortcuts, and no rubrics for novice analysts to follow and “always get it right.” The best method to impart these skills is close apprenticeship with experienced work practice practitioners. Field mentoring, a large component of this apprenticeship, involves securing a client engagement that the apprentice can participate in with the help of the mentor. This phase is the most time intensive and least predictable of all the training components, since each client engagement is unique in its requirements and timeline. Field mentoring also caused us the most trouble in terms of getting the apprentices certified. Our two main challenges were securing the client engagements, and conducting the studies while at the same time mentoring the apprentice. At first, we relied on CTC to secure client engagements for use as mentoring opportunities. Unfortunately, since this was an emerging market for them, CTC had difficulty securing a continuous stream of work practice eligible projects which could be leveraged for field mentorship. Our dilemma was whether to be open to any mentorship opportunity or to adhere to CTC’s preference that the projects be closely related to the work of CTC. Ultimately, we opted to execute several studies within CTC to stabilize the training pipeline. And over time, we convinced CTC to relax their preference for CTC-related projects, by telling them that the topic of the work practice study does not impact its effectiveness as a training opportunity; the skill of the mentor and the full application of work practice methodology are much more crucial to a successful field mentoring experience. Additionally, we found that by bringing apprentices outside of their comfort zone and encouraging them to apply their new skills to a completely new field, they improved their qualitative perspective beyond what they would have been able to achieve in a more comforting, familiar fieldsite. At its core, field mentoring is uniquely challenging because apprentices are mentored on actual work practice studies with accountability to deliver results to a customer. For both the mentor and the apprentice, the expectation is high, and the team must strive for productivity on the study as well as mentoring progress. To effectively accomplish these goals, our strategy has been to adhere to a 1-to-1 ratio of mentor and trainee during field mentoring. However, this strategy can create a bottle neck for certification progress and limits our ability to certify large numbers of people in rapid succession. Of course mentorship also occurs in the data analysis lab, where mentor and apprentice work together processing, analyzing, and synthesizing the multitude of data collected in the field: video recordings, audio recordings,

The Work Practice Center of Excellence

343

photographs, field notes, documents, and other artifacts. As instructors and mentors throughout this training project, we have constantly refocused our efforts on creating experiences that help our apprentices to engage in handson learning, to learn by doing. As critical numbers of analysts became certified, the team has organized sessions to report back to the growing work practice community within the organization. This has been very beneficial, as presenting the material to an external audience forces the analyst to strengthen their analytic argument. These sessions have also strengthened community interest, and increased management’s knowledge about how to support the work. Further, we have offered ongoing “refresher” and “data analysis” classes in order to maintain progress for the cohort of apprentices.

Conclusion The Work Practice Center of Excellence was established to help a business group deliver solutions that are founded on existing work practices and therefore fit better in their customer’s organizational ecology. The project developed a multilevel curriculum and method for teaching consultants without prior social science expertise how to conduct work practice analysis. Now in its third year, this work practice competency has gained traction within the organization and is being leveraged in other business units as a complementary skill set to Six Sigma quantitative analysis. Certified Work Practice Analysts have reported feeling more confident in working alongside their customers to understand the customer’s work from the customer’s perspective. In this sense, we succeeded in equipping the consultants with work practice skills they could utilize to build their business externally. As Maryann Fuhrmann, a CTC certified Work Practice Trainer, comments, “For the organization, I believe that the true value of Work Practice is the significant impact we can make on the development of more effective, people-centered work solutions that help our clients save money, improve their processes and grow their business.” So as the Work Practice Center of Excellence continues to flourish, so too will the ability of its analysts, project leaders, and mentors to have a positive impact on the customers they engage.

20

Transferring Ethnographic Competence Personal Reflections on the Past and Future of Work Practice Analysis Brigitte Jordan1

The Rise of Ethnography In the last few years, ethnography has taken on a new prominence and popularity in the business realm. As a consequence, many of our corporate partners are thinking about internalizing ethnographic expertise by “transferring” some measure of ethnographic skills to their employees.2 We have fielded sporadic requests for this kind of teaching for a very long time, both at PARC and the closely allied Institute for Research on Learning (IRL), but in the last few years they have become noticeably more frequent, sometimes attached to requests for research on recognized issues, such as technology development or understanding customers. In other words, teaching ethnographic field methods has become a product, an “offerable” for institutions like PARC. The tension between ethnography as research and ethnography as product is increasingly resolved by moving

1

2

344

As is clear from the text, I owe a tremendous debt to my current and former colleagues at PARC and IRL. Everything I say is, in some sense, a further commentary on the many, many discussions we have had over the years. I thank Susan U. Stucky and John Seely Brown for information and insights that vastly improved my memory. I particularly want to acknowledge Yutaka Yamauchi with whom I discussed many of these issues as we prepared a short article on teaching in the corporation (Jordan and Yamauchi, 2008). My deepest thanks go to Peggy Szymanski, Robert Irwin, and Jack Whalen for awesome editing and support in a trying time. I am adopting the “transfer” language here for consistency and convenience of a business audience though it is misleading in that it implies that bits of knowledge are “transferred” from a sender to a receiver, a teacher to a student. In fact, learning is fundamentally social and involves collaborative sense-making that builds shared knowledge communities and supports the collaborative exploration of a problem space. Managing (transferring) knowledge across organizational boundaries has been an issue at least since Shannon and Weaver (1949), Polanyi (1966), and Nonaka and Takeuchi (1995), and a variety of conceptual frameworks has been proposed. For an insightful and forward-looking review of these frameworks see Carlile (2004).

Transferring Ethnographic Competence

345

ethnography into a service function, a function that supports technologyfocused research as a validating rather than as a discovery science (Whalen and Whalen, 2004).

A History of Ethnographic Research and Teaching at PARC and IRL The full history of the relationship between IRL and PARC remains to be written. Here I offer my own recollections, based on my experience during the more than ten years of IRL’s existence, of how work practice analysis arose at PARC out of anthropologically grounded ethnography and ethnomethodology. All such accounts are socially constructed, drawn from a memory that is fallible and only rarely reflects “the facts.” Nevertheless, multiple histories are important for understanding where we have been, where we are now and where we are going. It is in this spirit that I recall some of the pivotal events that occurred after I accepted a joint appointment between PARC and IRL in 1988; I continue with a discussion of some of the issues that have emerged around the successes and failures of teaching about these methods in Xerox and other companies. The early years during which PARC emerged as a leader in the application of ethnographic methods to business concerns were shaped by a close alliance between WPT (the Work Practice and Technology group at PARC) and IRL. During that time, a major factor in energizing the connection between the two institutions were the activities and ideas that emerged around the weekly Interaction Analysis Laboratories (IALs) which I had introduced when I joined PARC/IRL. Grounded in anthropological participant observation, IALs were devoted to the analysis of video recordings from ethnographic field studies that were collaboratively microanalyzed by an interdisciplinary group of researchers, very similar to the lab sessions Harvey Sacks had conducted at the University of California at Irvine. The IRL/PARC labs were sometimes held at IRL, sometimes at PARC, but as a matter of policy were open to anybody who had a video recording to be analyzed. They drew international participants as well as frequent visitors from area universities and Silicon Valley companies and were instrumental in keeping the flow of ideas in the community lively and productive. While there was a great deal of overlap, two somewhat distinctive subcultures developed as actual work practice studies were carried out. The WPT group at PARC, under the leadership of Lucy Suchman (see Chapter 1, this volume) was more academically oriented in their analysis of airports,

346

Making Work Visible

lawyers’ offices, and bridge-building engineers.3 On the other hand, much of the work carried out at IRL in the 1990s focused on providing insights and recommendations to the companies that were funding IRL (of which Xerox was the most important), something that, at the time, was disparaged in academic settings. Nevertheless there was an exceptionally productive flow of personnel and ideas between PARC and IRL that generated an extraordinary alliance and strong collaboration. The tight coupling between these strands of endeavor produced over the years a distinctive approach to the analysis of work practices, workscapes, and later lifescapes,4 as well as a grounded understanding of the obstacles ethnographically based work practice studies were facing in corporations and how we could respond to them ( Jordan and Dalal, 2006). In my view, the main accomplishments of the IRL/PARC collaboration during these years were two-fold: *

*

The seminal theoretical and conceptual advances developed during that time around social learning, communities of practice (COPs), context sensitivity (situatedness), and knowledge as a collaborative achievement rather than as a collection of transferable items, and: The adaptation of anthropological research methods that had been developed in the study of exotic communities to large, technologydriven organizations.

Work practice analysis was originally internally focused, that is to say, on other parts of Xerox rather than on external clients. It began during an historic event in the late 1980s when Shirley Edwards, then VP of Real Estate for Xerox, came to PARC to conduct a meeting concerning facilities issues. At the meeting, I talked about how ethnographic methods could provide detailed data and insights about how people actually use space as they carry out their 3

4

Lucy’s degree is in anthropology, and like a number of anthropologists committed to understanding the endogenous organization of human practices, she had become very much drawn into ethnomethodology and conversation analysis. I had studied with Sacks, Garfinkel, and Schegloff but together with other anthropologists at IRL (such as Jean Lave), remained in the classic tradition of anthropological ethnography. Harvey Sacks himself had always seen CA as aligned with anthropology as much as with sociology and was preparing to do fieldwork in Mexico at the time of his death. The idea of “lifescapes” came out of early work at IRL and WPT. By the early 1990s our focus on work practice studies had expanded to include the more holistic notion of “workscapes.” In a project with on-the-move high-performance executives I coined the term “lifescapes” because it became clear that work was no longer confined to work-in-theworkplace but had spread into people’s “other” lives. The idea of “-scapes” as indicating horizontal cultural conceptual domains has been publicized by Appadurai with “ethnoscapes,” Cefkin with “rhythmscapes,” and many others (Appadurai, 1986; Cefkin, 2007).

Transferring Ethnographic Competence

347

work. I remember clearly that other attendants were distinctly nonplussed by what I had to say, probably expecting some technology-focused innovation ideas from PARC, but Shirley said “Oh, that sounds interesting. This is different.” Shirley became an advocate for ethnographically based work practice analysis. She was excited about the potential of ethnography for furthering her own objectives (which were to “increase productivity by supporting work”) and began to look for suitable sites where that approach could be introduced into Xerox operations. Eventually she brokered the first full-blown workplace study, the “Work Practice and Design” project at a Xerox call center in Dallas that demonstrated the value of the ethnographic approach to workplace-based issues. Ethnographic fieldwork produced recommendations for building a new call center that shaped the design and layout of work spaces for productivity and employees’ enjoyment, and generated proposals for redesigning Xerox training programs.5 While there was then no explicit effort to transfer ethnographic methods to the participants as part of our research, our presence and activities over more than a year implicitly carried the message that the desired transformation to a “learning organization” would need to be grounded in a deep understanding of employees, their work practices, and concerns. Significantly, this project was also our first opportunity to introduce and demonstrate the power of video-based Interaction Analysis to call takers and their trainers, an effort that taught them to “see the invisible” and generated insights about many of the issues inherent in their work (Bishop et al., 1994). We regularly invited call center workers to “pizza lunch meetings” where we showed videotapes of the customer calls they had dealt with, maybe just the day before. Looking at themselves at work and being able to talk about what they did was inherently interesting to them and with some support from us (we often pointed out things that they didn’t see) they gained insight into the ways in which the technologies and resources available at their workstations supported their work practices (or not). They became analysts of their own work. Out of this came employee-based 5

In this first large project William Clancey and I were co-Principal Investigators (Bishop et al., 1994). Shirley Edwards also funded the continuation of this project at the new call center in Lewisville (see Chapter 10, this volume) where Xerox was experimenting with reengineering their customer support work processes. IRL researchers Jack Whalen and Marilyn Whalen led ethnographic research in support of that experiment, which is described at length in Chapter 10 of this volume. Shirley went on to fund the Systematic Assessment Project at Xerox Business Services, where we partnered with an entire Xerox business division in a large-scale effort to create a Learning Organization (Aronson et al., 1995).

348

Making Work Visible

initiatives for generating change, some of which were favorably received by management. As a result, the call center trainers began to experiment with incorporating video in their repertoire, out of which emerged standards for what should be taught to novice call center workers. The trainers became strong advocates for using collaborative video analysis for coming to a deep and useful understanding of work practice issues. This was in effect a kind of indirect methods transfer and was successful largely because it directly dealt with workers’ and trainers’ issues. During these projects, almost without noticing, we regularly found ourselves attaching methods instruction to the discussions about research issues. Those instructions in many cases became workshops and courses. In other cases, they became routine instructional components of joint research meetings, as for example in a project with Nynex Science and Technology where methods discussion became an expected part of every research meeting that IRL- and Nynex-based researchers carried out. This period also saw the development of a number of project reports, white papers, and later journal publications that explained our approach and methods (Blomberg et al., 1993; Button et al., 2003; Jordan, 1996; Jordan, 1997; Ruhleder and Jordan, 1997; Suchman and Trigg, 1991). Building on the experience and confidence we had gained with early internal, Xerox-embedded efforts, we eventually branched out, developing a number of more or less formal course and workshop offerings that were designed to help large organizations develop internal ethnographic competence. In the long run, this led to the development of methods learning centers in PARC client companies (Kishimoto and Whalen, this volume; Plurkowski et al., this volume). What then have we learned during this long and complex history about embedding ethnographic competence in client companies? How can we teach employees who often have minimal research background about the benefits (and limitations) of using ethnographic field methods? I draw on my work and that of my colleagues at PARC and formerly at IRL to suggest that three levels of knowledge transfer need to be considered: expertise in basic field methods; analytic competence; and, ideally, strategic relevance. I will discuss these in turn.

Level-1: Basic Field Methods at the Toolkit Level Field methods epitomize the craft of ethnography. The bulk of the teaching for transferring ethnographic expertise is on the level of techniques. It is comparatively easy to teach and to learn the craft of ethnography and such

Transferring Ethnographic Competence

349

teaching shows impressive results in a short time. It provides basic grounding that differentiates ethnography from other research methods, especially focus groups and predesigned surveys. The goal of a first learning engagement is thus to provide corporate learners with a toolkit from which they can choose data collection techniques for their projects. A typical course of several days begins with historical background, that is, the transformation from exotic anthropological “jungle ethnography” to company-relevant “business ethnography.” Often a discussion is also required upfront in order to establish research as a data-based attempt to answer corporate questions. From this emerges a grounding conversation about what kinds of data can answer what kinds of questions, leading to an appreciation of what ethnography can and cannot do. Since we are convinced that most learning happens in the doing, our courses and workshops are interactive, experiential, and collaborative and often include role-playing during instructional sessions. They are built around field experiences of various lengths and complexity. We draw most materials for exercises and assignments from our own fieldwork, but often also include data that course participants contribute. For example, in Figure 20.1, a photograph of two friends listening to music on the same headset, generated a spirited discussion about the lack of a good design for such an obvious need that was followed not only by brainstorming about what other data would be useful to understand joint headphone use but also by redesign suggestions. Depending on participants’ needs, we teach and practice a variety of data collection methods, such as participant observation, on-site interviewing, documentation by photography, and audio and video recording. The overriding goal is “learning to see” in a new way. Or phrasing it more ambitiously, reconstructing their view of the world. For example, in business environments, thinking is heavily influenced by the Six Sigma approach, a business management strategy originally developed by Motorola. Six Sigma decontextualizes, abstracts, and aims to reduce variance. And as John Seely Brown (JSB) points out, it is the furthest thing from situated analysis or understanding how to take advantage of these complex situations. Its goal is to reduce variance while the purpose of ethnographic participant observation is to understand variance (Peggy Szymanski, interview with JSB, August 27, 2009). Early in the course of teaching, participants may continue to see themselves as objective outside observers – as data collection machines that document sites, events, and interactions objectively rather than as participant observers, continuing to attempt to embody the ideal of deductive, laboratory science (Jordan and Yamauchi, 2008). This attitude is particularly pronounced in those whose training and work experience tends to reduce

350

Making Work Visible

Figure 20.1 Joint headphone use

reality into abstract structures, consisting of boxes and arrows, as in the ubiquitous workflow diagrams. Such representations allow only a yes/no dichotomy without gradations, representing flow as a sequence of on/off states with no grey areas. Teaching ethnographic work practice analysis, even on a fundamental level, works towards counteracting such ingrained ways of seeing problem and solution spaces. Throughout our teaching sessions we focus on signal topics that are particularly important in ethnography such as how to build trust with research participants (critical for effective participant observation), or that distinguish it from other approaches, such as the difference between lived “practice” and documented “process.” We emphasize the fact that what people say is not necessarily what they do (often leading to a discussion of the proper and improper uses of focus groups), and the emic/etic or insider/ outsider perspective (distinguishing between data collected from the point of view of study participants and those that are based on the categories the

Transferring Ethnographic Competence

351

analyst brings in from the outside). We attempt to instill in participants the idea that the ethnographer needs to assume the role of learner, of student, and as much as possible, of apprentice. We might illustrate emic/etic issues, for example, by pointing out that for understanding the work of pilots from an informed experiential view, it would be useful to learn to fly, but it would also be good to complement participant observation by having experienced pilots talk about the typicality, normality, general pervasiveness (or lack thereof) of the activities observed. We pay particular attention to the importance of systematic (rather than impressionistic) data collection and ways of achieving that, for example, by constructing team-shared templates for transforming raw notes from the field into data usable for analysis. This then opens up a discussion of topdown vs. collaborative approaches to data collection and, more recently, the use of communication and collaboration technologies (cell phones, cameras, email) for transforming “informants” into collaborating, data collecting research colleagues (see Schiano and Bellotti, this volume). One essential component of teaching basic data collection is instilling in novice ethnographers a serious respect for the ways in which biases of various sorts can affect their work (Convertino et al., 2008). In today’s business world where we are frequently called upon to investigate interaction between (often globally) distributed workgroups, understanding what is known as the fundamental attribution error (Cramton, 2002:193) is of particular relevance. But the general idea here is to build a constant attitude of bias vigilance, including biases that are personal and grounded in participants’ experience and life history. The slogan here is: know thyself! Embedding ethnography in corporations is an exercise in culture change that almost always relies on rephrasing questions, reformulating metaphors, and surfacing deep-seated attitudes that can interfere with the research process. For many people, the idea that reality is simply “out there” to be described and that analysis should be researcher-independent is deeply ingrained, while we argue that this kind of analysis is inappropriate for workplace studies because workplaces are dynamic, ever-changing, organic systems where it is impossible to control for variance. Learning ethnographic techniques at this level produces a valuable close-grained documentation of what “really” is going on in the workplace (something that many managers are not aware of). It also allows identifying local problems and suggesting local solutions. For participants, such courses or workshops are immensely satisfying because of the resources they make accessible for “seeing differently” and the insights they generate rather quickly.

352

Making Work Visible

Level-2: Teaching and Learning Deep Analytic Competence It is comparatively easy to teach Level-1 ethnographic techniques, and companies readily see the value of having ethnographically informed members in their midst. What is much more difficult to teach is full analytic competency. In order to develop the ability to relate patterns emerging from field data to company-relevant issues, deep analytic competence is required that can draw on theory and general concepts that are applicable beyond the local fieldsite. For some companies, toolkit level competence is not enough. For example, fieldwork may reveal that employees spend too much time in meetings, leading to a bullet list of recommendations for overhauling meeting rules and meeting technologies. This is certainly useful but such local stopgap (“bandaid”) solutions do not touch the company’s larger issues which have to do with what kind of work these meetings need to accomplish and why so many meetings exist in the first place. And so some companies recognize that for addressing systemic, enterprise-wide problems a higher level of competence is required. This means that both teachers and learners have to move from mere observation of interaction in time and space to a structural level, where one might ask: what is it in the system that spawns the issues we observe in the field? In other words, how can we move from the specific, local, observable, to the general? Companies do want to fix local problems, but the big payoff comes for them when issues can be identified at a higher level, that is to say, are generalizable to a larger universe, for example, all of the company’s call centers, or all call centers within the industry, maybe including competitors. This is the promise of deep analytic competence. Still, Why Is Teaching Analytic Competence so Difficult? Professionally trained anthropologists typically acquire analytic skills during years of graduate education that include multiple stints of fieldwork. During that time they live within a community of similarly engaged fellow graduate students while coming under the increasingly close supervision of one or more major professors. A part of this process consists of an intensive study of prior expert thought and opinion (“the literature”), both in heads-down cramming sessions and in seminar conversations that link students’ emerging ideas to the history of their field and allow them to draw on patterns and insights from that literature. After seeing, reading, and hearing about a sufficient number of examples, students learn to recognize and classify them. They learn to determine

Transferring Ethnographic Competence

353

which elements of the situation or domain must be treated as important and which can be ignored (Dreyfus, 2009). In the academic context, this selection is shaped by the history and current interests of the discipline in which the student is embedded. The important aspects of this stage are repetition; critical, positive relationships with superiors and fellow students; and fitting methods, topics, and insights progressively into those of their professional community. But for competent analysis in the world of work, it is company-specific interests and features of the work culture that provide the analyst with the criteria for deciding which of the many possibly relevant features they should take up. In contrast to “pure research” in which analysis is often focused on abstract anthropological or sociological topics, corporate ethnography is responsive to the interests of the company. The institutional resources inherent in a professional graduate program are mostly lacking in the corporate environment. The question then is whether similar analytic skills can be conveyed to nonacademics during a comparatively brief training period. Are there approaches that could approximate the kind of in-depth learning that happens in graduate school education? We have given considerable thought to this issue and have devised a series of interventions and experiments that can grow some measure of analytic skill at a substantially faster pace than traditional graduate school attendance. We identify three basic requirements: participation in an ethnographic community of practice, the opportunity to acquire experience with knowledgeable mentors, and involvement with an appropriate body of literature. The distaste of corporate people for reading “academic theory” (a.k.a., hogwash, bullshit, and pie in the sky) is proverbial. We have had some success with offering (or joining) reading groups on corporate change where we would propose articles that include successful application of ethnographic methods. Equally effective is introducing articles that misapply quantitative, hypothesistesting, or data-mining approaches in order to generate an interest in ethnographic alternatives, while making it clear that there are many issues in corporate life for which other kinds of research methods are more appropriate. Experience shows that a single individual cannot be taught to attain full analytic competence. While methods-trained employees should be able to collect data on their own, for the analysis of such data to show systemic results, it has to be vetted by a community. Popular culture still champions the Marlboro Man’s solitary excursions (reflected in academic promotion criteria and corporate performance appraisals) but analytic competence is really a “team sport” in the sense that it requires a community within which analytic competence is encouraged and supported. It is through exposure to different kinds of analysis, considerable field experience, a tradition of group

354

Making Work Visible

reflection and sustained participation in collaborative analysis sessions that analytic skills develop. For this, a key requirement is constant mutual mentoring within the group as well as one-on-one mentoring that pairs novices with experienced researchers over the length of an entire project. To learn to actually do analysis requires periodic but frequent collaborative data mining sessions during which trainees analyze their field notes, sketches, recordings, photographs, documents, and other collaterals, exploring multiple data strands through insistent questioning. It is here that participants ask each other questions like: What do you see? What would somebody else see? What are you looking for? What else could you be looking for? How do we know this? What evidence do we have? Where else could we look for evidence? Has anybody else explored this question? Routinely engaging in this exercise (as we did in the Interaction Analysis Laboratories) opens up linear thinking to considering other options, causes, and influences, and tracks the flow of knowledge, people, and technologies that keeps systems interconnected. One outcome of communal data analysis sessions is that analysts learn to respect empirical data. It is here that the common human tendency to speculate, to make sense of the world by imputing motives and causes, can be overcome through an insistence on honoring the data. In these sessions, trainees learn to always “go back to the data” in order to hold unbridled conjecture in check. It is precisely this ability to step back from current hypotheses and explanations that differentiates an analyst from a person with opinions. In the IALs there was an often-enforced rule: when a video was stopped for comments, nobody could talk for more than five minutes without going back to the data by turning the tape back on. This fosters reliance on thinking about what the real world is telling us and starves unrestrained flights of fantasy. In the long run, this should lead to developing what we might call the ethnographic stance: an attitude that simultaneously honors data and theory by pulling in concepts and patterns and checking them against the data that have been collected. Built on reflection and repetition, it is these group analysis sessions that are a major factor for growing fledgling analysts’ expertise in pattern recognition. Video is especially helpful in this regard. In addition, guided reflexive exposure to relevant articles is remarkably effective in establishing common analytic ground among members of ethnographic teams. It must be recognized that companies who aspire to this level of ethnographic competence need to understand that they must commit two crucial resources for this enterprise: time and money. Unless they are prepared to invest both they might be better off reaping the substantial benefits of a Level-1 transfer.

Transferring Ethnographic Competence

355

Level-3: Growing Strategic Competence Let me say up front that at the strategic level, the transfer metaphor breaks down completely. Strategic competence means to have significant influence at the top level of the company to the point of changing the company’s culture, as was achieved, for example, in the Systemic Assessment Project mentioned previously and in the Eureka Project described in detail in the chapter by Whalen and Bobrow in this volume. Strategic success requires a group of ethnographers who are committed to upgrade their influence on higher level company goals and presupposes a cohesive group of ethnographic experts, an “ethno-team,” who have acquired a deep understanding of the company’s work culture. Some of its members would probably be inhouse or at least have become deeply embedded in the target company during an extended period of work. Thus strategic competence cannot be taught, it has to emerge almost naturally from the prior two levels of competence, fueled by the vision of an ethnographically trained core group that aspires to optimize the impact of ethnography at the strategic level. It requires a sophisticated understanding of the micropolitical climate and skill in the corporate dance that allows the team to navigate simultaneously within the macroorganizational bureaucratic structures of company politics and policies and the micro-organizational experiential world of employees and consumers that work practice analysis reveals. A base requirement for strategic impact is buy-in and ownership at all levels of the target company. We have found that in every one of our projects that were successful there was a strong connection between PARC or IRL project leader(s) and their complement in the company. Strategic projects thrive under the care of internal brokers, midwives, and boundary crossers who help the project and its results acquire a social life. When projects address issues that people care about, that help them achieve their goals, then the chances for success increase. When inside ownership is lacking, excellent research simply does not make it beyond the PowerPoint presentations; reports get stuck on the credenza of the executive office and recommendations fall into the proverbial “black hole.” Because of the zigzagging goal slalom, typical for corporate projects, much potentially useful information is collected but never acted upon because the parts of the company that could use it are not involved. For example, in a project focused on automating Intel chip-making factories, several dozen recommendations were produced which would have substantially improved training and operational efficiency in the factories. These recommendations were never taken up by company shareholders whose interests were narrowly

356

Making Work Visible

focused on automation (Jordan and Lambert, 2009). Thus “marketing” ethnographic results, positioning ethnographic projects with the right counterparts in the client company, becomes another essential strategic activity. Finally, strategic competence requires a grounded understanding of why projects fail and what can be done to raise the likelihood of success. Our own experience parallels that of Microsoft researchers Donna Flynn and Tracey Lovejoy who, in an insightful paper, talk about how ethnography attained strategic influence at Microsoft (Flynn and Lovejoy, 2008). They suggest that a major success factor is alignment of ethnographic research with the company’s product development cycle, and propose that ethnographic work should begin well before technology gets started in order to provide designers with relevant data from the field.6 Ethnography at the strategic level fails when ethnographic projects are (or become) misaligned with the client company’s top level objectives and business initiatives. Most ethnographic projects draw on Level-1 and Level-2 competence. They are concerned with product or customer characteristics. To achieve strategic competence, however, a plan for phased strategic engagement has to be put in place that looks at the deployment of ethnographic projects over the company’s product cycle, creating a strategic engagement model for ethnography with a vision of end-to-end research across the product cycle that aligns customer understanding with product strategy. The truth is that, in our experience, strategic success has rarely happened. In the early days, anthropologically based ethnography had an almost strategic function under JSB. Our mere presence made it necessary to take the social into consideration and thereby changed the nature of the dialogue at PARC. This was a fortunate constellation that provided the space for us to mature our academically grounded field methods into a work practice methodology that became enormously influential in corporate research.

Conclusion So what role will ethnography and work practice analysis play in the future of ethnography and work practice analysis in the corporate workscapes and lifescapes of the twenty-first century? I envision several possible scenarios: The first sees ethnography splintering – fragmenting into an assortment of 6

Microsoft ethnographers say that in their environment they need to build knowledge of target customers and product opportunities twelve to thirty six months ahead of corporate product teams in order to have relevant information ready when the project actually starts up. This, of course, means that the e-team must have access to corporate initiatives in the pre-deployment stage in order to strategically position such projects.

Transferring Ethnographic Competence

357

techniques that are sold to the corporate market place by free-standing specialists; the second sees corporate ethnography expanding and extending its reach from workscapes to lifescapes and beyond, retaining a commitment to systemic, deep ethnography. And there is an obvious third possibility of a development where both approaches exist contentiously side-by-side, each claiming name, history, and territory. Unfortunately, along with the rise in requests for competence transfer has come an expanding number of providers eager to respond to a market where it has become de rigueur to splice in an “ethnographic component” whether a project needs it or not. Some of these practitioners are not well trained and operate with minimal experience, often specializing in some particular technique. Data collection becomes commoditized as methods are unpacked and practitioners sell expert status in this technique or that which, in the long run, may pull the life blood out of what makes ethnography compelling, potentially leading to ethnographic piece work, deskilling, and theoretical anemia (Lombardi, 2009). But there is a difference, too often ignored, between adapting traditional anthropological methods in all their richness and the claiming of expert status that all too often relies on rote application of one or more techniques. A prime example is video ethnography, originally a type of data collection and analysis that has deep theoretical and practice foundations (Jordan and Henderson, 1995). In some contemporary applications, this has deteriorated into “video bites,” the equivalent of voice bites that merely serve to illustrate and validate some predetermined conclusion. Thus video has become more of a marketing tool that aligns internal and client stakeholders but has little to do with discovery ethnography. It does not enable finding out something new, but merely validates previously determined issues. A second future scenario is about expansion. We started, now so many years ago, with taking on work and learning as serious research foci and called what we did “work practice analysis.” We have retained the label even though we came to realize it was too restrictive. We are, after all, analyzing much more than people’s work practice; we also look at how they use space and time, how they live and shape the culture of their organization, and how the tools they use are part and parcel of how they work. What we study thus looked more and more like a landscape that had varying features and areas of interest through which one could take many different paths. And so we started to think of ourselves as doing “workscape analysis.” Eventually, our studies morphed into what I called “lifescapes analysis” because as work became mobile and the boundaries between work and private life blurred, our methodology was again able to adjust. (Meerwarth et al., 2008).

358

Making Work Visible

The third and I believe most probable future scenario is a contentious coexistence of the two versions with each claiming to “do” ethnography. This development is foreshadowed by current debates in professional meetings, sometimes heated, of who can call her/himself a (corporate) ethnographer. Such territorial wars carry heavy disciplinary baggage and are likely to backfire in the long run. What is important is not to get into territorial battles by insisting on the superiority of one type of analysis over another but to always pull the discussion back to the adequacy of different methods for particular situations. My hope is that once ethnography manages to get beyond the current hump of outsized claims and faddish popularity, we will again be able to “adapt and adopt” time honored ethnographic methods and teach what we have leaned to a changing world. This is truly where work practice analysis belongs.

References

Ackerman, M. (2000). The Intellectual Challenge of CSCW: The gap between social requirements and technical feasibility. Human-Computer Interaction, 15. 179–203. Ambrose, S. E. (1997). Citizen Soldiers: The U.S. Army from the Normandy Beaches to the Bulge to the Surrender of Germany, June 7, 1944-May 7, 1945. New York: Simon and Schuster. Anderson, M. (2008). Top ten predictions for 2009. Strategic News Services Podcast: http://www.tapsns.com/media.php Anderson, R. J. (1994). Representations and requirements: The value of ethnography in system design. Human-Computer Interaction, 9(2), 151–182. Anderson, R. J. (1997). Work, ethnography and systems design. A. Kent and J. G. Williams (eds.) The Encyclopedia of microcomputers, 20, 159–183. Andreoli, J., Castellani, S., Boissier, O., Bratu, M., Alloui, I. and Megzari, K. (2003). E-Alliance: A negotiation infrastructure for virtual alliances. Group Decision and Negotiation Journal, 12(2), 127–141. Appadurai, Arjun, (ed.) (1986). The Social Life of Things: Commodities in Cultural Perspective. New York: Cambridge University Press. Aronson, M., Bishop, L., Cefkin, M., Jordan, B., Lawrence, N., Sullivan, L., Preston, C., and Oesterle, J. (1995). Reflections on a journey of transformation: learning, growth, and change at Xerox Business Services. Final report on the XBS Systemic Assessment Project, Institute for Research on Learning, Menlo Park, CA. Bannon, L. (1991). From human factors to human actors: The role of psychology and human-computer interaction studies in system design. In J. Greenbaum and M. Kyng (eds.), Design at Work: Cooperative Design of Computer Systems, pp. 25–44. Hillsdale, NJ: Lawrence Erlbaum. Barley, S. R. and Kunda, G. (2001). Bringing work back in. Organization Science 12: 76–95. Barley, S. R. and Kunda, G. (2004). Gurus, Hired Guns and Warm Bodies: Itinerant Experts in a Knowledge Economy. Princeton, NJ: Princeton University Press. Barth, S. (2000). Eureka! Xerox has found it. http://kazman.shidler.hawaii.edu/ eurekacase.html. Bell, D. G., Bobrow, D. G., Falkenhainer, B., Fromherz, M., Saraswat, V., and Shirley, M. (1991). RAPPER: The copier modeling project. International Logic Programming Symposium, San Diego, CA, October. 359

360

References

Bell, D. G., Bobrow, D. G., Raiman, O., and Shirley, M. (1997). Dynamic documents and situated processes: Building on local knowledge in field service. In T. Wakayama, S. Kannapan, C. M. Khoong, and S. Navathe (eds.), Information and Process Integration in Enterprises: Rethinking Documents, Norwell, MA: Kluwer Academic Publishers. Bell, G. and Dourish, P. (Forthcoming). Telling Techno-Cultural Tales. MIT Press. Bellotti, V., Begole, B., Chi, E., Ducheneat, N., Fang, J., Isaacs, E., King, T., Newman, M. W., Partridge, K., Price, B., Rasmussen, P., Roberts, M., Schiano. D. J. and Walendowski, A. (2008). Activity-based serendipitous recommendations with Magitti mobile leisure guide. Proceeding of CHI ’08. New York, NY: ACM Press. Bishop, L., Cefkin, M., Clancey, W., Jordan, B., Oesterle, J. and Tudor, B. (1994). A learning organization in the making: A report on the work practice and design project at the Xerox CAC in Dallas. Menlo Park, CA: Institute for Research on Learning. Blomberg, J. (1987). Social interaction and office communication: Effects on users’ evaluation of new technologies. In R. Kraut (ed.), Technology and the Transformation of White-Collar Work. Hillsdale, NJ: Erlbaum. Blomberg, J., Giacomi, J, Mosher, A., and Swenton-Wall, P. (1993). Ethnographic field methods and their relation to design. In D. Schuler and A. Namioka (eds.), Participatory Design” Principles and Practices. Hillsdale: Lawrence Erlbaum Associates. Blomberg, J. and Henderson, A. (1990). Reflections on participatory design: lessons from the trillium experience. Proceedings of the SIGCHI conference on Human Factors in computing systems: Empowering people, 353–360. Bobrow, D. G. and Whalen, J. (2002). Community knowledge sharing in practice: The Eureka story. Reflections 4(2), 47–59. Bodker, K., Kensing, F., and Simonsen, J. (2004). Participatory IT Design: Designing for Business and Workplace Realities. Cambridge, MA: MIT Press. Boland, R. J. J., and Collopy, F. (2004). Design matters for management. In R. J. J. Boland and F. Collopy (eds.), Managing as Designing, pp. 3–18. Stanford, CA: Stanford University Press. Boswell, J. (2006). The Project Gutenberg EBook of Life of Johnson. [EBook #1564] http://www.gutenberg.org/files/1564/1564.txt Bowers, J., Button, G. and Sharrock, W. (1995). Workflow from within and without: Technology and cooperative work on the print industry shopfloor. In H. Marmolin, Y. Sundblad, and K Schmidt, (eds.) Proceedings of the Fourth Conference on Computer-Supported-Cooperative Work -ECSCW ’95, pp. 51–66. Dordretch: Kluwer Academic Publishers. Brooks, F. P. (1995). The Mythical Man-Month: Essays on Software Engineering. Reading, MA: Addison-Wesley. Brown, J. S. and Duguid, P. (1991). Organizational learning and communities-ofpractice: toward a unified view of working, learning, and innovation. Organization Science 2, 40–57. Brown, J. S. and Duguid, P. (2000). The Social Life of Information. Cambridge, MA: Harvard Business School Press. Brun-Cottan, F. (2009). The anthropologist as ontological choreographer. In M. Cefkin (ed.), Ethnography and the Corporate Encounter: Reflections on Research in and of Corporations (pp. 158–181). New York: Berghahn Books.

References

361

Brun-Cottan, F. and Wall, P. (1995). Using video to re-present the user. Communications of the ACM, 38(5), 61–70. Burt, C. W. and Hing, E. (2005). Use of computerized clinical support systems in medical settings: United Stated, 2001–2003. Advance Data from Vital and Health Statistics, #354, March 15, 2005. Button, G. (2000). The ethnographic tradition and design. Design Studies, 21 (4), 319–332. Button, G. and Sharrock, W. (1994). Occasioned practices in the work of software engineers. In M. Jirotka and J. Goguen (eds.), Requirements Engineering: Social & Technical Issues (pp. 217–240). London. Academic Press. Button, G., Sharrock, W., Swenton-Wall, P., and O’Neill, J. (2003, July). The Work Practice Consultancy Toolkit. (Available from Xerox Research Centre Europe, 6 Chemin de Maupertuis, 38240 Meylan, France). Carlile, P. R. (2002). A pragmatic view of knowledge and boundaries: Boundary objects in new product development. Organization Science, 13(4), 442–455. Carlile, P. R. (2004). Transferring, translating and transforming: An integrative framework for managing knowledge across boundaries. Organization Science 15(5): pp. 555–568. Carstensen, P, Schmidt, K. and Will, U. K. (1999). Supporting shop floor intelligence: A CSCW approach to production planning and control in flexible manufacturing. Proceedings of Group ‘99, November 14–17th, Phoenix, USA, 111–120. Castellani, S. and Grasso, A. (2002). Mixing manual and computer supported initiative for scheduling and negotiation across distributed organizations. In Proc. of ECEC 2002, Modena, Italy. 15 April 2002. Castellani, S., Grasso, A. O’Neill, J. and Roulland, F. (2009). Designing technology as an embedded resource for troubleshooting. In the CSCW Journal Special Issue on CSCW, Technology and Diagnostic Work, 199–227. Cefkin, M. (2007). Numbers may speak louder than words, but is anyone listening? The Rhythmscape and Sales Pipeline Management. Proceedings of EPIC 2007: 188–199. Cefkin, M. (ed.) (2009). Ethnography and the Corporate Encounter: Reflections on Research in and of Corporations. New York: Berghahn Books. Chang, B-W. (1998). In-place editing of web pages: Sparrow Community-shared documents. Computer Networks 30(1–7): 489–498. Churchill, E. F. (2008). A site for soar eyes: (Re)placing cognition. In T. Erickson and D. McDonald (eds.), HCI Remixed. MIT Press. Clancey, W. J. (1997). Situated Cognition: On Human Knowledge and Computer Representations. Cambridge University Press. Clarke, K. M., Harwood, M. J., Procter, R. N., and Rouncefield, M. (2001). The electronic medical record and everyday medical work. Health Informatics Journal 7(3–4):168–170. Convertino, G., Billman, D., Pirolli, P., Massar, J. P., and Shrager, J. (2008). The CACHE Study: Group effects in computer-supported collaborative analysis. Computer Supported Cooperative Work 17: 353–393. Coulter, J and Parsons, E. D. (1989). The praxiology of perception: Visual orientations and practical action. Inquiry, 33: 251–72.

362

References

Crabtree, A., O’Neill, J., Tolmie, P., Castellani, S., Colombino, T., and Grasso, A. (2006). The practical indispensability of articulation work to immediate and remote help-giving. In P. Hinds and D. Martin (eds): Proceedings of the 2006 Conference on Computer Supported Cooperative Work, Banff, Alberta Canada, November 4–8, 2006 (pp. 219–228). New York: ACM. Crabtree, A., Rodden, T., Tolmie, P. and Button, G. (2009). Ethnography considered harmful. Conference Proceedings 27th Annual CHI Conference on Human Factors in Computing Systems, Boston April. pp. 879–888. Cramton, C. D. (2002). Attribution in distributed work groups. P. Hinds and S. Kiesler (eds.), Distributed Work, 191–212. Cambridge, MA: MIT Press. Cross, N. (2006). Designerly Ways of Knowing. London: Springer. Dant, T. and Francis, D. (1998). Planning in organisations: Rational control or contingent activity? Sociological Research Online, 3(2). http://www.socresonline .org.uk/socresonline/3/2/4.html Darke, J. (1979). The primary generator and the design process. Design Studies, 1(1), 36–44. Davenport, T. H. and Prusak, L. (1997). Working Knowledge: How Organizations Manage What They Know. Cambridge: Harvard Business School Press. de Kleer, J. and Williams, B. C. (1987). Diagnosing multiple faults. Artificial Intelligence, 32, 97–130. Diggins, T. and Tolmie, P. (2003). The ‘adequate’ design of ethnographic outputs for practice: Some explorations of the characteristics of design resources. Personal and Ubiquitous Computing, 7(3), 147–158. Doubler, M. D. (1994). Closing With the Enemy: How GIs Fought the War in Europe, 1944–1945. Lawrence, KS: University of Kansas Press. Dourish, P. (2001). Where the Action Is. Cambridge, MA: MIT Press. Dourish, P. (2006). Implications for design. CHI 2006, 541–550. Dourish, P. (2007). Responsibilities and Implications: Further thoughts on ethnography and design. Proceedings of DUX07 Designing for User eXperiences 2007, 25. Dourish, P., Holmes, J., MacClean, A., Marqvardsen, P., and Zbyslaw, A. (1996). Freeflow: Mediating between representation and action in workflow systems. Proceedings of CSCW ‘96. November 16–20, Boston, USA, 190–198. Drew, P. and Heritage, J. (eds.). (1992a). Talk at Work: Interaction in Institutional Settings. Cambridge: Cambridge University Press. Drew. P., and Heritage, J. (1992b). Analyzing talk at work: An introduction. In P. Drew and J. Heritage (eds.), Talk at Work (pp. 3–65). Cambridge, UK: Cambridge University Press. Dreyfus, H. (2009) On the Internet. 2nd Edition. London and New York: Routledge. Driver, E. (2006). Untethering Information Workers: Rethinking Workplace Location and Layout. Information Workplace Series, Forrester Research. Engestrom, Y., and Middleton, D. (eds.). (1998). Cognition and Communication at Work. Cambridge: Cambridge University Press. Fitzpatrick, G. (2000). Understanding the paper health record in practice: Implications for EHRs [online]. M. Pradhan, J. Warren, S. Chu, E. Coiera, and A. C. L. Zelmer (eds.), HIC 2000: Proceedings (323–331). Brunswick, Vic.: Health Informatics Society of Australia.

References

363

Flynn, D., and Lovejoy, T. (2008). Tracing the Arc of Ethnographic Impact: Success and (in)visibility of our work and identities. Proceedings of EPIC 2008, 238–250. Garfinkel, H. (1967). Studies in Ethnomethodology. Englewood Cliffs, NJ: PrenticeHall. Garfinkel, H. (1991). Respecification: evidence for locally produced, naturally accountable phenomena of order*, logic, reason, meaning, method, etc. in and as of the essential haecceity of immortal ordinary society, (I) – an announcement of studies. G. Button (ed.) Ethnomethodology and the Human Sciences (pp. 10–19). Cambridge: Cambridge University Press. Garfinkel, H. (ed.) (1986). Ethnomethodological Studies of Work. London: Routledge. Gaver, W. W., Bowers, J., Boucher, A., Gellerson, H. Pennington, S., Schmidt, A., Steed, A., Villars, N., and Walker, B. (2004). The drift table: Designing for lucid engagement. In CHI 0 04: CHI 0 04 extended abstracts on Human factors in computing systems, 885–900. Goldmacher, A. (2008). Located mobility: Living and working in multiple places. In T. L. Meerwarth, J. C. Gluesing, and B. Jordan (eds.), NAPA Bulletin 30, Mobile Work, Mobile Lives: Cultural Accounts of Lived Experiences (pp. 118–127). WileyBlackwell and American Anthropological Association. Goodwin, C. and Harness Goodwin, M. (1996). Formulating planes: Seeing as a situated activity. Y. Engeström and D. Middleton (eds.), Cognition and Communication at Work (pp. 61–95). Cambridge: Cambridge University Press. Gouws, L., Stephan, A. and Trevelyan, J. (2006). The nature of engineering maintenance work: A blank space on the map. J. Mathew, L. Ma, A. Tan, D. Anderson (eds.), Engineering Asset Management: Proceedings of the 1st World Congress on Engineering Asset Management (WCEAM) 2006, 452–459. Greenbaum, J. M., and Kyng, M. (1991). Design at Work: Cooperative Design of Computer Systems. Hillsdale, NJ: Lawrence Erlbaum. Grudin, J. (1993). Obstacles to participatory design in large product development organizations. D. Schuler & A. Namioka (eds.), Participatory Design: Principles and Practices (pp. 99–119). Hillsdale, NJ: Lawrence. Erlbaum. Grudin, J. (1994). Computer-supported cooperative work: History and focus. Computer 27 (5): 19–26. Haraway, D. (1991). Simians, Cyborgs, and Women: The Reinvention of Nature. New York: Routledge. Harper, R. (Forthcoming). Texture: Human Expression in the Age of Communication Overload. MIT Press. Headland, T., Pike, K., and Harris, M. (eds.). (1990). Emics and Etics: The Insider/ Outsider Debate. London: Sage. Heath, C. (1986). Body Movement and Speech in Medical Interaction. Cambridge: Cambridge University Press (Editions de la Maison des Science de l’Homme, Paris). Heath, C., Hindmarsh, J., and Luff, P. (2010). Video in Qualitative Research: Analyzing Social Interaction in Everyday Life. London: Sage. Heath, C. and Luff, P. (1996). Documents and professional practice: ‘Bad’ organizational reasons for ‘good’ clinical records. In Computer Supported Cooperative Work, 354–363.

364

References

Henderson, A. (1986). The trillium user interface design environment. ACM, SIGCHI bulletin, 1986, Vol. 17, Issue 4, Special issue: CHI ’86 conference proceedings. Heritage, J. C. (1984). Garfinkel and Ethnomethodology. Cambridge, UK: Polity Press. Heritage, J. and Atkinson, J. M. (1984). Introduction. J. J. Atkinson and J. Heritage (eds.), Structures of Social Action: Studies in Conversation Analysis (pp. 1–15). Cambridge, UK: Cambridge University Press. Hing, E. S., Burt, C. W., and Woodwell, D. A. (2007). Electronic medical records use by office-based physicians and their practices: United States, 2006. In Advance Data from Vital and Health Statistics, CDC, #393 October 26, 2007. Hughes, J. A.; Randall, D.; Shapiro, D. (1992). Faltering from ethnography to design. Proceedings of the 1992 ACM conference on Computer-Supported Cooperative Work. ACM Press New York, NY, USA. pp. 115–122. Hughes, J., King, V., Rodden, T. and Andersen, H. (1994). Moving out of the Control Room: Ethnography in System Design. Proceedings of Computer Supported Cooperative Work ’94, 429–239. Hutchins, E. (1995). Cognition in the Wild. Cambridge: MIT Press. Iida, Y. (2000). Between the technique of living an endless routine and the madness of absolute degree zero: Japanese identity and the crisis of modernity in the 1990s. Positions 8(2), 423–464. Ikeya, N., Awamura, N., and Sakai (2010). Why do we need to share information?: Analysis of collaborative task management meetings. In J. Foster (ed.), Collaborative Information Behavior: User Engagement and Communication Sharing (pp. 89–108). Hershey, PA: IGI Global. Ikeya, N., Vinkhuyzen, E., Whalen, J., Yamauchi, Y. (2007). Teaching organizational ethnography. Proceedings of Ethnographic Praxis in Industry Conference (EPIC), 270–282. Japan Statistics Bureau. Statistical Handbook of Japan 2007. http://www.stat.go.jp/ english/data/handbook/c02c ont.htm#cha2_4 Johnson, J., Roberts, T. L., Verplank, W., Smith, D. C., Irby, C. H., Beard, M., and Mackey, K. (1989). The Xerox Star: A retrospective. Computer, 22(9), 11–26, 28–29. Jordan, B. (1997). Transforming ethnography – Reinventing research. Cultural Anthropology Methods Journal 9(3), 12–17. Jordan, B. (2008). Living a distributed life: Multilocality and working at a distance. In T. L. Meerwarth, J. C. Gluesing, and B. Jordan (eds.), Mobile Work, Mobile Lives: Cultural Accounts of Lived Experiences (pp. 28–55). Wiley-Blackwell and American Anthropological Association (NAPA Bulletin 30). Jordan, B. and Brinda Dalal, B. (2006). Persuasive encounters: Ethnography in the corporation. field methods 18(4):359–381. Jordan, B. and Henderson, A. (1995). Interaction analysis: Foundations and practice. Journal of the Learning Sciences 4, 39–103. Jordan, B. and Lambert, M. (2009). Working in corporate jungles: Reflections on ethnographic praxis in industry. In M. Cefkin (ed.), Ethnography and the Corporate Encounter: Reflections on Research In and Of Corporations. New York: Berghahn Books.

References

365

Jordan, B. and Yutaka Yamauchi, Y. (2008). Beyond the university: Teaching ethnographic methods in the corporation. Anthropology News 49(6):35. Kaplan, A. M. and Haenlein, M. (2010). Users of the world, unite! The challenges and opportunities of social media. Business Horizons, 53(1), 59–68. Kayser, T., (1995). Mining Group Gold: How to Cash in on the Collaborative Brain Power of a Group. McGraw-Hill. Kishimoto, K., Shioda, T. Takeda, H. (2008). Using fieldwork to hand down skills of expert systems engineers. FUJITSU, 59(6). Kishimoto, K., Terasawa, M., Hirata, S. (2009). Work style innovation through business ethnography and organizational monitoring. FUJITSU, 61(6). Klein, M., Dellarocas, C., and Bernstein, A. (eds.) (2000). Special issue: Adaptive workflow systems. Journal of Computer Supported Cooperative Work, 9(3–4). Koenemann, J., and Belkin, N. (1996). A case for interaction: A study of interactive information retrieval behaviour and effectiveness. Proceedings of CHI ‘ 96, 205–212. Lave, J. (1988). Cognition in Practice: Mind, Mathematics and Culture in Everyday Life. Cambridge University Press. Lave, J., and Wenger, E. (1991). Situated Learning: Legitimate Peripheral Participation. Cambridge University Press. Lawson, B. (2006). How Designers Think: The Design Process Demystified (4th ed.). Oxford, UK: Elsevier. Levy, D. (2001). Scrolling Forward: Making Sense of Documents in the Digital Age. New York: Arcade. Li, C., and Bernoff, J. (2008). Groundswell: Winning in a World Transformed by Social Technologies. Boston: Harvard Business School Press. Lindquist, S., Westerlund, B., Sundblad, Y., Tobiasson, H., Beaudouin-Lafon, M., and Mackay, W. (2007) ‘Co-designing communication technology with and for families – Methods, experience, results and impact’ in Streitz, N., Kameas, A., and Mavrommati, I. (Eds.) (2007). The Disappearing Computer: Interaction Design, System Infrastructures and Applications for Smart Environments. Springer. Llewellyn, N. and Hindmarsh, J. (eds.) 2010. Organisation, Interaction and Practice: Studies in Ethnomethodology and Conversation Analysis. Cambridge: Cambridge University Press. Lombardi, G. (2009). The de-skilling of ethnographic labor: Signs of an emerging predicament. Proceedings of EPIC 2009, 41–49. Luff, P., Adams, G., Bock, W., Drazin, A., Frohlich, D., Heath, C., Herdman, P., King, H., Linketscher, N., Murphy, R., Norrie, M. Sellen, A., Signer, B., Tallyn, E., and Zeller, E. (2007). Augmented paper: Developing relationships between digital content and paper. N. Streitz, A. Kameas, and I. Mavrommati (eds.). The Disappearing Computer: Interaction Design, System Infrastructures and Applications for Smart Environments. Springer. Luff, P., Hindmarsh, J., and Heath, C. (eds.). (2000). Workplace Studies: Recovering Work Practice and Informing System Design. Cambridge: Cambridge University Press. Lynch, M. (1997) Scientific Practice and Ordinary Action: Ethnomethodology and Social Studies of Science, Cambridge: Cambridge University Press. Maegawa, K., Sugiyama, Y., Kang, S., Yamaki, K. (2009). Commercialization of fieldwork. Organizational Science, 42(4).

366

References

Manzenreiter, W. and Horne, J. (2006). Leisure and consumer culture in Japan. Leisure Studies, 25, 4: 411–415. Martin, David (2006). Who and what are electronic patient records for? An ethnomethodological ethnography of system development in the NHS. Proceedings of Symposium on Current Development in Ethnographic Research in the Social and Management Sciences. Liverpool University, 13–14th September, 2006. Martin, D. and Sommerville, I. (2004). Patterns of cooperative interaction: Linking ethnomethodology and design. ACM Transactions on Human-Computer Interaction. Vol. 11 Issue 1. © Association of Computing Machinery, pp. 59–89. Martin, D., Wastell, D., O’Neill, J., and Varey, R. (1999). Pushing all the right buttons: Lessons from two case studies of touch-screen kiosks. M. A. Sasse and C. Johnson (eds.), Interact’99. Proceedings of Interact, Edinburgh, Scotland, August 30 to September 3, 1999. IOS Press. Martin, D., Rouncefield, M., O’Neill, J., Hartswood, M., and Randall, D. (2005). Timing in the art of integration: That’s how the Bastille got stormed. Paper presented at Group 05, Sanibel Island, Florida, USA, November 6–9. Martin, D., O’Neill, J., Colombino, T., Roulland, F., & Willamowski, J. (2008). ‘Colour it’s just a constant problem’; An examination of practice, infrastructure and workflow in colour printing. D. Randall (ed.), Proceeedings of the 8th International Conference on the Design of Cooperative Systems (COOP 2008), May 20–23. Carry-le-Rouet: France. Martin, D., O’Neill, J. Colombino, T., Roulland, F. and Willamowski, J. (2010), Practical troubles with infrastructure and workflow in colour printing. From CSCW to Web 2.0: European Developments in Collaborative Design’ Springer: London pp 21–42. Martin, N. and Wall, P. (2008). The secret life of medical records: A study of medical records and the people who manage them, In Being Seen: Paradoxes and Practices of Invisibility (Ethnographic Praxis in Industry Conference) (15 October 2008) McGrath, D. (2009). Smartphone shipments to grow 6–11%, iSuppli says. EE Times. http://www.eetimes.com/news/semi/showArticle.jhtml?articleID=215800530 Means, J. and Adams, T. (2005). Facilitating the Project Lifecycle: The Skills & Tools to Accelerate Progress for Project Managers, Facilitators, and Six Sigma Project Teams. Jossey-Bass. Meerwarth, T. L., Gluesing, J. C., Jordan, B. (eds.) (2008). Mobile Work, Mobile Lives: Cultural Accounts of Lived Experiences. Wiley-Blackwell and American Anthropological Association (NAPA Bulletin 30). MIME (2001) Project Website, http://mime.cs.nott.ac.uk/ Moore, R. J. (2008). When words fail: Referential practice in face-to-face service encounters. Language in Society 37(3): 385–413. Moore, R. J. and Whalen, J. (forthcoming). Organizations in action: Work orders and document practices. N. Llewellyn and J. Hindmarsh (eds.), In Organization, Interaction and Practice. Cambridge: Cambridge University Press. Moran, T. P. (1989). What is EuroPARC? Proceedings of the Computer-Human Interaction, May. 51–52. Muller, M. J. (1991). PICTIVE–an Exploration in Participatory Design. Paper presented at the CHI ‘91: Proceedings of the SIGCHI conference on Human factors in computing systems: Reaching through technology.

References

367

Mumford, E. (1983). Designing Human Systems for New Technology: The ETHICS Method. Manchester, UK: Manchester Business School. Newman, S. (1998). Here, there, and nowhere at all: Distribution, negotiation, and virtuality in postmodern ethnography and engineering. Knowledge and Society 11: 235–267. Nikkei Electronics (2008). No. 970. Nonaka, I. and Takeuchi, H. (1995). The Knowledge-Creating Company: How Japanese Companies Create the Dynamics of Innovation. New York: Oxford University Press. O’Neill, J., Martin, D., Colombino, T., Roulland, F. & Willamowski, J. (2008). Colour management is a socio-technical problem. Proceedings of CSCW ‘08, November 8–12th, San Diego, USA, pp 599–608. ACM press. O’Neill, J., Castellani, S., Grasso, A., and Tolmie, P. (2005a). Using real-life troubleshooting interactions to inform self-assistance design. M. F. Costabile and F. Paterno (eds.), Proceedings of Interact, Rome, Italy, September 12–16, LNCS 3585, Springer, 377–390. O’Neill, J., Castellani, S., Grasso, A., Tolmie, P., and Roulland, F. (2005b). Representations can be good enough. H. Gellersen, K. Schmidt, M. Beaudouin-Lafon and W. Mackay (eds.), Proceedings of 9th European Conference on Computer Supported Cooperative Work, Paris, France, September 18–22, Springer, 267–286. O’Neill, J., Martin, D., Colombino, T., Watts-Perotti, J., Sprague, M., and Woolfe, G. (2007). Asymmetrical collaboration in print shop-customer relationships. L. J. Bannon, I. Wagner, C. Gutwin, R. H. R. Harper and K. Schmidt (eds.), Proceedings of ECSCW 2007. September 24–28th, Limerick, Ireland, pp. 231–250, London, UK: Springer. Orr, J. (1996). Talking about Machines: An Ethnography of a Modern Job. Ithaca, New York: Cornell University Press. Orr, J. (2006). Ten years of talking about machines. Organization Studies, 27(12), pp. 1805–1820. Philips Research, (2001). http://www.research.philips.com/technologies/projects/ ami/background.html. Plowman, L., Rogers, Y., and Ramage, M. (1995). What are workplace studies for? ECSCW. 309–324. Polanyi, M. (1966). The Tacit Dimension. New York: Doubleday and Company, Inc. Puffer, M. J., Ferguson, J. A., Wright, B. C., Osborn, J., Anshus, A. L., Cahill, B. P., Kamath, J., and Ryan, M. (2007). Partnering with clinical providers to enhance the efficiciency of an EMR. Journal of Healthcare Information Management 21(1):25–32. Randall, D., Harper, R. and Rouncefield, M. (2007). Fieldwork for Design: Theory and Practice. Springer Verlag, New York. Reason, J. and Hobbs, A. (2003). Managing Maintenance Error: A Practical Guide. Ashgate Publishing. Riordan, M. (2005). Variation in Premedia Color and the Potential Automation of Imaging Tasks. Technical Report – PICRM-2005–05, Printing Industry Research Center at Rochester Institute of Technology. Roam, Dan. (2008). The Back of the Napkin. New York: Penguin.

368

References

Rouncefield, M., Hughes, J. A., Rodden, T., and Villar, S. (1994). Working with “Constant Interruption”: CSCW and the Small Office. Computer Supported Cooperative Work ’94. pp. 275–286. Cambridge, MA: ACM. Ruhleder, K. and Jordan, B. (1997). Capturing distributed activities: Video-based interaction analysis as a component of workplace ethnography. A. S. Lee, J. Liebenau, and J. I. DeGross (eds.), Information Systems and Qualitative Research (pp. 246–275). Proceedings of the IFIP TC8 WG 8.2 International Conference on Information Systems and Qualitative Research. London: Chapman and Hall. Sacks, H. (1963). Sociological description. Berkeley Journal of Sociology, 8, 1–16. Sacks, H. (1987). On the preferences for agreement and contiguity in sequences in conversation. In J. R. E. Lee and G. Button (Eds.), Talk and Social Organisation, (pp. 54–69). Bristol, UK: Multilingual Matters. Sacks, H. (1992). Lectures on Conversation Vols I and II. (ed. Jefferson, G.), Cambridge, MA: Blackwell. Sacks, H., & Schegloff, E. A. (1979). Two preferences in the organization of reference to persons in conversation and their interaction. G. Psathas (ed.), Everyday language: Studies in Ethnomethodology (pp. 15–21). New York: Irvington. Sacks, H., Schegloff E. A., and Jefferson G. (1974). A simplest systematics for the organization of turn-taking for conversation. Language 50(4), Part 1, pp. 696–735. Schegloff, E. A. (2007). Sequence Organization in Interaction: A Primer in Conversation Analysis, Volume 1. Cambridge: Cambridge University Press. Schegloff, Emanuel A. and Sacks, H. (1973). Opening up closings. Semiotica 7: 289–327. Schiano, D. J.; Elliott, A.; Bellotti, V. (2006). Between cram school and career in Tokyo. Ethnographic Praxis in Industry Conference (EPIC) Proceedings, 2006(1): 155–159. Schiano, D. J., Elliott, A., and Bellotti, V. (2007). A Look at Tokyo Youth at Leisure: Towards the Design of New Media to Support Leisure Outings. Computer Supported Coop. Work 16, 1–2: 45–73. Schmidt, K. (1999). Of maps and scripts: the status of formal constructs in cooperative work, Information and Software Technology, 41.6, pp. 319–29. Schmidt, K. (2000). The critical role of workplace studies in CSCW. P. Luff; J. Hindmarsh; and C. Heath (eds.): Workplace Studies: Recovering Work Practice and Informing System Design (pp. 141–149). Cambridge University Press, Cambridge. Schmidt, K. and Bannon, L. (1992). Taking CSCW Seriously: Supporting Articulation Work. Computer Supported Cooperative Work 1: 7–40. Schön, D. A. (1987). Educating the Reflective Practitioner: Toward a New Design for Teaching and Learning in the Professions (1st ed.). San Francisco, CA: Jossey-Bass. Schön, D. A. (1990). The design process. In V. A. Howard (Ed.), Varieties of Thinking: Essays from Harvard’s Philosophy of Education Research Center (pp. 110–141). New York: Routledge. Schon, D. A. (1995). The reflective practitioner. Ashgate. Schuler, D., and Namioka, A. (eds.) (1993). Participatory Design: Principles and Practices. Hillsdale, NJ: Lawrence Erlbaum Associates. Schwarz, R. (2002). The Skilled Facilitator. Jossey Bass.

References

369

Scollon, R. and Wong Scollon, S. (2003). Discourses in Place: Language in the Material World. London: Routledge. Sellen, A., and Harper, H. R. (2003). The Myth of the Paperless Office. Cambridge, MA: MIT Press. Shannon, C. and Weaver, W. (1949). The Mathematical Theory of Communications. Urbana, IL: University of Illinois Press. Shapiro, D., Hughes, J., Harper, R., Ackroyd, S., and Soothill, K. (1991). Policing Information Systems: The Social Context of Success and Failure in Introducing Information Systems in the Police Service. Technical Report EPC-91–117. Rank Xerox Limited, Cambridge EuroPARC. Shirky, C. (2008). Here Comes Everybody: The Power of Organizing Without Organizations. New York: Penguin. Simon, H. A. (1968). The Sciences of the Artificial. Cambridge, MA: M.I.T. Press. Sprague, B. (2009). Customer service is crucial in a downturn. Business Week: May 15. Stewart, T. A. (1997). Intellectual Capital: The New Wealth of Organizations. New York: Doubleday. Strappers, P. J., Sanders, E. B-N. (2004). D. McDonagh, P. Hekkert, J. Van Erp, and D. Gyi (eds.) Design and Emotion: The Experience of Everyday Things (pp. 77–81). CRC Press. Strauss, A., S. Fagerhaugh, B. Suczek and C. Weiner (1985). Social Organization of Medical Work. Chicago, University of Chicago Press. Streitz, N., Kameas, A., and Mavrommati, I. (eds.) (2007). The Disappearing Computer: Interaction Design, System Infrastructures and Applications for Smart Environments. Springer. Suchman, L. (1983). Office procedures as practical action: Models of work and system design. In ACM Transactions on Office Information Systems, 1 (4), 320–328. Suchman, L. (1987). Plans and Situated Actions: The Problem of Human-Machine Communication. New York: Cambridge University Press. Suchman, L. (1993). Technologies of accountability: Of lizards and aeroplanes. G. Button (ed.) Technology in Working Order: Studies of Work, Interaction, and Technology. London: Routledge. 113–26. Suchman, L. (1995). Making Work Visible. In Communications of the ACM: Special Issue on Representations of Work, L. Suchman, ed. 38 (9): 56–64. Suchman, L. (1997). Centers of coordination: A case and some themes. L. Resnick, R. Säljö, C. Pontecorvo & B. Burge (eds.), Discourse, Tools, and Reasoning: Essays on Situated Cognition (pp. 41–62). Berlin: Springer-Verlag. Suchman, L. (2000). Making a case: ‘Knowledge’ and ‘routine’ work in document production. P. Luff, J. Hindmarsh and C. Heath (eds.) Workplace Studies: Recovering Work Practice and Informing System Design. Cambridge: Cambridge University Press. Suchman, L. (2007). Human-Machine Reconfigurations: Plans and Situated Actions. 2nd Ed. Cambridge: Cambridge University Press. Suchman, L., Blomberg, J., Orr, J. and Trigg, R. (1999). Reconstructing technologies as social practice. American Behavioral Scientist 43: 392–408. Suchman, L. and Trigg, R. H. (1991). Understanding practice: Video as a medium for reflection and design. J. Greenbaum and M. Kyng (eds.), Design at Work: Cooperative Design of Computer Systems. Hillsdale: Lawrence Erlbaum Associates.

370

References

Suchman, L., Trigg, R., and Blomberg, J. (2002). Working artifacts: Ethnomethods of the prototype. British Journal of Sociology 53 (2): 163–79. Szymanski, M. (1999) Re-engaging and dis-engaging talk in activity. Language in Society (1999), 28. pp. 1–23. Tolmie, P., Pycock, J., Diggins, T., Maclean, A. and Karsenty, A. (2002). Unremarkable computing. Proceedings of the 2002 Conference on Human Factors in Computing Systems, pp. 399–406, Minneapolis: ACM Press. Trigg, R., Blomberg, J. and Suchman, L. (1999). Moving document collections online. Proceedings of the Sixth European Conference on Computer-Supported Cooperative Work (ECSCW): 331–350. Ventres, W., Kooienga, S., Marlin, R., Vuckovic, N. and Stewart, V. (2005). Clinician style and examination room computers: A video ethnography. Family Medicine 37(4): 276–281. Viller, S. and Sommerville, I. (1999). Coherence: An approach to representing ethnographic analyses in system design. Human-Computer Interaction, 14: 9–41. Vinkhuyzen, E. (1999). Communication and coordination in mission control: A study of flight dynamics work practices. RIACS report 99.01. Vinkhuyzen, E., Szymanski, M. H. (2005). Would you like to do it yourself? Service requests and their non-granting responses. P. Seedhouse, K. Richards (eds.), Applying Conversation Analysis, 91–106. Basingstoke, UK: MacMillan. Vinkhuyzen, E., and Whalen, J. (2007). Expert system technology in work practice: A report on service technicians and machine diagnosis. S. Hester and D. Francis (eds.), Orders of Ordinary Action. Aldershot, U.K: Ashgate. von Hippel, E. (2005). Democratizing Innovation. Cambridge, MA: MIT Press. Wall, P., and Mosher, A. (1994). Representations of work: Bringing designers and users together. Proceedings of the Participatory Design Conference (Palo Alto, CA), pp. 87–98. Computer Professionals for Social Responsibility. Watts-Perotti, J., Sprague, M. A., Wall, P., McCorkindale, C., (2009). Pushing new frontiers. Examining the future of paper and electronic documents, in the Proceedings for the Ethnographic Praxis in Industry Conference, Chicago, IL. Weiser, M. (1991). “The computer for the 21st Century,” Scientific American, 265(3): 94–104. Weiser, M. (1994). The world is not a desktop. ACM Interactions, January: 7–8. Wenger, E. (1998). Communities of Practice: Learning, Meaning and Identity. New York: Cambridge University Press. Wenger, E., McDermott, R. and Snyder, W. (2002). Cultivating Communities of Practice: A Guide to Managing Knowledge. Cambridge: Harvard Business School Press. Whalen, J. and Vinkhuyzen, E. (2001). Expert systems in (inter)action: diagnosing document machine problems over the telephone. In P. Luff, J. Hindmarsh, and C. Heath (eds.), Workplace Studies: Recovering Work Practice and Information System Design (pp. 92–140). Cambridge: Cambridge University Press. Whalen, J., Whalen, M.; Henderson, K. (2002). Improvisational choreography in teleservice work. British Journal of Sociology. June; 53 (2). 239–258. Whalen, J., Zimmerman, D. H., and Whalen, M. R. (1988). When words fail: A single case analysis, Social Problems, Vol. 35, No. 4, pp. 335–362.

References

371

Whalen, J. and Zimmerman, D. H. (2005). Working a call: Multiparty management and interactional infrastructure in calls for help. In C. D. Baker, M. Emmison, and A. Firth (eds.), Calling For Help: Language and Social Interaction in Telephone Helplines. Amsterdam: John Benjamins. Whalen, M. and Whalen, J. (2004). Studying workscapes. P. LeVine and R. Scollon (eds.), Discourse and Technology: Multimodal Discourse Analysis, (pp. 208–229). Washington, DC: Georgetown University Press. Whalen, M. and Zimmerman, D. H. (1987). Sequential and institutional contexts in calls for help. Social Psychology Quarterly 50:172–85. Whalen, M. and Zimmerman, D. H. (1990). Describing trouble: Practical epistemology in citizen calls to the police. Language in Society, 19:4. World at Work. (2007). February Edition. Telework trendlines for 2006. A report by World at Work, based on data collected by the Dieringer Research Group. Yamauchi, Y. (2009). Power of Peripheral Designers: How Users Learn to Design. Paper presented at the 4th International Conference on Design Science Research in Information Systems and Technology, Malvern, PA. Yamauchi, Y., Whalen, J., and Bobrow, D. G. (2003). Information use of service technicians in difficult cases. In Conference Proceedings CHI 2003: New Horizons. Zimmerman, D. H. (1992). The interactional organization of calls for emergency. P. Drew and J. Heritage (eds.), Talk at Work (pp. 418–469). Cambridge, UK: Cambridge University Press.

Index

advisory board, 123–126 analytic competence, 352–354 Anderson, B., 3, 35 Anderson, S., 4 Barley, S., 9 Bauer, B., 4 Bentley, D., 36 Blomberg, J., 4, 27, 28, 75 Brown, J. S., 2, 26, 349 Brun-Cottan, F., 4, 75, 82 Burton, R., 2 Button, G., 3 computer supported cooperative work (CSCW), 9, 225–226 converging methods approach, 90 Conversation Analysis, 34, 81 de Kleer, J., 2 design directions document, 122 diary study, 89, 90–92, 103, 112–114 Dourish, P., 36 Drew, P., 9 Edwards, S., 346 embodiment, 67, 231–233 Ethnomethodology, 37–41 Eureka, xxiii–xxiv, 3, 257–284 EuroPARC, 3, 35

Henderson, A., 2, 3, 75 human-centered design, 7–8 Industrial Design/Human Interface group (IDHI), 4, 28, 75 Institute for Research on Learning, xxiv, 3, 344–348 interaction analysis laboratory, 345 interviews, 111–112 Jordan, B., xxi, xxiv, 2, 3, 32–33, 185 Levy, D., 2, 31 lifescapes analysis, 357 Luff, P., 36 member’s perspective, 65–67 Moran, T., 35 Mosher, A., 4 naturalistic observation, 5 Newman, S., 30 O’Brian, J., 35 organizational change, 280–282 Orr, J., xxii, xxiv, 28

field methods, 348–351

Palo Alto Research Center, 1, 2, 75, 258 PARC, Incorporated, 10 Phased Interactive Learning (PhIL), 15, 192–197 process v. practice, 159 Pycock, J., 36

Garfinkel, H., 2, 9, 41 Goodwin, C., 9, 30 Grasso, A., 3 Gyricon, 240

Raiman, O., 3 Raymond, G., 9 recorded data, 5–6 root cause analysis, 47, 48, 49, 165

Harness Goodwin, M., 9, 30 Harper, R., 35 Heath, C., 9

Sacks, H., 2, 34, 42 Scollon, R., 9 Smith, B., 2

373

374 sociotechnical systems, 161–162 Sparrow, 213 strategic competence, 355–356 Suchman, L., xxi, 2, 12, 75 Tolmie, P., 3, 13, 35 Trigg, R., 31 video podcast, 123 Wall, P., 4, 75 Weiser, M., 54, 55, 65, 72

Index Whalen, J., xxi, xxiv, 3 Whalen, M., xxi, xxiv, 3 Wong Scollon, S., 9 Workplace Project, 30–31 workscapes, 3, 6, 10, 207 Wynn, E., 2 Xerox Research Center Webster, 1 Xerox Research Centre Europe, 1, 3, 35–36, 60–62 Zimmerman, D., 9

LEARNING IN DOING: SOCIAL, COGNITIVE AND COMPUTATIONAL PERSPECTIVES

Computation and Human Experience Philip E. Agre Situated Cognition: On Human Knowledge and Computer Representations William J. Clancey Communities of Practice: Learning, Meaning, and Identity Etienne Wenger Learning in Likely Places: Varieties of Apprenticeship in Japan John Singleton, Editor Talking Mathematics in School: Studies of Teaching and Learning Magdalene Lampert and Merrie L. Blunk, Editors Perspectives on Activity Theory Yrjö Engeström, Reijo Miettinen, and Raija-Leena Punamäki, Editors Dialogic Inquiry: Towards a Socio-cultural Practice and Theory of Education Gordon Wells Vygotskian Perspectives on Literacy Research: Constructing Meaning Through Collaborative Inquiry Carol D. Lee and Peter Smagorinsky, Editors Technology in Action Christian Heath and Paul Luff Changing Classes: School Reform and the New Economy Martin Packer Building Virtual Communities: Learning and Change in Cyberspace K. Ann Renninger and Wesley Shumar, Editors Adult Learning and Technology in Working-Class Life Peter Sawchuk

Vygotsky’s Educational Theory in Cultural Context Alex Kozulin, Boris Gindis, Vladimir S. Ageyev, and Suzanne M. Miller, Editors Designing for Virtual Communities in the Service of Learning Sasha A. Barab, Rob Kling, and James H. Gray, Editors Bakhtinian Perspectives on Language, Literacy, and Learning Arnetha F. Ball and Sarah Warshauer Freedman, Editors Beyond Communities of Practice: Language, Power and Social Context David Barton and Karin Tusting, Editors Human-Machine Reconfigurations: Plans and Situated Actions, 2nd Edition Lucy Suchman Psychotherapy in Everyday Life Ole Dreier Thinking as Communicating: Human Development, the Growth of Discourses, and Mathematizing Anna Sfard Assessment, Equity, and Opportunity to Learn Pamela A. Moss et al., Editors From Teams to Knots: Studies of Collaboration and Learning at Work Yrjö Engeström Acting in Anaesthesia: Ethnographic Encounters with Patients, Practitioners and Medical Technologies Dawn Goodwin The Materiality of Learning: Technology and Knowledge in Educational Practice Estrid Sørensen How Students Come to Be, Know, and Do: A Case for a Broad View of Learning Leslie Rupert Herrenkohl and Véronique Mertl The Learning in Doing series was founded in 1987 by Roy Pea and John Seely Brown.