Peter M. Asaro

Beckman Institute

Departments of Philosophy and Computer Science

University of Illinois at Urbana-Champaign

asaro@uiuc.edu

4369 Beckman Institute

405 N. Mathews

Urbana, IL 61801

USA

TEL: (217) 352-7737

*The author wishes to thank Geof Bowker, Randi Markussen, Andy Pickering and Leigh Star for their many helpful comments on earlier drafts of this manuscript. An unpublished version of this manuscript has been mistakenly cited in numerous places as being a book of the same title published by the University of Illinois Press in 1996. While it has been interesting to trace the proliferation of this error, there is no such book.

Abstract

This essay attempts to shed historical light on some of the social, political, and ethical issues that have arisen from two disparate perspectives on technology which have both come to integrate an explicit consideration of social factors into systems design. It presents several distinct historical traditions which have contributed to the current field of Participatory Design (PD) methods-Joint Application Design (JAD^®), Engineering Codevelopment, British "socio-technical systems" and Scandinavian "collective resources" approaches to PD-and which in practice involved the end-users in the design process differently. It examines how their differences were based upon their differing perspectives on workers, differing professional relationships to technology, and differently stated goals. One interest in examining the independent development of methodologies within these perspectives is that, despite their many and fundamental differences, the various approaches ultimately converged on a considerably large set of shared concerns and very similar practices.

The paper also examines the relation of these approaches to transformations in the theorization of business organization and the manifold trend of corporate restructuring which helped to secure a place for variants of related user-involving methodologies in major U.S. and multinational corporations. It concludes with an examination of some broader issues addressing the relationship between technology and society and the prospects for critical technological practice. I argue that PD and its related methodologies are best understood as models of the integration of technological practice into social critique. Rather than seeing PD as merely the insertion of public dialog within technological design practice, as several observers have urged, we should see it as an exemplar of a critical practice of designing technological systems.

1. Introduction

2. North American Systems Design: early 1970s to mid 1980s

2.1 Background

2.2 IBM's Joint Application Design

3. European Systems Design: mid 1960s to mid 1980s

3.1 Background

3.2 The Scandinavian Approach: Collective Resources

3.3 The British Approach: Socio-Technical Systems

3.4 Participatory Design

4. Redesigning American Systems and Corporations: mid 1980s to mid 1990s

4.1 Background

4.2 The Re-invention of the American Corporation

4.3 Xerox's Engineering Codevelopment

4.4 Facilitating Communication Between Designers and Users

5. Reflections on Technology and Critical Practice

Everywhere we remain unfree and chained to technology, whether we passionately affirm or deny it. But we are delivered over to it in the worst possible way when we regard it as something neutral; for this conception of it, to which today we particularly like to do homage, makes us utterly blind to the essence of technology.

-Martin Heidegger from "The Question Concerning Technology"

1. Introduction

While technological "progress" is not without many vocal and compelling critics, the fact that technology permeates our society is undeniable. The insight into the deep connection between the form of technology and the form of human life, so eloquently expressed by Heidegger in the quote above, has led to some very interesting work in the design of socio-technical systems. This essay attempts to come to terms with some of the social, political, and ethical issues that arise from this work in a single, though broad, domain of socio-technological development: information systems design. I have chosen this particular domain because of an interesting convergence which has occurred. Beginning from disparate perspectives on technology, a field of subtly different methodologies have emerged for integrating an explicit consideration of social factors in systems design. At one pole the primary concern is with technological precision and product quality while the other pole is primarily concerned with the social impact due to the implementation of new technologies. The various methodologies ranging between these perspectives I will call Participatory Design (PD), and earn this description in virtue of the fact that they each seek to involve the users of an information system in the processes by which that system is designed. PD is generally more narrowly applied only to the Scandinavian tradition and its heirs, which are here considered as a branch of a larger family tree. I will use the term a broader sense to include the methodologies which bear a family resemblance in virtue of their explicitly involving users in their design practices. I will present several distinct approaches to design which have contributed to PD and which in practice involved users in different ways consequent upon the designers' differing perspectives on workers, professional relationships to technology, and stated goals. The methodologies to be discussed include Joint Application Design®, Engineering Codevelopment, Socio-technical Systems Theory, Collective Resources, and Participatory Design proper. I will then examine the subsequent development of these methodologies and their relation to transformations in the theorization of business organization and trends in corporate restructuring which helped to secure a place for variants of their central practices in major U.S. and multinational corporations. I conclude with an examination of some broader issues in the relationship between technology and society and the prospects for a critical study of technological practice. I argue that PD is best understood as an exemplar of a critical practice of designing technological systems.

My historical presentation traces the history of the two poles of PD which have only recently converged into the rather heterogeneous field of practices. The first of these poles is the development of the user-oriented design methods which originated in large North American corporations producing office information technologies, and take IBM as my principle example. IBM's design practices began systematically involving users in a methodology first developed in 1977. Called Joint Application Design (JAD^®),⁽¹⁾ it was derived as an extension of an existing IBM design methodology, Business Systems Planning (Carmel, Whitaker & George, 1993, p. 41). I begin the historical narrative with IBM because it represents a fairly linear extension of older design methodologies motivated by commercial goals. While not often recognized as a major contributor to PD, not only has JAD addressed the involvement of users in systems design, it also provides some insight into the corporate culture which would later adopt variants of the PD practices originating in Europe.

The other, and better recognized, pole which gave rise to the current field of PD had its roots in the post-war work of social scientists at the Tavistock Institute of Human Relations in London, but really began its historical development in 1960 with a series of four labor organization experiments called the Norwegian Industrial Democracy Project. That project led to two different research programmes: one in Britain, the "socio-technical systems" approach; and one in Scandinavia, the "collective resources" approach. These programmes subsequently grew back together in the early 1980s but didn't find a broad influence in North America for many years.⁽²⁾ In the literature, this recombined research programme is referred to as the "Scandinavian Approach," but I will refer to it as the European approach to acknowledge the contributions of British as well as Scandinavian researchers.

One interest in contrasting the independent development of methodologies from each end of such polar perspectives is that, despite their differences, they ultimately converged on a set of shared concerns and very similar practices. From a closer and more practical perspective, each approach recognized a set of problems surrounding the position of the "user" in systems design. Yet the articulation and resolution of these problems took very different turns and expressed different values. From a broader cultural and political vantage, these approaches have very different origins, and it is thus surprising that they should find as much common ground as they do. There are, of course, critical issues which arise as the two traditions grow closer together, and the details of this historical development become crucial to our critical understanding of it.

After recounting the history of PD, I will return to the transformations which occurred in American business and which further altered system design methodologies in the late 1980s. These transformations included the widespread adoption of PD-related principles and practices of user-involvement by North American corporations. This occurred first through the singular, though high-profile, work at Xerox Corporation, which was strongly influenced by the European researchers in the early 1980s, but continued to spread, and much more quickly, in the late 1980s this pattern of growth coincided with the more general movement of corporate restructuring. From among many similarly relevant projects during this later period, I examine a single project, called the "Class Project," to illustrate the impact of the changing conceptions of corporate organization on design methodology.

The methodology which emerged out of the Class Project was called Engineering Codevelopment (EC). The project, sponsored by the Commission on Preservation and Access (a private, non-profit organization), began in 1989 and was a joint venture between Xerox and the Cornell University libraries to develop proprietary digital-image technologies for the preservation of, and on-line access to, delicate rare books in the libraries' collections. Unlike the two principal traditions we will investigate, JAD and PD, it is not really a tradition so much as an exemplar of an interesting design perspective which lies somewhere in field of convergence between the North American and European approaches. Xerox's design methodologies were numerous and varied, arising out of a more academically and experimentally oriented research perspective than was typical of most American companies. Focusing in the early 1980s on cognitive ergonomics and human-computer interaction at its Palo Alto Research Center (PARC), and by the late 1980s on systems design organization with its newly created Work Practice and Codevelopment Group, Xerox spanned the poles between the North American and European approaches. I examine in detail only one of the methodologies in this middle area because it represents an interesting point in the convergence between these approaches-researchers at Xerox were clearly influenced by both approaches. It thus serves as a useful point of reference in understanding how involving users reshapes design practices and, because it represents many of the features sought after by management theorists in their discourse on Business Process Reengineering, it also provides an illustration of one way in which PD methods were able to align with this field of discourse. I chose to examine Xerox's Class project, despite Xerox's consideration of this project as a failure, because it provides insights into the complex social factors involved in technological design, and the ways in which these factors influenced design practice.

These two main methodological approaches to systems design embodied traditions of systems design practice which held very different perspectives on technology and the role of technology in the workplace. This led to very different ways of conceiving of the "problem" of integrating technologies into the workplace. Moreover, each perspective and set of problems developed in very different contexts; these arose in different kinds of organizations-academic, public, trade-union, commercial-and in situations which placed designers and users in very different relationships. There are two interesting phenomena which resulted from this and which I wish to emphasize. The first is that at some point, within each of these traditions it was decided that involving users systematically in the design of technological systems was central to achieving desirable goals. The second is that despite seeing the users' participation in the design process as an essential local objective, how each tradition conceived of the users' role in that process was shaped by their global objectives. This configuration turns out to bare a striking resemblance to the situation which confronted anthropologist when reflecting upon their role in colonialism, and Participatory Design researchers came to address many of the same critical issues. With these issues in mind, I turn now to the histories of these different approaches to user participation in systems development.

2. North American Systems Design: early 1970s to mid 1980s

2.1 Background

Systems design grew out of traditional engineering design in large organizations and sought to develop proprietary systems according to explicitly formalized "requirements." The consequence of this heritage was a highly structured and procedural approach to design.⁽³⁾ If users needed to be consulted, this was done unsystematically through informal interviews, a process called "requirements gathering," from which designers derived a set of system requirements. Systems design was seen as an engineering task to design the best possible system for a given set of needs, and even in the stage of requirements gathering expert designers employed a strong technical rationality. Winograd and Flores have identified the maxims governing the rationality of this computer culture:

1. Characterize the situation in terms of identifiable objects with well-defined properties.

2. Find general rules that apply to situations in terms of those objects and properties.

3. Apply the rules logically to the situation of concern, drawing conclusions about what should be done. (Winograd & Flores, 1986, p. 15)

Thus, establishing system requirements was seen as a more or less straightforward and value-neutral step occupying designers only briefly in the earliest stages of the overall development process, where the real skill lay in designing the best system for the given set of requirements. The result of this step-wise development process, according to Winograd and Flores as well as most of the systems designers from the fields we are about to examine, was a preference objective, formal, and reductive descriptions of the system's intended user-context and did not stop to reconsider these descriptions during the later stages of design.

2.2 IBM's Joint Application Design

The development of Joint Application Design (JAD) is itself a response to the difficulties encountered by rationalistic system design methods. The conventional wisdom of systems design was embodied in Business Systems Planning, IBM's principle development methodology during the 1970s. A classic problem for any form of centralized planning is that those doing the planning are too far removed from the activities of those the plans are being made for, due to a lack of effective communication. In IBM's case, system designers were finding it difficult to formulate system requirements from their labs while users were being frustrated by systems that failed to suit the needs of their office. Informed by insights in group dynamics and social psychology, JAD was developed in 1977 by IBM employees Chuck Morris and Tony Crawford as an extension of the existing design methodology. The intended objective of the methodology was to reduce the time required for the System Development Life Cycle (SDLC) while, hopefully, simultaneously increasing quality and reducing overall costs. The impetus for this methodological change was thus only minimally theoretical and overwhelmingly practical.

To achieve these goals, the methodology sought to integrate structured meetings between designers and users into the SDLC; it is these meetings which stand as the essential defining characteristic of JAD. The meetings were arranged so as to occur several times at the early stages in the SDLC, where meetings in the earliest stages focused on high-level user concerns and objectives while meetings in the later phases of design demanded increasingly detailed and specific information from users, with the ultimate goal of creating a single Design Document. The JAD Design Document was intended to provide a basis for specific system requirements and consisted of a list of user requirements approved by everyone attending the meetings, thus constituting both an object of group consensus and a technical resource for design.

A universal feature of JAD meetings was that they were highly structured by a concern for maintaining social control in situations which might have otherwise called into question the relationships between experts, managers, and workers or digressed into an unproductive chaos:

The JAD methodology emphasizes structure and agenda. This is evident in the JAD literature that reads somewhat like cookbooks. Everything is explained in great detail: "to do" lists are included, as are masters of useful forms. There are four necessary building blocks for a well-run JAD meeting:

1. Facilitation. A designated leader (or leaders) manages the meeting. Some JAD practitioners consider the meeting leader to be key to process success, even more so than the act of gathering the users in one place, the essence of JAD.

2. Agenda setting/structure. The meeting must have a plan of action.

3. Documentation. One or more designated scribes carefully document everything in the meeting. Various lists are rigorously maintained.

4. Group Dynamics. Group dynamics techniques are used for inspiring creativity (e.g., brainstorming), resolving disagreements (e.g., airing facts, documenting them as "issues," taking notes), and handling speaking protocols (e.g., enforcing "only one conversation at a time"). (Carmel, et al., 1993, p. 41)

In addition to the facilitator and scribe, the key individuals involved in these meetings were users and designers. It is important to note that the "users" in these meetings were supposed to consist of managers and veteran workers with detailed knowledge of the work process. The implication is that "user satisfaction" consisted in not only satisfying the requirements of the work process, but in satisfying the interests of the people in charge of overseeing and managing those processes. As Carmel, et al. (1993) report,

we have observed numerous North American JAD meetings in which operational employees are overlooked as participants. This results in a meeting room filled with middle managers and supervisors unable to specify details of day-to-day operations (e.g., what 17 fields are needed to fill out form A345). This organizational failure stems in part from an unjustified lack of confidence that "front-line" workers can meaningfully contribute to the design process. (p. 46)

This raises many issues regarding the implications of information technology in organizational control and the politics of the workplace which will be addressed more carefully in the final section.

Besides limiting the voice of the worker as a "user" through the explicitly management-dominated organization of meetings, JAD also served to protect and promote the authority of technical experts. Indeed, the ostensive objectives of the designers are embodied in the structured nature of the meetings by their simultaneously performing two functions: 1) the extraction of knowledge-beliefs, impressions, desires-from users in a controlled fashion through designer-established agendas, and 2) the rationalization or "selling" of the system to users by design engineers. Function 2) was achieved in part through the use of elaborate visual aids⁽⁴⁾ which sought to enable design experts to, among other things, describe their system to users and justify technical constraints and in part because everyone in attendance at the JAD sessions was considered to have "signed off" on the Design Document those sessions produced. The literature seeking to improve on the JAD methodology generally focused on slight alterations in the presentation aids, forms, or the overall organization of the meetings themselves. The dual-function of the JAD meeting allowed the technical experts to represent users' needs as objective information in the technical design phase by using the data contained in the JAD Design Document. Thus the technical design process could largely retain its rationalistic procedures while the users' influence on design was conveniently reduced to a well-structured functional input to the design process, a process which always remained in the control of the expert designers. It is precisely this highly structured nature of the process which was touted by how-to books on JAD (August, 1991; Crawford, 1994; Wood & Silver, 1989). So while JAD did seek to integrate users into the design process, it was unwilling to call into question or transform the fundamental technical rationality, practices, and political organization of that process.

3. European Systems Design: mid 1960s to mid 1980s

3.1 Background

The European approach has its roots in a very different socio-political sensibility. The Norwegian Industrial Democracy Program consisted of four experiments carried out by researchers from the Tavistock Institute of Human Relations and the Norwegian Work Research Institute between 1964 and 1967 (Emery & Thorsrud, 1976). These studies investigated how social groups organized around production technologies and sought to reform job distribution and wage systems for workers. After these four experiments, two research programmes developed along different trajectories: Scandinavian researchers focused on union empowerment through "collective resources" and British researchers focused on autonomy in work-group organization through "socio-technical systems design." Each felt that they had chosen the most promising set of objectives for what they saw as being feasible in the democratic reform of workplace technology and hence each saw the objectives of the other as tangential to the central issue. The British saw a union-centered approach as only being viable in the political environment of Scandinavia and as failing to theorize the organization of labor on a fundamental level, while the Scandinavians saw group dynamics as being ineffectual because it failed to consider the predominating power struggles of class and capital. Both approaches, however, were motivated by a shared concern for workplace democracy and the humanization of work and both contributed to the broader Quality of Working Life movement then beginning to take shape.⁽⁵⁾

3.2 The Scandinavian Approach: Collective Resources

The Scandinavian collective resource approach originated when the Norwegian Computer Centre (NCC) began working with the Norwegian Iron and Metal Workers Union in 1970 to educate union officials on how technology affects working conditions and might be made to serve union interests. The expressed goal was to assist unions in devising technological control activities and policies. The basic methodology was to set up union mechanisms to gather and analyze information about specific technologies and their effects on workers, with the belief that by doing so unions could offset the employers' natural advantage in technological knowledge and make it possible for them to put technological issues on the bargaining table. Research revolved around an elaborate "negotiation model" which sought to depict the bureaucratic process of introducing new technologies on the shop-floor with the expectation that such a model would allow trade unions to intervene in response to management's technological proposals (Ehn & Kyng, 1987, p. 42). The "collective" here indicated that the intention was the empowerment not of individuals in their workplace but of the trade union collective in bargaining situations, while "resources" indicated the emphasis on the information resource gathering of trade unions. The initial projects did not seek to involve the workers directly into the design process, nor did they identify the design process as a particularly significant locus of interest. In fact, worker participation (in the processes of information and requirements gathering) was seen as problematic in that 1) the workers involved might become experts and join management thereby threatening union solidarity, 2) such participation could give management undue access to "shop floor information," 3) it could prevent effective trade union participation, or 4) that it could even become a managerial strategy for worker manipulation (Ehn & Kyng, 1987, p. 40). Here the concern was not in the democratization of the technological design process, but of the bureaucratic decision-making process through which a company would seek to introduce a new technology on to the shop-floor.

By the nature of the Norwegian Work Environment Act (1977), which provided participatory rights to all (not just unionized) workers, issues of workplace democracy were seen as requiring locally specific actions and solutions, which resulted in several highly specific local projects. And even though the law provided for individuals' rights in workplace co-determination, due to the political relationship which existed between managers, workers and unions, only union-initiated activity was seen as having a viable impact on workplace organization. This was also a consequence of the theoretical framework which motivated the researchers, who saw themselves as trying to find viable alternatives to the Tayloristic rationalization of work. Inspired by Marxist critiques of the technological rationalization of work from authors like Braverman, Noble and Winner,⁽⁶⁾ they believed that unions were the only viable point of resistance to the otherwise inevitable capitalist processes of deskilling and increasingly centralized control through the division of labor, even though they expressed some doubts:

However, there were also practical and empirical anomalies that could not be explained by these theses. Work was not deskilled in all cases. More collective forms of work organization than the Tayloristic were sometimes proposed by management. It happened that workers gained from the introduction of new technology, etc. But this does not mean that the Marxist approach to understanding changes of the labor process in a capitalist economy has to be rejected. (Ehn & Kyng, 1987, p. 36)

Yet it was precisely this insistence on the Marxist critique which had motivated their rejection of other theoretical approaches up until the continuance of their own work necessitated a new theory of the design process. And it was an explicit theorization of design processes that the Scandinavian researchers were to later discover in "socio-technical system design" and which would mend the schism between Scandinavian and British researchers after a decade of independent work.

Early collective resource studies targeted heavy manufacturing technologies, which in themselves were not particularly flexible, and areas where the unions had found it difficult to translate their interests into negotiable demands. Researchers and union officials studied and analyzed workers' feelings about existing technologies (through surveys, union meetings and other methods adapted from psychoanalysis), lobbied and bargained to require employers to disclose information about technological reorganization proposals (such as the technical specifications and organizational policies involved in introducing numerically controlled machines and computer-based planning systems into the production line), and produced textbooks and formed classes for the education of union officials and workers (Ehn & Kyng, 1987, p. 28). These methods met with limited success in achieving those goals which could be easily formulated as collective demands, such as requiring the retraining of workers displaced by a new technology, but it was difficult to make any gains on qualitative humanistic concerns regarding technology through bargaining. The approach failed to spread as a general union movement because most local unions could not spare the necessary time and financial resources required to make it work.⁽⁷⁾

A shift in industrial focus and the conception of design came about with the "second generation" of the Scandinavian approach marked by the Swedish-Danish UTOPIA project in 1981, the first recognizable PD development project. Conceived in response to the discouraging results of the earlier trade union projects which had found that the lack of flexibility in existing technologies limited the possibilities of workers to influence workplace organization, UTOPIA targeted technological development as a prospective site for worker involvement and influence. According to the Marxist critiques, the technological dehumanization of work through deskilling, intensified division of labor, rigid and routine practices and the shifting of control toward the top of organizations was an inevitable result of the introduction of new technologies which necessarily served the interests of management and owners. Since the existing technologies were presumably all being developed to satisfy the interests of their purchasers, the business owners, and hence to increase productivity, control, and efficiency, the only effective means of empowering workers in competitive industrial markets was seen to be the creation of alternative technologies designed around workers' interests. Thus, the researchers sought to realize the ideals of the Scandinavian Technology Agreements established in the late 1970s, which had empowered workers with control over workplace technology in writing but not in practice, by designing a technological system with workers' skills and interests in mind (Greenbaum & Kyng, 1991, p. 11).

In cooperation with the Nordic Graphic Workers Union, the UTOPIA (both an acronym and an ideal) project studied a group of newspaper typographers working without computer support in order to develop a state-of-the-art graphics software product for skilled graphics workers. The objective was to create a commercial product which the unions could then demand as an alternative to other deskilling technologies. The commercial product was ultimately unsuccessful due to a small market and the shortcomings of the company which owned and marketed the software. The completion of this project in 1985 coincided with a renewed acceptance of the socio-technical systems design research being conducted in Britain.

3.3 The British Approach: Socio-Technical Systems

The British researchers were interested in the phenomena of group dynamics that had been observed in the early Tavistock inquiries into "leaderless groups," originally motivated by founding member Eric Trist's personal fascination with the efficiency of German panzer tank divisions (Mumford, 1987, p. 61) and Britain's War Office Selection Boards' interest in choosing and training military officers in order to utilize the phenomena of emergent group dynamics (Trist, 1993, p. 42).⁽⁸⁾ The actual experiments involved observing coal-miners and how they organized their labor practices around technological systems. Miners working in large shafts were organized around assembly-line type machinery and showed little variation in their repetitive tasks while miners working in small shafts or exposed faces, where the machinery could not be installed, developed novel dynamic and efficient work practices. Amenable to psychological and biological theory,⁽⁹⁾ the researchers developed their "organic" theoretical approach from systems theory, and focused particularly on the notion of "open systems," which itself drew heavily on a bio-organic metaphor (Mumford, 1987, p. 65). Thus work organization was theorized as an organic relationship between workers and technology-the socio-technical system-which ought to be analyzed according to criteria of "health" instead of the raw productive output measures of rationalized and mechanistic analyses.

From their studies of miners working under Tayloristic management, they concluded that the observed inefficiencies resulted from optimizing the mechanical aspects of the system at the expense of the human aspects. The optimization of the socio-technical system as a whole would thus require a joint optimization of both aspects with an eye towards the social and psychological impact of technology on workers (Mumford, 1987, p. 63). Research focused on the concept of the "autonomous work group" in which workers were allowed to spontaneously develop their own work routines, make decisions, and change tasks with little or no supervision.⁽¹⁰⁾ It was this theoretical conception of the production process that would motivate much of the later development of PD.

During the time that the Scandinavian researchers were concerned with empowering unions, the British researchers had been developing "socio-technical design principles" and management philosophies.⁽¹¹⁾ It was just these projects which were at the time being highly criticized by Marxist theorists in Scandinavia for promoting values that were fundamentally capitalist-increasing productivity and decreasing worker resistance. While these criticisms carried some force with respect to how the principles were often applied within the workplace, they still depended on a conception of the labor process as a mechanism driven by capitalist interests pushing against humanist concerns under great pressures which left no real space for compromise, and always threatened to be just about to burst.

It was against this same background of concerns that a general international social movement, called the Quality of Working Life movement, began demanding more humane work environments and transforming the general conception of a worker's relation to their work away from a purely economically motivated conception to one which included an emotional investment in, and personal attachment to, work. This was in part achieved through a reconception of the identity of the worker as an enterprising agent seeking personal fulfillment through job satisfaction. Miller & Rose (1995) have argued that by conceiving of the labor process as an organic process in which the most productive system was the healthiest and happiest system, socio-technical systems research made a significant contribution to this movement by offering a space for the negotiation of humanistic and economic concerns. They go on to argue that it is not as important to note the success or failure of the movement as to recognize the alignment of ethical, political, economic and technological elements through which it was possible for the Quality of Working Life movement to establish a new identity for the worker and to restore the legitimacy of the corporation in industrial democracies.

The alignment of heterogeneous conceptions of work in the Quality of Working Life movement also opened a novel space for theorizing technology in which it could hence be seen as serving multiple interests and values. So while it had been observed by Marxist critics like Noble and Braverman that technology could exploit and subjugate workers, it was now seen as also being able to promote workers' interests by making work more interesting, reintroducing skill, and by making practically feasible the "autonomous work group." The key to this lay in granting the worker direct control over the nature of the technology encountered in their day-to-day job. This also allowed the Scandinavian researchers to explain why the trade union approaches had failed-because the available technologies were not being designed in step with the new conception of workplace autonomy, union-supported technology had been just as mechanistic as management-supported technology. By 1985 the British and Scandinavian traditions had rejoined under a common banner of democratizing technological systems design. The consequence was to be an increased emphasis on the involvement of the worker in technological design, which had already begun in the UTOPIA project. This was to be the essential feature of the tradition from that point on.

3.4 Participatory Design

PD researchers encountered two main barriers to the successful participation of users/workers in the design process. The first of these was a lack of appreciation by workers for their own knowledge of what they do-as one researcher reported,

It is a widespread opinion among workers that they themselves know nothing about technology, and that the necessary information must be obtained from management. This paralyzes the workers as far as actions are concerned. . . .[It] is at least as important to collect and prepare the knowledge of the workers, a knowledge they have obtained through their jobs. (Kensing quoted in Clement & Van den Besselaar, 1993, p. 29)

There was also a reluctance among technical experts to give project control to users, as this threatened their technical authority and traditional work practices. As a worker in one project remarked to the researchers, "But you don't always listen to us-you do what you think is right for us and the project. And, you are the expert; so who are we to dispute your decisions?" (Nurminen & Wier, 1991, p. 297). These two barriers together constituted the socio-political resistance to the democratization of the design process. Both users/workers and design experts found it difficult to leave their traditional socio-political roles in order to participate as intelligent and capable equals in a shared project. The common response to this problem was to send the experts "into the field." Rather than trying to rely exclusively upon special interviews or meetings to learn about users' work practices (as the union projects had done), researchers utilized "action research" methods whereby they spent a great deal of time observing and interacting with workers in their workplace. The reported consequences of this were an enhanced appreciation on the part of both workers and experts for workers' knowledge, and an increased understanding by workers of technology and its influence on their work practices.

Because the second generation PD projects sought to establish democratic participation among the workers influenced by a given technology, they saw their initial objective as breaking down traditional concepts of work and expertise among the design group (of both workers and experts). A second key concern for practitioners of PD was the enlisting of external support for their projects and methods. Almost all projects reported resistance within or friction between the different organizations involved in the projects. Also, due to the nature of the legal, political and economic conditions in which the European projects were situated, they sought to realize their objectives at a highly local contextual level, resulting in few immediately generalizable design principles. The confluence of these factors had led to a very nebulous concept of just what was entailed in utilizing a PD methodology-it started to become an ideological approach rather than a prescriptive set of techniques. As a consequence of this, North American systems designers were very skeptical of PD and worried that it was only viable in a context such as that which existed in Scandinavia. This is to say that as a design philosophy, during most of the 1980s PD was unable to "sell itself" to a North American market still committed to its own tradition of software engineering practices and ideals. Skeptical system designers asked what methods it prescribed and what techniques it utilized, while PD's proponents insisted that the key factors were the promotion of democratic ideals and an enormous creative effort on the part of designers.

4. Redesigning American Systems and Corporations: mid 1980s to mid 1990s

4.1 Background

Several changes in the late 1980s and early 1990s were to alter the way in which Participatory Design and other user-involving techniques were perceived and which would both transform it and carry it across the Atlantic. Early Scandinavian PD focused purely on democratic participation and on overcoming various difficulties in achieving this, moreover it was primarily conceived and pursued by an academic community that held little concern for business interests and only limited responsibility for producing actual systems. This was in turn a very different set of problems from those which gave rise to JAD and similar methodologies in North American corporations, where the problems addressed by system design professionals focused on promoting business goals by increasing the efficiency and effectiveness of technical design. The interesting exception to this was Xerox PARC which was an early supporter of PD in North America, though it stood in relative isolation in this regard. The consequence was that for more than a decade both JAD and PD developed in isolation with almost no interaction between developers working in the two traditions. JAD grew up in the halls of North American industry⁽¹²⁾ and received very little academic attention, while PD was conceived by academics and grew up in the progressive social democracies of Europe.

In recent years, the two traditions have grown close enough together to find themselves publishing in the same professional journals and exchanging techniques and tools. This convergence was not the result of deep theoretical insights stemming from either side, but of separate movements by each toward filling a space opened by a newly emerging organizational regime. This new regime, described by Agre (1995) as the "empowerment and measurement regime,"⁽¹³⁾ was constituted by elements taken from management theories concerned with product quality and business process efficiency, accountancy methods seeking more precise cost and expenditure measurements, and nationalist political rhetoric over the global competitiveness and security of American industry and its workers. This regime not only transformed the discourse of management theory, when combined with global information networks, this new regime made possible global factories and very large distributed control systems through modularized organization, out-sourcing, and the electronic transmission of office communications, design requirements, software products, and even programming labor. Co-evolving with and informed by these changing conceptions of workers and business organization, the current heterogeneous field of PD claims the twin goals of increasing efficiency (of both technical experts and users) and increasing democracy (primarily for users).

4.2 The Re-invention of the American Corporation

In the mid 1980's, North American business experienced a trend toward maximizing the efficiency and flexibility of its organization.⁽¹⁴⁾ This movement toward "reinventing the corporation"⁽¹⁵⁾ led to a series of hot topics in professional management literature, the most significant to PD being Total Quality Management (TQM) and Business Process Reengineering (BPR). While these management theories each claim principled distinctions between them, their historical development reflects their joint participation in a larger trend and they have been combined along with other elements under the rubric of corporate restructuring and down-sizing. More importantly, they each constitute a field of discourse which has given shape to the empowerment and measurement regime (Agre, 1995) and the impact of each can be seen distinctly on the face of PD literature.

This led to the development of a Requirements Management reengineering team which sought to eliminate the "churn" by forming cross-functional design groups (with members from marketing, service, management, and customer representatives in addition to engineers), restructuring the requirements-gathering process into an iterative "listen-define-validate" model that relies on continual feedback instead of an initial exhaustive establishment of requirements (Hutchings & Knox, 1995, p. 74). Though not all companies approached this new problem in the same way, this is essentially the basic process by which many North American companies began using more participative design methodologies.⁽¹⁷⁾ Through a demand for the representation of the user/customer in the design process, BPR and TQM transformed both the problematic of designing new work organizations and the standards of value for judging the design process and its products.

4.3 Xerox's Engineering Codevelopment

In many ways, Xerox's Engineering Codevelopment (EC) exemplifies the impact of socio-technical system design and customer-centeredness on North American systems design. Various other projects could serve this purpose as well-see, e.g., Holtzblatt & Beyer (1995)-but EC does a good job of bringing together many of the themes and issues with which this paper is concerned. EC was more daring than most design methodologies in that it recognized that the successful integration of the user/customer in that process will necessarily transform the work practices of engineers. It approached the methodological problems of design with essentially the same general objectives sought by reengineering-reducing development time, improving product quality and customer satisfaction-but with a slightly different set of values. Besides improving the product through the integration of users in design, EC sought to improve the practices and skills of engineers by engaging them in novel situations where traditional practices and routines could not be readily applied:

Working directly with users and supporting their day-to-day work requires engineers to be committed to helping users in a personal way. Hierarchical dependency relationships between engineers and managers do not work in a codevelopment effort that bridges two different enterprises. In addition to independence, team members are encouraged to develop a diverse set of technical, interpersonal, and often interdisciplinary skills. Finally, individuals develop their own direct and informal contacts within their own and the users' organizations. (Anderson & Crocca, 1993, p. 49)

This "radicalization" of engineers' work was not, actually, an objective at the beginning of the experimental project, and was in fact one of the causes of the project's failure. Thus it was neither arbitrary nor purely philanthropic, but reflected researcher's committed objective to experiment with and develop improved engineering methodologies-they were willing to try it in order to "see what would happen." Xerox's desire to participate in the Class project with Cornell librarians was primarily motivated by a theoretical rather than a purely practical concern, but one which still sought to achieve business objectives valued by an office technologies company seeking to improve its technological advantage through organizational change. It should also be noted that the Class Project was undertaken by Xerox researchers outside of PARC, though PARC researchers did consult on the project.

The EC methodology was characterized by two essential features. The first of these was a customer-centered prototyping methodology. The system being designed was a scanning and retrieval system for some 1,000 brittle books in Cornell's libraries. The project consisted not just in designing an information archiving system, but in the development of the physical digital scanning and printing devices necessary for its effective use. Hence, the methodology placed a great deal of emphasis on obtaining customer reactions to working prototypes placed in the customers' workplace. The objective was thus to "tune" the artifact to the work environment in which it was to serve. In addition, EC was characterized by its willingness to let engineering practices develop open-endedly around the requirements of prototype development. This was reflected in the enormous amount of autonomy granted to the design team by their own corporate management. Where traditional engineering methods are organized according to a hierarchical and functional structure, the division of design labor tends to be rigidly controlled by management, who maintain authority in key decision-making situations. The semi-autonomous EC group dissolved its internal organizational hierarchy and also obtained some degree of executive autonomy from their corporate supervisors (Anderson & Crocca, 1993, p. 54). This limited autonomy resulted in a profound transformation of engineering practices in the design process, it also contributed to the political tensions which ultimately led to the project's poor reception within the company.

Traditional design methods at Xerox placed market research and development groups in charge of producing detailed system specifications, often through survey research. These specifications were in turn transmitted to design groups and subgroups via the established management hierarchy and functional sub-divisions in the engineering organization. EC by contrast sought to develop a system for a customer with needs that could not be formalized by market research methods. The design team was thus charged with the novel task of evoking and verifying the system requirements through personal interactions with users. The idea was to establish the needs of users and codevelop the prototype in real-time by letting users judge the adequacy of the prototype's features (rather than engineers as was traditional).

One consequence of this new demand on the designers in such a situation, at least in the Class project, was a radical transformation of engineering practices. Whereas design traditionally began with the division of tasks and responsibilities among designers, the EC group members "floated" among tasks and leadership roles through the course of the project based on personal initiative and experience. This had five advantages as reported by group members: 1) decisions were made by those most involved with a problem rather than by a manager with no antecedent knowledge of the matter, 2) because designers were not locked inside functional "black-boxes" they were each aware of the overall situation at hand and how their activities were situated within it, resulting in a better overall design, 3) designers were able to participate in diverse tasks which broadened their skills and knowledge while enriching their work experience, 4) all members were fully employed and did not spend large amounts of time waiting for others to complete their tasks, and 5) group members were chosen and valued for their particular and unique skills rather than their conformity to a homogenizing structure (Anderson & Crocca, 1993, p. 54).

In contrast to JAD-like methodologies which sought to alter the hierarchical and functional design process by merely adding a new functional component to that process, Xerox's EC was willing to forego the organizational structure of its design process in the hope that the flexibility provided would produce a working system where its traditional market research methods could not reach. While Xerox's explicit objectives were to design a highly-customized system for their customer, they were also interested in studying the novel methods and group dynamics that would evolve in a design team organized around a customer-centered design process and granted a great deal of autonomy. The experiment not only resulted in a product which satisfied the customer but also in a reconstitution of the engineers' work organization, which led to the promotion of some empowering humanistic values-job satisfaction, diversity of experience, skill appreciation, personal autonomy, and educational development.

It is important to note that the project's leader, Bill Anderson, began his career at Tavistock, and it is not merely coincidental that the account of engineering practice presented as an outcome of this project echoes the virtues of the "semi-autonomous workgroup." Moreover, the promotion of these humanistic values was limited to the engineers of the technological system, and did not extend to the users. While interactions with the users stimulated this organizational transformation, the users' organizational situation was not profoundly changed. The introduction of a new technological system in the librarians' workplace certainly transformed their work practices in various ways, but from a design perspective there was an established objective to maintain a value-neutral approach to these transformations: "Even though engineers are changing the customer's work practice, they need to avoid interfering with the social and political dynamics that characterize that workplace." (Anderson & Crocca, 1993, p. 55). The result of pursuing such a methodology was an engineering work practice very much like the ideal "semi-autonomous workgroup" originally sought by the socio-technical systems researchers at Tavistock, yet here it has again shifted its focus towarc the design process and away from the consequences of technological change on the organization of workers' practices. And, like the Scandinavian PD researchers, EC recognized the difficulties of communicating design concepts between engineers and librarians.

4.4 Facilitating Communication Between Designers and Users

Scandinavian PD had always been concerned with how technology alters work practices. At its inception, however, it did not consider the technological design process itself to be a key point of interest. Once projects like UTOPIA had begun to problematize technological design, they saw their challenge as being to overcome the traditional roles, power relations and preconceptions of designers and workers. Very quickly they added to this the problem of "communication." What researchers found most difficult, once socio-political barriers had been bridged, was that designers and workers tended to talk past one another. Researchers also encountered, though were not surprised by, problems in uncovering knowledge of tacit skills or embodied routines and realized a necessity for respecting the fact that workplaces have a rich local vocabulary that takes time to master and is not always easily translated to individuals outside the workplace.⁽¹⁸⁾ It was expected that bringing the workers into the design process would also bring their tacit knowledge into the technological product-this, however, would ultimately be realized by bringing the designers into the workplace.

What researchers found to be most difficult was effectively communicating a technical system's design to workers who lacked technical knowledge. While users were not participants in the actual technical execution of the design, they found it difficult to understand the various system designs which the engineers proposed. The initial response to this in the UTOPIA project was to develop exemplary "screen shots" of what the potential design would look like. This helped, but users found it impossible to judge whether such a design would satisfy their needs since they could not conceive of how it would actually operate. In response to this problem, the researchers developed many innovative means for communicating the practical functionality of various designs and design elements to potential users. The PD literature has since produced numerous articles on "prototyping," "visualization," "mock-ups," "storyboarding," "metaphorical design," and "future workshops" which all have the expressed purpose of offering suggestions of how to develop and use videos, transparencies, functional prototypes, and even cardboard boxes and plywood to give users a sense of how a proposed system will work.⁽¹⁹⁾

(Kensing & Munk-Madsen, 1993, p. 80, Table 1)

What is interesting about this model is that without ever mentioning "anthropology" or "culture," they have effectively presented an ethnographic model of inter-cultural discourse. Moreover, they do not see the problem of communication as getting information from one domain to another, but as one of bridge-building-synthesizing a new field of knowledge through the confluence of two different fields of discourse and sets of practices. The objective is to align the concepts and representations of both workers and engineers around a common discourse and set of practices through which the desired technological artifact can evolve and find use.