Designing the New Information Services
by Philip E. Agre
Everybody knows that information technology will change the world, but nobody knows how. Basic changes in technology have created a vast new design space, including both new ideas made conceivable by the technology and old ideas made practicable by the technology's widespread adoption. Technological change is often an occasion for institutional change, and interest groups are announcing that exponential improvements in information technology make their own institutional change agendas both possible and inevitable. In the case of higher education, for example, database vendors such as Oracle imagine the university being rebuilt online around large multimedia courseware servers. Others imagine a market in finely divided educational services, such that students purchase each course or each lesson separately. Such agendas can become self-fulfilling to a certain extent if they become inscribed in the basic categories of software architecture and public policy. Before this happens, before the possibilities of technology are foreclosed by the limits of imagination, we need to discuss the full range of technological and institutional options that are actually available at this seemingly pivotal time.
How do technology and institutions interact? One story that used to dominate much of the world, but that has since fallen out of fashion, identifies information technology with rationalization. Indeed, in its original definition,1 the phrase information technology referred specifically to the use of computers to replace human judgment by rational decision-making in organizations. On one level, this story extended the practices of industrial automation from the factory floor to the manager's office; on another level, it was part of a centralized command-and-control ideology that was just as deeply identified with capitalism as it has come to be with socialism.
Strange as it would have seemed in the 1960s, information technology has lost its close cultural association with rationality. In part, this change has been driven by actual organizational experience. Businesses that have tried to organize their computer systems around a single consistent information architecture have largely failed. In business schools, this ill-conceived attempt to create a universal language has been replaced by a healthy respect for the difficulty of reconciling the vocabularies and practices of different business disciplines such as marketing, manufacturing, and finance.2 This kind of epistemological pluralism is found on both the left and the right. Feminists such as Sandra Harding emphasize the ways in which knowledge is embedded in cultural worldviews and ways of life, and classical libertarian economists such as Friedrich Hayek argue that a centrally organized social order could never take into account the detailed knowledge that individuals have of their local environments.3
As a result, a new story about information technology and institutional change has taken hold, one that embraces the opposite extreme of decentralized self-organization. Central to this story is the concept of disintermediation, which courtesy of the miracle of Latin morphology is readily understood as the obsolescence of all institutions that play a mediating role between buyers and sellers, borrowers and lenders, students and knowledge, citizens and laws. The new institutional order is imagined to be perfectly fluid, so that anybody can be connected to anybody or anything at any time and to anybody else or anything else at any other time. It is order without centralized direction and with a perfect devolution of all decision-making power from hierarchies to peripheries.
The desirability of this story rests largely on its antipathy to institutions, which it would dissolve. Institutions are envisaged as encumbering and static, selfishly imposing an artificial and outdated order on lives and relationships that could otherwise be freely chosen. These mediating institutions had existed to create a path from point A to point B, goes the story, but now that information technology provides such paths by its very workings, the institutions exist only to resist change and postpone their own fall. This story depicts institutions in a certain way: constraining and not enabling, conduits and not repositories of memory in their own right. It is surely a story with significant elements of truth. And yet our responsibility in imagining an evolved institutional order in a new technological world is to ensure that other elements of truth are heard as well and are incorporated into the common sense of our times.
To evaluate these stories and others, we must get beyond some simple oppositions. For example, it is useless to favor or oppose technology. The point is not that technology is as undeniable as gravity but rather that technology is plastic: we can be certain that the most basic quantitative aspects of information and communications technologies will continue to improve for the foreseeable future, but nothing follows from this about the qualitative organization of the technology that we will actually choose. Information technology is singularly open to social shaping and creative appropriation, and opposition to a particular story about information technology does not imply opposition to technology as such.
We must also get beyond particular disciplinary frames. Notwithstanding the emergence of new cultural stories, the day-to-day language and practices of computer science still embody a great deal of the control orientation that they derived from the institutional environments of their birth. For example, the main tradition of computer systems design is inherently hostile to privacy, in its assumption that everyone and everything in sight needs to be identified and in its inability to comprehend, much less to adopt, the privacy-protection technologies that eliminate conventional identifiers or render them difficult to reconstruct.4 Clearly the disciplinary frame of computing can benefit from constructive interference from other fields. The disciplinary frame of economics, powerful though it certainly is as a way of looking at the world, interacts in complex ways with technical and political perspectives, particularly in the fraught area of the economics of information.5 The political perspectives, for their part, supply important conceptualizations of the legitimacy and illegitimacy of institutions while lacking some of the very detailed analytical powers of technology and economics. Finally, since language is at the heart of computer systems design and the human interactions that computers mediate, linguistic and discourse-analytic perspectives will be required as well, each with its own somewhat incommensurable way of framing the issues that arise. The great promise of information studies is to provide the disciplinary switchboard through which these fields and others can come together as we try to imagine and manage new information technologies as phenomena of human social life.
At the point of intersection of these various disciplinary perspectives is the need for a new integrated design practice, one that draws on a great diversity of converging disciplines to design a new generation of information services. Technologies and institutions can be, and should be, designed together, and the distinctive opportunities and requirements of each should be brought to bear on the other. No longer will it suffice to imagine new gadgets without any conception of the relationships and activities within which they will be used, and no longer can we assume that the traditional and familiar institutional forms will remain well-suited to the opportunities that new technologies provide. Something of the necessary convergence of design disciplines can be seen in novel forms of business organization that take advantage of computer networking to increase spans of control and to flatten hierarchies, in the invasion of the territory of ergonomically minded user interface design by graphic designers, many of whom now call themselves "information designers" and "interaction designers,"6 and in the emergence of a new generation of information appliances whose basic conception is driven as much by industrial design as by circuit design.7 But much more can be done to bring these various design disciplines into productive dialogue and stepwise synthesis; the design of the new information services is a good place to start. Observe that the object of design is not the gadget but the service. As such, the new design practice works outward from the experience of generations of librarians and commercial information providers who have designed information services for diverse audiences. New technology can lead to either greater fragmentation or greater integration of knowledge in this area, depending on who takes the initiative to define it.
What would such an integrated design practice be like? First, every design practice presupposes a particular relationship to power. Architecture, for example, presupposes a client who can cause buildings to be built. Computer systems design is partially the design of the human activity of using the systems, and as such it presupposes that somebody is able to persuade some people to organize their activity differently. It is little wonder, then, that computer systems design went through significant changes during the transition from the military and business markets, where such persuasive power exists in abundance, to the consumer market, where it does not. The relation of design to power is particularly crucial in the context of emerging information technologies, which are organized around compatibility standards that arise in global markets. The U.S. government, for example, is rapidly learning that it can no longer set standards, through its own unilateral action, in an area such as information security and that it must instead build multilateral political alliances while using export controls in an attempt at least to prevent vendors from settling on standards that it does not like. And what is true for the government is particularly true for lone designers and firms: new designs will be adopted only to the extent that they become standards, and they will become standards only if they are compatible, both in timing and in substance, with a remarkably complex environment of coevolving standards and the public and private stakeholders whose interests they affect. Some designers do have the good fortune to be allied with a firm, such as Microsoft, that possesses unilateral leverage over a particular portion of the standards environment. All others will need some worked-out sense of their participation in a larger process, and the practicalities and strategies of this participation will be indissociable from even the smallest details of the design work.
Some concepts will be indispensable to this new design practice. One fundamental concept is that of network effects.8 In its narrowest use, a good in a market exhibits network effects to the extent that its value to a consumer depends on the number of other consumers who have it. The classic example is telephone service, which becomes more valuable as more potential callers subscribe. Information goods and services such as network protocols and data formats generally exhibit network effects when they must be compatible with the protocols and formats employed by other users. But the concept of network effects can also be applied more broadly. Vast and cumbersome though it is, the institutional system of the European Union exhibits network effects whose power is visible in the number of countries that are lining up to join. Organizational practices likewise exhibit network effects when they facilitate the circulation of learning and expertise from one site of practice to another. Network effects favor compatibility over adaptation to local circumstances, and they tend to create winners and losers. The diverse processes by which these winners and losers are chosen will be of great concern to all well-informed and self-interested participants. A designer whose products obtain a critical mass of users will succeed, whereas a designer who comes to market too early or too late, or who lacks persuasive power, or whose competitors are better able to act in a coordinated way at many points in a complex global market, is likely to fail.
These standards dynamics lie on the boundary between economics and politics. In certain aspects, they can be understood as rational actors' reflexively interconstraining choices, whose outcomes can be modeled quantitatively using the tools of game theory. But at the same time, they can be seen as points of public debate: a standard succeeds in large measure by becoming a self-fulfilling prophecy whose prophetic power rests largely on symbols and ideology and rhetoric and history.9 Legal scholars have observed that standards act as a kind of law and that the establishment of standards is therefore in the deepest sense a political process through which a community comes to govern itself.10 Each perspective is valid, and it is important to maintain a sense of the tension between the two. Only then can we usefully ask (1) under what conditions it is even possible to establish a standard, (2) what kind of standard is likely to emerge, and (3) who is likely to win and lose as a result.
In practice, network effects usually occur in close conjunction with another economic phenomenon, economies of scale. Economies of scale are found whenever a good can be produced in quantity more cheaply through a substantial initial investment. Information goods are famous for the economies of scale because virtually all of the cost of producing them goes into the first copy.11 Economies of scale often reward the establishment of standards, in that the same information good can be circulated and used more widely. A standard computer operating system, established through network effects, will create economies of scale for applications vendors, and the resulting variety and low prices of software will further reinforce the dominance of the standard operating system, whereupon the same effects will probably lead to standardized winners among the applications as well. Analogous phenomena are found in the organizational realm: a firm that buys a competitor and standardizes the practices of the combined firm will enjoy amplified economies of scale in a wide range of information and knowledge work, thus leading to layoffs in those areas and to a competitive advantage, all other things being equal, over any remaining competitors.
Taken together, these phenomena cast considerable light on the nature of computing as a social practice. Technologists characterize computers narrowly in terms of the outputs they provide when certain data are provided as inputs. But the bigger story concerns the origins of those inputs and the conditions under which the outputs are meaningful out in the world. If a computer derives inputs from the four corners of the country, or from all around the world, then its outputs will be meaningful only if those data were defined and captured in a uniform way. This in turn requires some institution to establish and regulate a uniform set of practices across great distances and a potentially great diversity of physical and cultural environments. These are, in a broad sense, the infrastructural conditions of computing.12
With these concepts in mind, we can evaluate the radically decentralized story about both information technology and social order that I sketched at the outset. The story turns out to be complicated, with centralization and decentralization interacting in several ways. The establishment of standards, for example, usually requires centralized coordination, whether by a government, a large private firm, or a professional organization. The establishment of the metric system, for example, required both the rationalizing ideology of the French Revolution and the centralized state apparatus that the Revolution had inherited from the ancien régime. Once a standard is established, however, it will usually be self-perpetuating due to network effects, which will create a continuing incentive for everyone to do what everyone else is doing. The centrally established order will continue to be felt whether the center continues to impose it or not. Economies of scale, for their part, tend to produce industry concentration, which is certainly a kind of centralization, but the same technologies that create those economies also permit organizations to decentralize operational decision-making by using standardized practices of measuring and monitoring to retain centralized control.
These interactions between centralization and decentralization create several dangers that any well-thought-out design practice must avoid. First is the danger of premature implementation. Many good ideas in the computer industry fail because they require as prerequisites the establishment of standards, or a critical mass of potential users, or a sufficiently widespread infrastructure, or all three. Likewise a proposed design can fail when it bundles several functionalities that are viable only as separate standards, each establishing economies of scale by being incorporated into a wide range of applications. An example might be Electronic Data Interchange (EDI), which bundled a special-purpose networking protocol instead of waiting for a general-purpose network, such as the Internet, to become widely established.
A second danger arises when an institution is incapable of creating the centralized coordination that it needs to establish a new technical or process standard. When this happens, the institution can remain stuck in an inefficient diversity of practices, unable to interoperate, share learning, or establish economies of scale. Decentralization is thus not an unambiguous good and can cause serious backwardness.
A third danger occurs when a standardization process must move so quickly that it does not have time for democratic mechanisms, broad representation, and reflective deliberation. This is an urgent matter for due-process-oriented standards organizations that increasingly find themselves bypassed by private standards-setting consortia, which are usually dominated by large vendors even more than are the formal standards bodies.13 Democratic values are also endangered by the very fact of standardization. Inasmuch as the potential damage from a security breach is multiplied on a network of standardized systems, the need for military and police intervention to secure those systems against malicious disruption is greatly expanded.
A final danger derives from the dynamics of network effects themselves. Under ideal economic conditions, the standards that emerge from competition among different networks will closely reflect consumers' interests as expressed in their market choices. But ideal economic conditions are difficult to obtain in extremely complex and fast-moving high-technology markets, and as a result, network effects create a significant bias in favor of rapid time-to-market and against product quality. The result can be the nearly irreversible entrenchment of a low-quality standard -- a phenomenon familiar in the personal computer software market.
These observations relate to the nature of institutions and the institutional dimensions of design. The word institution is often loosely used to mean "organization," for example in the ways that an organization manages to perpetuate itself and its traditions even as particular individuals come and go. But the mechanisms I have been describing lead to the establishment of institutions in a deeper sense: with self-perpetuating rules and structures of human relationships. The uniform interfaces and practices that arise in the world govern people's lives without necessarily being reproduced by any conscious agency. They become woven into the fabric of everyday life. And the design of the new information services must of necessity be concerned with the ways in which information and its use are embedded in this fabric and with the dynamics by which this fabric changes and resists change, quite independently of its participants. Information services have always been woven into the institutionally organized activities of the people who use them, whether through paper, radio and television broadcasts, or the telephone. The Internet, however, is capable of weaving information services much more intricately into people's lives, reaching -- in much more interactive and specialized ways -- into particular settings where the services might be of use. Although information technology has a reputation for encouraging disembodiment, something closer to the opposite is likely to be the case. As miniaturization and low-cost wireless data communications liberate information services from cumbersome computer monitors and as less expensive networked printers decentralize the manufacture of printed documents, everyday activities can be reorganized less and less in terms of their tethering to information resources and more and more in terms of those aspects that make more legitimate demands on the body: travel, face-to-face conversation, interaction with the physical machinery that one is building or repairing, aesthetic judgments, recreational activities, cooking and eating, rest, and so on. Information services will be called on to conform ever more creatively to the shapes and patterns of bodily activities. For example, although a library will continue to be a place, simply because of the bodily demands of meeting and studying, we will increasingly speak of "librarying" as an aspect of a wide range of activities, and library services will be designed in such a way that they can be projected whenever, wherever, and however they are needed.
Numerous value issues will arise in the design of the new information services. As communities of practice increasingly form through the mediation of computer networks and other media, it will be increasingly necessary to decide whether services are to be designed for individuals, in the name of egalitarian uniform access, or for specific communities, in the name of efficient but potentially exclusionary specialized adaptation. One potential resolution of this tension lies in an understanding of the institution's role in relation to everyday information use. Routinized information use is embedded in activities and relationships, and to be socialized into a community means, among other things, to be trained in the discovery, evaluation, and use of a specific repertoire of information resources. The design of information services can be divided between supporting the settled-down steady state of these routinized uses of information and supporting the process by which a newcomer, especially one who is just joining a community or who lacks community support, is able to become integrated into the community's practices. To support an already routinized information use, an information service can reach into a specific niche in a system of activity. The latter function, getting someone oriented in the seemingly infinite and chaotic space of information services, is considerably more complex. Yet it is crucial for purposes of social mobility.
The new information services, then, will be composed of both institutions and standards, and they will arise through an interaction between institutions and standards. Technologists will be constrained by the economics of information technology, and so good design will require taking the market dynamics of standards into account. All of the existing institutions that deliver information services, both public and private, will be challenged to define new roles, which will no doubt be broader in some ways and narrower in others. But institutions that cannot negotiate the complex new terrain may find themselves losing their legitimacy. It will be particularly important to articulate a broad conception of access. Who will be the advocates for broad-based accessibility of information services in a more technologically mediated future? Designers, librarians, entrepreneurs, regulators, educators, and social movements will all presumably have their roles. But only if their diverse perspectives can be integrated will we achieve the truly integrated design practice that is clearly needed.
1. Harold J. Leavitt and Thomas L. Whisler, "Management in the 1980's," Harvard Management Review 36, no. 6 (1958): 41 - 48.
2. Thomas H. Davenport, Information Ecology: Mastering the Information and Knowledge Environment (New York: Oxford University Press, 1997).
3. Sandra Harding, Is Science Multicultural? Postcolonialisms, Feminisms, and Epistemologies (Bloomington: Indiana University Press, 1998); Friedrich A. Hayek, The Counter-Revolution of Science: Studies on the Abuse of Reason (Glencoe, Ill.: Free Press, 1952).
4. Philip E. Agre, "Beyond the Mirror World: Privacy and the Representational Practices of Computing," in Philip E. Agre and Marc Rotenberg, eds., Technology and Privacy: The New Landscape (Cambridge: MIT Press, 1997).
5. See, for example, James Boyle, Shamans, Software, and Spleens: Law and the Construction of the Information Society (Cambridge: Harvard University Press, 1996).
6. See, for example, Robert Jacobson, ed., Information Design (Cambridge: MIT Press, 1999); Kai Krause, ed., In Your Face: The Best of Interactive Interface Design (Rockport, Mass.: Rockport, 1996); and Richard Saul Wurman, Information Architects, ed. Peter Bradford (New York: Graphis Press, 1996).
7. Donald A. Norman, The Invisible Computer: Why Good Products Can Fail, the Personal Computer Is So Complex, and Information Appliances Are the Solution (Cambridge: MIT Press, 1998).
8. Michael L. Katz and Carl Shapiro, "Systems Competition and Network Effects," Journal of Economic Perspectives 8, no. 2 (1994): 93 - 115.
9. See, for example, Philip E. Agre, "The Internet and Public Discourse," First Monday 3, no. 3 (1998).
10. See, for example, Joel Reidenberg, "Lex Informatica: The Formulation of Information Policy Rules through Technology," Texas Law Review 96 (1998): 553 - 93.
11. Kenneth J. Arrow, Collected Papers, vol. 4, The Economics of Information (Cambridge: Harvard University Press, 1984), 142.
12. Geoffrey Bowker, "Information Mythology: The World of/as Information," in Lisa Bud-Frierman, ed., Information Acumen: The Understanding and Use of Knowledge in Modern Business (London: Routledge, 1994).
13. Paul A. David and Mark Shurmer, "Formal Standards-Setting for Global Telecommunications and Information Services," Telecommunications Policy 20, no. 10 (1996): 789 - 815.