Source
Automatically imported from: https://pages.gseis.ucla.edu/faculty/agre/oksnoen.html
Content
Design for Democracy
#### Philip E. Agre Department of Information Studies University of California, Los Angeles Los Angeles, California 90095-1520 USA pagre@ucla.edu http://polaris.gseis.ucla.edu/pagre/ Paper prepared for the 1994 Oksnoen Symposium. This is a draft. Please do not cite it or quote from it. Version of 28 February 1994. 6100 words.
1 Introduction
The design of computer systems has not historically been organized in a democratic way. Designers and users have had little interaction, and users have had little control over the resulting systems, except perhaps through the indirect routes available to them through resistance in the workplace and the refusal to purchase relatively unuseable systems for their own use. Yet over the last ten or twenty years, a growing movement, originating in Scandinavia but now increasingly influential in other industrialized countries, is attempting to reform the design of computer systems in a more democratic direction (Bjerknes, Ehn, and Kyng 1987, Schuler and Namioka 1993). This movement, sometimes known as participatory design, invites the participation of, and in many cases gives formal control over the design process to, the people whose work-lives the system affects. Like all democratic movements, the participatory design movement faces an uphill struggle against both the institutional structurings of technical practice and a deeply ingrained belief-system that tends to reify and fetishize technology -- and thereby to suppress sophisticated awareness and analysis of the forms of social practice that go on around it.
This paper is a theoretical contribution to the project of democratic design. It proposes several categories for analyzing the social dimensions of computer technologies and design practices. Central among these categories is the notion of a design formation, which pertains to the systematic ways in which computers are institutionally and imaginatively knitted into the structures of human activity around them. In particular, I propose to contrast, in a broad and general way, two design formations: rational design and democratic design. Rational and democratic design are not wholly stable and unified phenomena; they each have their respective histories and they each encompass a moderate diversity of sometimes conflicting subtraditions. They nonetheless represent two distinctive approaches to conceptualizing and organizing the social process of computer system design. Rational design is currently predominant through most of the technical world, still deeply and subtly presupposed by the standard language and methods of design. Democratic design, on the other hand, is still very much an emerging tradition, and one of my aims is to assist its would-be practitioners in surfacing and working through the rationalist inheritance in their unfolding practice.
Section 2 discusses some basic analytical issues pertaining to the relation between technology and society; it then introduces the general idea of a design formation. A design formation presupposes, and helps to reproduce, a specific vision of human relationships through the social organization of design and of computer-mediated activity.
Section 3 describes rational design. Rational design takes up two complex and inter-related stances toward language. The first concerns the language of formalization, through which worldly things are assimilated to mathematics for purposes such as computer programming. The second concerns the linguistic metaphors through which computer design methods formalize human activities. These uses of language articulate with particular unfortunate types of imaginative and institutional relationships between the designers and users of computer systems.
Section 4 describes democratic design. Democratic design begins with a distinctive approach to language and its role in design. Whereas rational design employs formalization to assimilate human activities to mathematics, democratic design makes formalisms into objects of human reflection and action. And whereas rational design endows computers with useful functionality through systematic capture of the activities they support, democratic design fashions media within which communities of practice can organize their activities for themselves.
2 Design formations
The central analytical problem in social studies of technology concerns the relationship between two categories frequently treated as autonomous, "technology" and "society". Determinist positions lie at two extremes: technological determinism holds that technological development proceeds according to its own internal logic and then proceeds to have "effects" in society, and social determinism holds that technologies come into being to serve functions that arise in historically specific settings. Both positions are untenable. Machines do not determine the ways in which they will be used, but neither are machines wholly neutral in social terms. Rather, forms of technology coevolve with forms of human relationship. The dialectical motion of this coevolution brings people and machines into ever more intimate connection, but this connection is never absolute. Every form of machinery is compatible with a range of social relationships and forms of activity.
This much is relatively uncontroversial within social studies of technology, and a variety of authors have developed theoretical frameworks to explicate more specifically the interrelationships of social forms and technical forms. One of these is Latour (1987), whose actor network theory draws attention to the complex networks of social actors (a category which, for Latour, includes both people and machines) within which both human activities and the operation of machinery takes place. A computer, for example, is always interpolated within a considerable range of human activities, from design to sales to maintenance to repair to managerial planning, and the functionalities of computers are frequently modeled upon, and interlock in complex ways with, the tasks undertaken by people. Despite all of the significant phenomena to which it draws attention, however, this analysis does little justice to the specific ways in which particular forms of technology, and particular forms of technical design, actually work.
An alternative tack, pursued here, takes as its unit of analysis the *design formation*. A design formation is a historically specific discursive structuring of the imaginative and practical relationships between designers and the people whose lives are to be affected by the artifacts being designed. The term "design formation" is modeled on Foucault's notion of a "discursive formation", the difference being that a design formation is rooted in the practices around a particular category of machinery. The unit of analysis, then, is not a category of physical artifacts, but rather the system of discursive forms that surrounds, and in large measure serves to constitute, a given category of artifacts. These discursive forms might include metaphors, distinctions, genres and registers of language, styles and formats of writing, forms and structures of conversational interaction, and so forth.
A brief example may help to bring these abstractions down to earth. Research on "planning" in artificial intelligence employs a notion of "hierarchy" that has a rich and interesting lineage. The original hierarchies, of course, were administrative. Panofsky (1957) has argued that hierarchy was a pervasive metaphor in the cultures of Europe during the Middle Ages, and Weber (1947 [1922]) made administrative rationality into a central topic of sociological theory. Simon (1947) proposed a theory of administrative decision-making based upon a complex complementarity between individual cognition and bureaucratic information flows. Hierarchies, Simon argued, systematically compensate for the limitations of individual cognition, with the result that individuals and organizations fit together in complex ways. Far from being independent and self-defining agents, individual human beings exhibit jagged contours whose concavities are actively met by the procedures of the organization. In particular, Simon argued that individuals receive their goals and values from their organizations, provided that these dictated actions that lay within their particular "zones of indifference", a concept borrowed from Barnard (1962 [1938]).
Simon, of course, went on from his organizational studies to found artificial intelligence, and the first computer models that he and his colleagues built had a tremendous influence on virtually all subsequent work in the field -- and most particularly upon the field's conception of the individual. These models did not follow Administrative Behavior in great detail, for the simple reason that the programming methods then available, including the highly innovative methods of symbolic programming that Simon and his colleagues invented to construct these models, were inevitably not expressive enough to support the full complexity of Simon's theory. These models, therefore, inherited the ideas of Administrative Behavior through a kind of filter. Lost in that filter were virtually all of the jagged edges along which Simon imagined individuals fitting into organizations; what remained were the individualistic conception of formal problem-solving then current in psychology and the idea that individuals inherit their goals from their superiors in hierarchies. The notion that a program is assigned a goal from outside was soon presupposed by the vast majority of the formalisms taught in classes on AI and extended through further research in the field.
AI's conceptions of hierarchy had other sources, of course, through both Simon and others. The grammatical analysis of sentences by linguists such as Chomsky had a tremendous influence, and hierarchy as a principle of rational organization is central to AI works as otherwise heterodox as Minsky's Society of Mind (1986). The conception of hierarchy that has (until lately anyway) grown such deep roots in the technical practice of artificial intelligence is a simple example of a design formation. Although the ideological connection between hierarchical schemes of rational control and hierarchical metaphors for human cognitive organization in AI are clear enough to radical critics (Berman 1989), they are nonetheless wholly obscure to the vast majority of AI practitioners themselves.
The point, then, is not that design formations embody conscious conspiracies. Rather, rhetorical and genealogical research can uncover structures in technical discourses -- each of which, once elucidated, describes a particular mode of dialectical articulation through which the technical design practice conjoins forms of machinery to forms of human social life. Design formations, in other words, are simultaneously forms of technical imagination and forms of social imagination. They provide the discursive raw material through which lines of interconnection can be, and are, drawn between the machines and the social arrangements around them. These lines of interconnection can be numerous and diverse, and they deserve fresh inquiry in each case. The category of "design formation" can orient and focus that inquiry.
3 Rational design
Rational design is the design formation that is currently found in the vast majority of computer design settings worldwide. The category of rational design identifies some deep and usually unarticulated features of computer design work; as such, it casts a wide net and encompasses a reasonable amount of internal diversity. It will not be my purpose to catalog this internal diversity, nor to identify varieties of rational design that might be considered better or worse by any particular ethical or political criterion. The underlying discursive structures that constitute rational design are extraordinarily stable and receive little discussion, critical or otherwise, within mainstream computer science. As a result, they have rarely been made objects of analysis, experiment, or reform.
Rational design can be characterized in terms of two distinctive uses it makes of language, formalization and capture. After discussing these notions separately, I will draw out certain common themes and attempt a summing-up.
Formalization is the discursive process through which practitioners of a technical discipline such as economics or computer programming assimilate non-mathematical things to mathematics. I will give a systematic account of this process in another paper (Agre forthcoming); the account here will be brief and doubtless unsatisfying, given how little critical attention has been given to this highly consequential phenomenon.
Formalization begins in the earliest stages of mathematical education with word problems such as "John has four apples; Mary has three apples; how many apples do they have altogether?". Such problems are not simply sugar-coating for arithmetic exercises such as 4+3=7. To the contrary, they are the earliest stages in the cultivation of a particular kind of mediated relationship between mathematical formalisms and the worldly states of affairs which they are held to represent. A child who is faced with the brief tale about John and Mary will not be encouraged in exhibiting curiosity about John and Mary and their apples, much less in reserving judgement until John and Mary have been able to tell their own side of the story.
The point is that mathematical knowledge has a specific institutional form. Math classes are not organized into classes on fruit, classes on cars, and classes on factories. Instead, they are organized by particular mathematical schemata; a student who is faced with a word problem can be assured that it is an addition problem or a multiplication problem or a rate-of-work problem, simply because that is the lesson plan for the day. In this context, a problem such as "one drain can empty the pool in 12 hours; another drain can empty the pool in 8 hours; how long will it take them to drain the pool together?" can be understood as perfectly homologous to a problem such as "Jim can paint the house in 12 hours; his Dad can paint the house in 8 hours; how long will it take them to paint the house together?", even though the latter problem does not provide nearly enough information to derive an answer if Jim and his Dad are imagined to be actual people with an actual house and a potentially complex personal relationship.
This kind of math teaching organizes a particular experiential relationship between mathematics and things, which Heidegger (1977 [1953]) referred to as "enframing". Each type of word problem corresponds to a conceptual frame such as "rate-of-work problems"; to "apply" mathematics to circumstances in the world, one looks to fit those circumstances to one's repertoire of available frames. Many technical fields, such as operations research and game theory, consist of the application of a family of such frames. Practitioners of these fields are, as Heidegger would say, challenged forth to construe material circumstances in the world as fitting those fields' constitutive frames, while drawing attention away from any aspects of the world that exceed the enframings that are constructed through this process. Particular enframings may be tried out, found wanting, replaced, refined, extended, or compared with others, without the basic practice of enframing being thematized or questioned.
The fictional nature of word problems supports this kind of structured and distanced relationship to the sites of technical practice. A word problem is a kind of code, whose decryption reveals the mathematics hidden underneath and facilitates the goal of "setting up" and "working out" the necessary arithmetic. Various small refinements of this procedure are certainly found, especially in the the most progressive mathematics curricula of the last ten years, but it is still the rare mathematical curriculum that leaves students equipped to employ mathematical tools in a discerning way, always conscious of the provisional nature of all formalizations and the latent claims of the bits that will inevitably get left out. Lacking this kind of "feeling for the organism" (Keller 1983), it is a straightforward transition from the fictional distantiation between word problems and their objects to the institutional distantiation between technical professionals and their sites of practice. This distantiation, of course, is not a total disconnection; it is, rather, a highly structured epistemological relationship that supports and assists in reproducing the project of enframing, which is itself partially constitutive of the practice and which helps underwrite its claims to expertise.
Computer systems design is itself a technical discipline in this sense. Many of its frames are drawn from existing disciplines, and it contributes its own through the catalogs of fundamental algorithms with which every programmer becomes acquainted. Instruction in programming, like instruction in basic applied mathematics, employs fictional scenarios organized not in thematic terms but in terms of the frames of algorithms and the conventional structures of programs and data. Together the frames of computer system design provide an extraordinarily rich and versatile medium of formalization. To design a computer system for a particular application domain, one begins by enframing the central concepts of that domain in mathematical terms that are commensurable with some of the conventional frames of computing. This work may be performed by a systems analyst through some kind of systematic investigation of information flows and the like, or it may be a more informal process of gathering enough of a conception of the domain's "requirements" to allow a practicable, or at least a plausible, enframing to take form.
In another work (Agre 1995), I have built upon numerous empirical studies of programming to describe certain aspects of programmers' imaginative and institutional relationships to the people who will use the systems they produce. This "technical conception" imagines the user simultaneously to share the perspective of the programmer (so that numerous questions of intelligibility, knowledge, and meaning cannot be faced squarely) and to be a component of the designed system (one locus of information-storage and information-flow among many). This ambiguous and divided conception of the human subject, and the particular type of human relationship which this conception presupposes and helps reproduce, form one component of rational design. As a design formation, it is not necessarily part of anybody's conscious intentions; rather, it is carried in particular kinds of language and methods and institutional organization.
A second aspect of rational design is the phenomenon of capture, which I have described in detail elsewhere (Agre 1994). The word "capture" is a term of art among computer people; when a computer receives certain information as input, it is said to have captured that information. The point is that, whereas "input" is a wholly formal notion (for example, an eight-bit integer quantity), "captured information" has some definite semantic relationship to the world outside (for example, a diagnostic code). In its most general signification, capture refers to the process by which a computer maintains an up-to-date representation of some complex state of affairs with which it is interacting.
The most important situation for present purposes arises when a computer is designed to support some kind of human activity. Within the worldview of rational design, a computer supports an activity by maintaining a representation of that activity, so that the activity throws a kind of "shadow" upon the internal workings of the machine. After all, goes this way of thinking, a computer can only compute with what it can represent, and it can only represent what it can capture. Human activities are understood on a linguistic metaphor; they are conceived as assembled from a vocabulary of standardized parts (for example, transaction types or speech acts) through a set of rules specifying the sequences into which they might be arranged (for example, accounting procedures or conversational conventions). This "grammar of action" may be drawn on the users' own language for discussing their activities, though it need not. Also, the grammar need not express the full complexity of the activity, so long as it provides a formal framework into which particular stretches of activity can be fitted by their participants.
Grammars of action are normally accompanied by an ideology which regards them as transparent representations of the activities they represent. The reality, though, is almost invariably more complex. The installation of new computers inevitably involves changes in the activities with which the computers interact. Some of these changes will be consciously planned concomitants of automation, and others will take place through improvised accommodations to the newly computerized regime. These accommodations might include the practicalities of data entry, the interpretive work necessary to categorize activities within a particular formal grammar, selective use of the system, and reorganization of the activities with an eye to the organizational consequences of the captured representations. Capture, then, requires that human activities be reorganized, through both technical and social means, to make them more easily trackable by computers.
As elements of a common design formation, formalization and capture have several common features. Each one involves a special stance toward language that helps structure the relationships between computers and the social worlds around them. In particular, each one uses language to assimilate elements of human experience to mathematics, either the formalisms employed by designers or the formalisms constituted within the operating computer itself. Furthermore, each one involves a type of ideology and practical work through which this process of assimilation is hidden from view, as if the domain of practice in question were already a mathematical structure. Of course, some domains of practice already have substantial elements of mathematical structure, but many do not, and even the most mathematically intensive domain of human activity has its interpretations and judgements through which the mathematical structures are understood as referring to states of affairs beyond themselves.
Formalization and capture are not just ideas about the relationships among computers, language, and human activity; they are forms of social practice through which definite relationships -- both material and ideological -- among computers, language, and human activity come into being, and moreover they are systematic properties of these newly arisen relationships themselves. This is precisely what allows us to interpret formalization and capture as elements of a design formation -- rational design. The relationships imagined and instituted through formalization and capture do not wholly determine the contents of particular computer-mediated activities, nor do they wholly determine the sociotechnical dynamics of the implementation and use of any given system. An understanding of rational design in any given site depends upon an elucidation of most everything else that is going on in that setting, from industry structure to managerial ideology to the prevailing traditions and practices of resistance.
Rather, then, than well on these underdetermined aspect of the practices surrounding rational design, I wish to conclude this section with a comment on the ideology, or rather, the form of social imagination, that is implicit in formalization and capture. Each of them specifies and enacts a kind of trajectory, a representational translation and practical transformation, from encountered states of affairs (to be treated as puzzled and as sites of intervention) to mathematical abstractions (in the minds of designers and the functioning of computers). Although individual designers are certainly capable of ordinary friendliness and concern, the social vision described by this trajectory has no place in it for engagement with people. That is, rational design understands "users" as part of a field of intervention and not as subjects and agents of the design process. Users may on occasion be asked to "participate", but this participation is in no way constitutive of the design process, which is defined in terms of the transformation of a given site of intervention from a problematic and disorganized condition to a systematic corrspondence with formally defined mathematical structures and processes.
In an important way, rational design regards these mathematical structures and processes as a higher reality than the lived worlds of the users, which cannot be understood within rational design except in terms of their correspondence to, or divergence from, the formalisms that are invented in the design process and realized in the functioning machine. Noble [where] identifies this attitude as a philosophico-religious system that has underlain a great deal of the historical development of technology; he describes it with exquisite precision as "masculine transcendentalism", the project of raising up corporeal reality to a more perfect level of crystalline order. This need not be anybody's consciously formulated intention, of course. But its tacit presence in the technological imagination -- and most particularly around applied computing -- helps explain, for example, the peculiar social movement around technologies such as virtual reality, which can be understood as a masculine transcendentalist image of the millennium of disembodied abstraction.
The social imagination of masculine transcendentalism, of course, does not fully determine the uses to which computers are put, nor the social relations that surround them. It does, however, obscure these things, and in doing so it frustrates the project of democratic design. I hope that a clearer theoretical analysis of the problem may help clear the ground for a fresh approach, and it is to the possibilities for such an approach to which I now turn.
4 Democratic design
What would a democratic design formation be like? The conceptual inertia of rational design is so deep and obscure that it will take great care even to ask the question. It is best to start from first principles. Democratic design, for one thing, does not refer to a distinctive class of machines. The design formation, as a unit of analysis, identifies a discursive structuring of the imaginative and institutional interconnections between machines and their social environments. In particular, a design formation goes a long way toward defining what a given category of machines is, so far as the machine's connections to the world itself is concerned. Forms of machinery coevolve with forms of social relationship, so an existing computer, considered narrowly as a physical artifact, might conceivably fit into a democratic form of social practice, just as an animal might conceivably find a niche in an alien ecosystem. As previous theorists and practitioners of democratic technology have understood, the necessary unit of analysis is sociotechnical and not simply technical (Trist 1981). The next step is to take up this line of reasoning with greater attention to the specific qualities of computers as a category of useful technical artifacts.
And so we might ask, would computer technology that coevolved with democratic forms of social practice look considerably different, when compared in the narrowest technical terms, than the computers with which we are acquainted today? Must we give up the transistor to practice democracy? The local area network? While it is intriguing to imagine a radical critique of existing computing technology "all the way down to the bits", it not very useful to speculate at such an abstract level. Nor is it particularly useful to ask, in an abstract way, how this or that type of machinery can "support" democratic social practices as opposed to other types. Having identified certain aspects of rational design which articulate with an undemocratic social vision, it will be important to up the matter on this more concrete level through a reconsideration of the relationships between language, machinery, and social imagination in computer system design. First, though, it is necessary to back up and recall a few things about democracy.
An American scholar, recently returned from three years of college teaching in Denmark, told me a story that illuminated for me the decay of democratic practice in my own country. One day, he said, some students came to his office with complaints about one of his classes. They told him, "we don't like this", and "we don't like that", and "we want this", and "we think you should do that", and so on, and as they did so it gradually became apparent to him that "we" actually meant the whole class. The class, it would seem, had met outside of the scheduled time, discussed issues of concern to them, and sent representatives off to deliver their concerns to Randy, confident that they would be able to conduct a successful negotiation. To my mind this arrangement is vastly preferable to the "evaluation" survey administered to my students at the end of each term, which presupposes and exacerbates a kind of passive-aggressive attitude on the part of students who have never learned how to organize themselves to change their environments. My friend's experience reminded me that democracy is not simply, or even mostly, a set of formal mechanisms and institutions. Much more fundamentally, it is a set of skills and habits that are best understood as cultural -- a way of live which one acquires with all of the ease, and all of the difficulty, of the partially articulated ways of being of a culture.
The actual content of the "skills and habits" of democracy will, of course, vary by culture and by the type of activity. "Manuals" of democratic social practice have been prepared by people involved in social movements, unions, cooperative businesses, and experiments in collective living, and by authors sympathetic to these initiatives [refs]. The development of democratic design should start with a full awareness of these practices and not, for example, with attempts to evaluate the political qualities of particular technical proposals. This process may be difficult, and indeed, democratic design may seem a more burdensome activity than rational design, for the simple reason that democratic social practice cannot call upon the relations of power that are necessary to impose forms of activity and technology that have been imagined in rational abstraction beforehand.
Rather than attempt a systematic exposition of democratic social practice and its consequences for design, I will concentrate on the particular issue that was central to a critique of rational design -- the form of social imagination implicit in a given way of using language. For rational design, the purpose of language is to mediate an assimilation of practical reality to the mathematics that underwrites a particular vision of social order. For democratic design, the purpose of language is considerably different: to assist people in organizing their own lives by building and maintaining relationships across the boundaries of distinct cultural and professional worldviews. As Robinson and Bannon (1991) have pointed out, these distinct worldviews are present in a computer's operating environment whether this heterogeneity has been recognized by the design process or not. They describe some of the reinterpretations that computational formalizations routinely undergo, and they ask how a design practice might facilitate the process, as opposed to trying to suppress it.
Robinson (1991) suggests that the answer lies in "double level languages". A computer system and the social arrangements around it are said to employ double level languages when distinctions are clearly maintained between the mathematical formalisms employed by the computer and the cultural languages employed by its users. This is not the norm in rational design, which seeks to suture a computer into the semantic order that surrounds it through the establishment and maintenance of a system of ambiguities between the users' cultural languages and the machine's constitutive formalisms. A display on a computer screen, for example, might use terms like "employee" and "price", on the assumption that certain formalized versions of these terms correspond to the way these terms are employed by the system's users. Robinson regards this assumption as dangerous and recommends, to the contrary, that the computer's formalisms be presented as formalisms, and in particular as tools that can be used -- that is, interpreted by the users -- in an unbounded range of ways. Users may wish to establish a relatively stable set of correspondences between formal items and cultural terms, but these correspondences must be understood as contingent and reversible. It follows that users must be supported in interpreting troubles with the system as evidence for an unsatisfactory assignment of cultural meanings to elements of the system's formalism.
Robinson's suggestion is extraordinarily important, and its consequences and practicalities will require substantial effort to work out fully. For present purposes, I am simply concerned to portray it as the core of an emerging design formation -- and therefore to point at the conception of human relationships that it offers as an alternative to the conception that is implicit in rational design. Whereas rational design melts down human subjectivity, agency, and diversity into the false objectivity of a mathematical order, democratic design gives a central place to the heterogeneity of socially organized standpoints and thus to the practical work of engagement among those standpoints. This work of engagement has numerous dimensions, from the narrowly linguistic negotiation of working glosses and "translations" of one-another's vocabularies to the overtly political negotiation of interests and agendas.
This activities are obviously conditioned by a wide variety of historical and situational factors. Among these factors is the participants' consciousness of such social processes, and in particular their consciousness of the role of language in the social arrangements surrounding computers. Rational design and its practices of formalization and capture organize a definite set of false beliefs and mystified experiences of technologically mediated activities, and the reader can easily be forgiven for being unaware of any bonds of domination radiating from his or her Macintosh. Insight on these matters may be more readily available to those whose activities are, or soon will be, captured in great detail and in real time by computers that they did not choose or design. In developing this insight, it presumably does not suffice to be exposed to a theory of formalization and capture.
As a practical matter, consciousness of design will manifest itself within the activities of design and use themselves. Although rational design is most straightforwardly consistent with a strict division of labor between designers and users, a variety of movements employing the language of "participation" have explored alternative ways of involving users in the design process. Uncritical attitudes toward this concept of participation have delayed the important work of mapping out these various projects and their properties. Toward one extreme are located the formative experiments of the participatory design movement, in which designers presented users with mock-ups of potential systems and users had a degree of genuine control over the process, at least to the extent that they could withdraw their participation in it (e.g., Bodker et al 1987). Toward another extreme are attempts, within much more conventional organizational frameworks, to solicit users' understandings of their activities as one of the inputs to a basically rational design process. These latter activities are capable of devolving into the kind of appopriation of workers' production knowledge that has been decried by a considerable literature in the union movement, typified by Braverman (1974). The point is that participation must always be understood as a contested terrain (Banks and Metzgar 1989). The purpose of democratic design is not to eliminate the conflicts of interest and worldview that are played out through the participation process, but simply to acknowledge these conflicts and allow them to be confronted and worked through in a democratic fashion.
The role of consciousness in design participation has been made particularly evident by the emerging practice of organizational communication. Resistance to new work practices has been a commonplace occurrence throughout the history of technology. To address this resistance, management theorists and practitioners have developed organizational communication, a kind of public relations aimed at an organization's own employees. The basic procedure is to map out the employees' cognitive processes in regard to work activities and the social arrangements that surround them, to formulate the "perceptions" to which these cognitive processes give rise, and to design new technologies and the communications that surround them (announcements, instructions, training, and so forth) with a view to circumventing or dissolving those perceptions. In a more extensive analysis elsewhere (Agre 1994), I have referred to this process as "design for perception", in which users' cognitive processes are understood as being nearly continuous with the operation of the machinery itself. On this view, resistance to new work arrangements is a malfunction of the sociotechnical system, just as a "bug" causes a malfunction of the techncial system.
The practice of design for perception is not yet widespread, at least in its most sophisticated forms, but it is valuable to contrast it to the form that might be taken by a more genuinely democratic practice of technology. Would-be democratic designers cannot simply regard themselves as designers of both technologies and forms of consciousness. This unilateral conception of design must be replaced by something much more difficult: a conception of design for which the very conception of a computer is, as a matter of course, open to negotiation as part of the engagement of worldviews that takes place throughout the design process. This kind of openness is entirely foreign to rational design, which regards its sites of practice as chaotic terrains requiring reconstruction through systematic assimilation to a rational order. Placing the entire substance of computer science on the negotiating table on each occasion of design certainly seems a radical proposal within such a frame of reference; the point is that this is the only proposal that is consistent with an end to the ideologies and illusions of rational design.
All of the foregoing should be regarded as preliminaries to the construction of a democratic design formation. The actual substance of this construction will take place through the innovations that take place within particular design projects, as the conceptual inertia of formalization and capture begins to wear off. It will be necessary, among other things, to formulate a conception of a computer's functionality that does not depend upon capture -- that is, upon the maintenance of a systematic correspondence between internal memory-states and the human activities outside. An alternative begins, once again, with the democratic goal of facilitating the interpenetration of worldviews among the parties to a given activity. This suggests employing computers as media that support the communication, cultural forms, and collective memory of a community. The sociotechnical conditions for supporting the life of a community in this way are not well understood. Numerous authors (e.g., Coate 1997, Rheingold 1993, Smith and Kollock 1999), for example, are exploring the properties of the "virtual communities" on the Internet. My own sense, though, is that electronic mail as currently understood is much better at communicating information and arguments within a worldview than at facilitating a meeting of minds between worldviews. A better approach, perhaps, is to look at computer-mediated communications as part of a larger communicative ecology within the collective life of a community. After all, the purpose of democracy is not the rational design of machinery but the conscious choice of a way of life.
Acknowledgements
Some parts of this paper were stimulated by my reading of an unpublished manuscript by Mike Robinson. It has benefitted from comments by Philip Fleischer.
Bibliography
Philip E. Agre, Surveillance and capture: Two models of privacy, The Information Society 10(2), 1994, pages 101-127.
Philip E. Agre, Conceptions of the user in computer system design, in Peter Thomas, ed, Social and Interactional Dimensions of Human-Computer Interfaces, 1995.
Philip E. Agre, Rhetoric and math in technical language, forthcoming.
Andy Banks and Jack Metzgar, Participating in management: Union organizing on a new terrain, Labor Research Review 14, 1989, pages 1-55.
Chester I. Barnard, The Functions of the Executive, Cambridge: Harvard University Press, 1962. Originally published in 1938.
Bruce Berman, The computer metaphor: Bureaucratizing the mind, Science as Culture 7, 1989, pages 7-42.
Gro Bjerknes, Pelle Ehn, and Morten Kyng, eds, Computers and Democracy: A Scandinavian Challenge, Aldershot, UK: Avebury, 1987.
Susanne Bodker, Pelle Ehn, John Kammersgaard, Morten Kyng, and Yngve Sundblad, A UTOPIAN experience: On design of powerful computer-based tools for skilled graphic workers, in Gro Bjerknes, Pelle Ehn, and Morten Kyng, eds, Computers and Democracy: A Scandinavian Challenge, Avebury, 1987.
Harry Braverman, Labor and Monopoly Capital: The Degradation of Work in the Twentieth Century, New York: Monthly Review Press, 1974.
John Coate, Cyberspace Innkeeping: Building community online, in Philip E. Agre and Douglas Schuler, eds, Reinventing Technology, Rediscovering Community: Critical Explorations of Computing as a Social Practice, Greenwich, CT: Ablex, 1997.
Martin Heidegger, The question concerning technology, in Basic Writings, translated by William Lovett, New York: Harper and Row, 1977. Originally published in German in 1953.
Evelyn Fox Keller, A Feeling for the Organism: The Life and Work of Barbara McClintock, San Francisco: Freeman, 1983.
Bruno Latour, Science in Action: How to Follow Scientists and Engineers through Society, Cambridge: Harvard University Press, 1987.
Marvin Minsky, The Society of Mind, New York: Simon and Schuster, 1986.
David Noble, [masculine transcendentalism]
Erwin Panofsky, Gothic Architecture and Scholasticism, New York: Meridian Books, 1957.
Howard Rheingold, The Virtual Community: Homesteading on the Electronic Frontier, Reading, MA: Addison-Wesley, 1993.
Mike Robinson, Double-level languages and co-operative working, AI and Society 5(1), 1991, pages 34-60.
Mike Robinson and Liam Bannon, Questioning representations, in Liam Bannon, Mike Robinson, and Kjeld Schmidt, eds, Proceedings of the Second European Conference on Computer-Supported Cooperative Work, Amsterdam: Kluwer, 1991.
Douglas Schuler and Aki Namioka, eds, Participatory Design: Principles and Practices, Hillsdale, NJ: Erlbaum, 1993.
Herbert Simon, Administrative Behavior: A Study of Decision-Making Processes in Administrative Organization, New York: Macmillan, 1947.
Marc A. Smith and Peter Kollock, eds, Communities in Cyberspace, Routledge, 1999.
Eric Trist, The Evolution of Socio-Technical Systems: A Conceptual Framework and an Action Research Program, Toronto: Ontario Quality of Working Life Centre, 1981.
Max Weber, The Theory of Social and Economic Organization, translated by A. M. Henderson and Talcott Parsons, New York: Oxford University Press, 1947. Originally published in German in 1922.