USABILITY VERSUS COMPUTABILITY: SOCIAL ANALYSES BY COMPUTER SCIENTISTS Rob Kling Department of Information and Computer Science University of California - Irvine Irvine, Ca 92717 714-856-5955 email...@ics.uci.edu Draft 5d - October 1993 To appear in "INFORMATICS, ORGANIZATION AND SOCIETY," SAVVAS A. KATSIKIDES (Editor). Vienna-Munich: OLDENBOURG VERLAG, Reihe OCG, Austrian Computer Society. 1993. ------------ INTRODUCTION[1] Computer specialists' views of the relationship of the society in which they live and work to computerization are very important. They permeate professional discussions about what should be computerized, whose interests merit attention, how computerization should proceed, and the good and bad consequences of different approaches. Computer specialists do not agree on these matters. In fact, they hold several different views of the relationship of society to computerization [2]. Computer scientists are one important group of computer specialists. They have a specially important role in developing advanced computer technologies, in educating a large fraction of computer specialists in universities, and in acting as spokespeople about computing matters within elite scientific circles. Their perspectives shape curricula and public discourse about what should be computerized and how. I will briefly sketch five perspectives about the relationship of society to computerization which are held by computer scientists. USABILITY VERSUS COMPUTABILITY This paper characterizes five different perspectives which computer specialists hold about the relationship of the society in which they live and work to computerization. These perspectives differ in a key dimension that underlies many key debates about the nature of computer science research and education -- computability versus usability. The computability tradition in computer science examines the capabilities and efficiencies of computer-based systems in mathematical terms. The mathematical theories of computation anchored in automata theory and theories of algorithmic complexity are the cornerstones of this tradition. In contrast, the usability tradition is concerned with understanding and developing computer-based systems which gracefully fit into and support (or transform) human activity in various social settings. This line of inquiry is represented in computer science by human factors studies of computer systems, concerns for reducing the social risks of computer-based systems, etc. The primary underlying theories derive from the social sciences rather than from mathematics. This dimension -- computability versus usability -- absorbs many important lines of research and teaching in computer science, but does not completely characterize the field. Studies of novel computer architectures, programming environments, and artificial intelligence, for example, sometimes emphasize neither computability nor usability criteria. But they are often focussed on developing novel technologies which expand the range of activities that can be readily done with computer-based systems. In practice, effective computer-based systems should be very usable, although we have many examples of clumsy technologies which are widely used (e.g., the vi editor in UNIX, programming languages such as COBOL, FORTRAN, and C, DBASE III, etc.). Since many people like responsive systems, the speed of interaction is one critical dimension that facilitates the usability of computer based systems. For highly interactive tasks, like text editing and conducting computer-based searches, swift response times can make a system a pleasure to use. Very usable systems often rely on fast algorithms for searching, sorting, managing screens, etc. But people also relish other features of interfaces which are not amenable to mathematical analysis such as coherent command sets, harmonious colorful displays, and readily intelligible output (Shneiderman, 1987). FIVE PERSPECTIVES I have identified five perspectives and ordered them by the richness of their views of the society -- from the simplest to the most complex.[3] ============================================= TABLE #1 #1: COMPUTING IS A MATHEMATICAL AND ENGINEERING VENTURE WITH NO DEEP SOCIAL CONTENT #2: SOCIETIES ARE CUSTOMERS FOR A CAVALCADE OF IMPROVED TECHNICAL PRODUCTS #3: COMPUTERIZATION CAN AND SHOULD REVOLUTIONIZE SOCIETIES #4: PROFESSIONAL RESPONSIBILITY TO SOCIETY IS ESSENTIAL #5: COMPUTERIZATION IS A FUNDAMENTALLY SOCIAL PROCESS ------------------------------------- Table #1 FIVE PERSPECTIVES ABOUT COMPUTERIZATION AND SOCIETY ================================== Perspective #1: Computing is a mathematical and engineering venture with no deep social content One commonplace view of computer science and engineering is reflected in a 1989 report by the ACM Task Force on the Core of Computer Science. (Denning, et. al., 1989). Computer Science and Engineering is the systematic study of algorithmic processes -- their theory, analysis, design, efficiency, implementation and application -- that describe and transform information. The fundamental question underlying all of computing is, What can be (efficiently) automated? .... The roots of computing extend deeply into mathematics and engineering. Mathematics imparts analysis to the field; engineering imparts design (Denning, et. al. 1989). In this view, computer science focusses primarily upon algorithms rather than systems that anyone would actually use. It has no explicit social content. The report suggests an interest in implementations and applications -- and consequently the usability of computer-based systems through its focus on "engineering" as well as mathematics. But the only legitimate analytical approaches are mathematical and hence, a-social. The ACM Task Force's report is not completely consistent. For example, it suggests that courses be offered in human-factors studies without acknowledging that psychology, sociology (and perhaps physiology), rather than mathematics, provide the analytical underpinnings of these inquiries. This view of computing is very commonplace among academic computer scientists. It shapes most computer science curricula so that they ignore the social dimensions of computerization. Consequently, most of the 40,000 college students who graduate each year with computer science degrees in the US. develop little understanding of the social assumptions about design or development that they will make as practitioners (Kling, 1993). The expansion of computing (and Computer Science) has depended upon the improved usability of computer based systems base, not simply their efficiency. The "micro revolution," for example, has placed tens of millions of computers and terminals on the desks of many managers, professionals and clerks in industrial countries. Networking has made electronic mail accessible to millions of managers and professionals. Whatever theories help explain the expansion of computing applications, which ones work well or badly (and in what terms), they are not fundamentally mathematical studies of algorithms or systems optimization. This perspective contributes important insights to "computing in the laboratory." But it cannot readily help computer specialists understand who computerizes, what it takes to computerize effectively, and the consequences of different computerization strategies in the larger world. Perspective #2: Societies are Customers for a Cavalcade of Improved Technical Products Many computer specialists expect continuous technological progress. They focus on the improvements in discrete technological innovations - whether they are novel technologies, such as optical disks or hypertext, or incremental improvements in more mature technologies such as databases. This sensibility is reflected in many technical and trade journals. It is visible through special issues and articles devoted to the nuances of a specific new technology without corresponding articles devoted to examining the social meanings of these technologies, the ways they are likely to be used, the social choices involved, etc.. This perspective is somewhat similar to Perspective #1 in focussing on discrete artifacts. But its approach to evaluation may include human factors and other psychological criteria for refined systems engineering. While specialists of this persuasion are often keenly aware of the applications of computing technologies, they do not publish coherent analyses of how they add up in concrete settings. To take a simple example, a popular technically-oriented North American microcomputing journal, Byte Magazine, publishes hundreds of notes and articles each year about a variety of new products and computing techniques. Each product is treated in isolation, and in a way that encourages Byte's readers to desire better technologies. These technologies are anchored in diverse vendor worlds -- IBM, Apple, Atari, Sun, etc. But the best technologies from each of these vendor worlds do not necessarily add up into a well integrated and affordable computing world. However, Byte's editors and writers do not examine how participants in actual organizations work with and try to integrate diverse technologies of varying age into workable computing milieus. Perspective #3: Computerization Can and Should Revolutionize Societies Many organizations are adopting computing equipment much more rapidly than they understand how to organize positive forms of social life around it. However, some fervent advocates of computerization portray the actual pace of computerization in schools, offices, factories, and homes slower than they wish. These "computer revolutionaries" argue that many key institutions -- such as schools, businesses, family life, public agencies -- can be progressively reformed through the appropriate application of computer-based systems. Computer revolutionaries make five key assumptions: (a) Computer- based technologies are central for a reformed world; (b) Improved computer-based technologies can further reform society (c) No one loses from computerization; (d) More computing is better than less, and there are no conceptual limits to the scope of appropriate computerization; and (e) Perverse or undisciplined people are the main barriers to social reform through computing (Kling and Iacono, 1988). Computer revolutionaries portray computer technologies as essential instruments for solving important and previously intractable social problems (Feigenbaum and McCorduck, 1984; Papert, 1980; Yourdon, 1986). But they underplay the ways in which the forms of computerization which they advocate can also create significant problems. Computer revolutionaries rely on an "undersocialized" conceptual scheme which isolates computer use from key elements of social life and ignores the details of computerization processes. They help inspire readers to trust in a "computer revolution" made by acquiring and using computer equipment under commonplace social conditions. It is too facile to simply characterize computer revolutionaries as "optimistic." But they evaluate computer based systems stripped out of the particular contexts in which people live with them. The computer revolutionary discussions exaggerate the social power of new computer technologies and their use by simplifying the social conditions which make them attractive to some participants, and the complexities and sluggishness of institutional change. Computerization is a complex set of social practices, not just the use of a computer system. Different computerization strategies may better serve different social interests and have different social consequences. The limits to computerization may be set by a powerful array of institutionalized social practices. Computer revolutionaries usually ignore these important aspects of computerization because they compromise the innocence, power, and purity of new technologies and their advocates (Kling and Iacono, 1991). The questions that underlie the claims of computer revolutionaries are interesting and important. But I haven't found good answers framed in the rhetoric of "computer revolution." The best answers come from close empirical observation of computer systems in use which suggests the real possibilities, limitations, paradoxes and ironies of computerization situated in identifiable social settings (see for example, Kling, 1987; Kling, 1992). Perspective #4: Professional Responsibility to Society is Essential. Some computer specialists are specially concerned that computing technologies should be "sound products." In this view the computing professions should be responsible to their clients by delivering practical systems which are usable, reliable, and safe. The main foci of attention have been to identify computer systems which can be major threats to physical safety or civil life and to identify discrete solutions to reduce these risks. Some forms of computer technologies can be harmful because of unreliable software (e.g., life-critical information systems; election counting systems; social security payments; fly-by-wire aircraft; military command and control systems.). Other kinds of computer based systems threaten to diminish personal privacy. Both kinds of threats have been the subject of a specialized topical literature, the concern of organizations like Computer Professionals for Social Responsibility[4], and special forums, like the ACM sponsored "Risks" computer bulletin board. In this perspective the primary reforms will come through improved software quality and certain changes in organizational practices (e.g., privacy protections; administrative guidelines to insure safe software and data handling practices). Perspective #5: Computerization is a Fundamentally Social Process Computerization is a fundamentally social process for designing and deploying technologies in social settings. A starting point to appreciate this perspective is the observation that the development and use of the vast majority of computer-based systems involves many social choices (Kling and Jewett, in press). Advancing the argument that some technology should be developed or adopted are social acts which usually make claims about how that technology will alter the lives of people who use technologies or others in their orbit (e.g., their clients, family members). Designers and developers make important choices in identifying preferences ("requirements") for systems, setting priorities, and in convincing people and organizations to alter their practices when they adopt new technologies. Some aspects of computer systems design are a form of social design, for many of the same reasons that the design of buildings and streets is a kind of social design, rather than only applied engineering or abstract artistic work with steel, concrete, and glass. The design choices open some social opportunities and close off others -- thereby shaping social life. The dominant explicit paradigms in academic computer science focus on computability, rather than usability (Perspective #1). They do not help technical professionals comprehend the social complexities of computerization. For example, the 1989 ACM Task Force on the Core of Computer Science quoted above claims that all the analyses of computer science are mathematical. This view is so widespread that many computer scientists use the term "theory" to mean "mathematical theory," and thereby deny the possibility of social or psychological theories of computing! This view much too narrow-minded to be helpful, and in fact it does not withstand much scrutiny. The lines of inquiry where it might hold are those where mathematics can impart all the necessary analysis. But there are whole subfields of computer science -- such as artificial intelligence, computer-human interaction, social impacts studies, and parts of software engineering -- where mathematics cannot impart all the all the necessary analysis. The social sciences provide a complementary theoretical base for studies of computing which examine or make assumptions about human behavior. Unfortunately, many behaviorally oriented computer scientists feel intimated by their mathematical colleagues theoretical imperialism. In the classic Turing machine models of computability, there is no significant difference between a modest microcomputer like an Apple Macintosh microcomputer and a Cray supercomputer. Of course, these two machines differ substantially in their computational speed and memory. But the Mac, with software designed to match its graphical interface, is a much more usable computer for many small-scale tasks such as writing memos and papers, graphing data sets of 500 data points, etc. Unfortunately, the computability paradigm doesn't help us understand a Mac's relative advantage over a Cray for tasks where the ease of a person's interacting with software, rather than sheer CPU speed or file size, is the critical issue. This contrast between Macs and Crays is a small-scale machine- centered example. More significant examples can found whenever one is asking questions like: What should be computerized for these particular people; How should computerization proceed?; Who should control system specifications, data, etc.? These are "usability" questions on a social scale that link computerization to the behavior of organizations. Paradigms that focus on the nature of social interaction and organizational behavior provide much better insights for designing computer systems in support of group work than does the computability paradigm (Ehn, 1988; Ehn, 1989; Kling, 1987; Kling, 1992; Kling, 1993; Winograd, 1988). Organizational Informatics is the field which studies the development and use of computerized information systems and communication systems in organizations (Kling, 1993a, Kling, 1993b). It includes studies of their conception, design, effective implementation within organizations, maintenance, use, organizational value, conditions that foster risks of failures, and their effects for people and an organization's clients[5]. Mathematical analyses help us learn about the properties of computer systems abstracted from any particular use, such as the potential efficiency of an algorithm or the extent to which computers can completely solve certain problems in principle (e.g., the question of undecidability). As long as a computer scientist doesn't make claims about the value of any kind of computer use, mathematical analyses might be adequate. But most computer scientists do make professional claims about how people and organizations should computerize, although often informally as members of committees and consultants. In fact, they often recommend technologies which are either easy to use or which represent some kind of technological advance (e.g., teaching programming in Pascal, Ada, or C++ rather than BASIC). The advances in computer interface design which led to the Mac, and other graphical interfaces, rested on psychological insights, rather than mathematical insights. Similarly, advances in programming languages, software development tools, and database systems have come, in part, through analyses of what makes technologies easier for people and organizations to use. While some of these technologies have a mathematical base, their fundamental justifications have been psychological and social. Claims about how groups of people should use computers are social and value-laden claims (Kling, 1983; Dunlop and Kling, 1991, Berleur, et. al., 1991). For example, suppose that a computer scientist argues that organizations should computerize by developing networks of high performance workstations, with one on every person's desk. This claim embodies key assumptions about good organizational strategies, and implicitly rules out alternatives DD such as having a large central computer connected to employees' terminals, with a few microcomputers also available for those who prefer to use micro-oriented software. These two alternative architectures for organizational computing raise important social and political questions: Who will get access to what kinds of information and software? What skill levels will be required to use the systems? How much money should be spent on each worker's computing support? How should resources and status be divided within the organization? Who should control different aspects of computerization projects? And so on. These questions emphasize the social and political organization of work, and their relationship to the technological organization of computer systems (Clement, in press). Many computer scientists are keenly interested in having advanced computer-based systems widely used, not just studied as mathematical objects. These hopes and related claims rest on social analyses and theories of social behavior. Mathematical frames of reference lead analysts to focus on behavior that can be formalized and "optimal arrangements." Concerns for optimality make most sense when the objective functions of a group are known, and are consistent. But they are unable to answer the question of what should be optimized and why. For example, should programming environments be optimized to minimize their usage of computer memory or the time that it takes people to design and develop software? Social analyses usually help analysts identify the ways that different participants in a computerization project may have different preferences for the way that computing technologies are woven into social arrangements (Kling, 1987; Kling, 1992; Kling 1993b). Their "objective functions" may differ, and even be incompatible, to the extent that people know what they are. Participants are most likely to have different preferences when computerization projects are socially complex and a project cuts across key social boundaries -- identified by characteristics such as roles, occupational identity, reward structures, culture, or ethnicity. We do not see social (and organizational) analyses as panaceas. But they help identify the kinds of opportunities and problems that participants will actually experience and respond to in their daily lives. CONCLUSION I have briefly described five perspectives about the relationship of computer specialists to society, and the role of social analysis in the field. These range from Perspective #1 which emphasizes computability and places social change as other people's responsibility to Perspective #5 which views the usability of computing in meaningful social settings as most important. The computability perspective appears to be the most "scientific" and "elegant" because it focuses on mathematical analysis and a laboratory conception of engineering design. It helps give computer scientists prestige by identifying their analyses with mathematics ("queen of the sciences"). Unfortu- nately, it is impotent in understanding many key aspects of computerization in the world -- even basic questions, such as what really makes computer-based systems effective, why computer- based systems of different kinds have been adopted by different kinds of organizations, and what have been the consequences of their implementation. The other four perspectives acknowledge that social choices play a key role in understanding the value and problems of different approaches to computerization. Perspective #2 (societies are customers for a cavalcade of improved technical products) captures much of the excitement of computerization, but ignores most of the key social choices and dilemmas. Not every solution can come in a cardboard box or shrink-wrapped package. Perspective #3 (computerization can and should revolutionize societies) places computerization on a large and dramatic social stage. The key dilemma is that computer revolutionaries tend to fetishize equipment (like Perspective #2), to overestimate the rate at which computer-based systems can act as catalysts for social change, to ignore the extent to which key social choices are linked to different forms of computerization, and to ignore the potential problems of large scale, intensive computerization. These two perspectives rest on technological utopianism as key assumptions (Dunlop and Kling, 1991). Perspective #1 is socially isolated and non-responsible and Perspective #2 is producer-oriented. Perspective #4 focuses on identifying and minimizing the socially harmful effects of specific kinds of computer-based systems. Perspective #4 also counterbalances these by being consumer oriented, socially responsible, and focused on tractable problems. Perspective #5 goes beyond Perspective #4 by promising an analytical approach to understanding the opportunities, choices, and dilemmas of computerization (kling, 1993a; Kling, 1993b). It relies on the social sciences rather than mathematics for its analytical underpinnings. The education of hundreds of thousands of computer science students has been shaped by Perspective #1. They leave academic computer science programs with some skills in designing software systems and programming them. They usually take courses in data structures and algorithms in which they learn to appreciate and carry out mathematical analyses of computer performance. But they leave systematically ignorant of Organizational Informatics -- of the ways in which social analyses of computer systems provide comparably important insights into the effectiveness of computing in the world. This view was more sustainable 20 years ago when mainframes and computation (accounting or scientific) were the major forms of computing. But the computing world has gone through a major sea change in which highly flexible work stations and computer networks now form the computing milieux for millions of men and women who manage text and pictures. While Organizational Informatics is pertinent to all forms of computerization, it is inescapable in this new wave of flexible locally-organized computing. Many segments of the computing community would much better understand computerization and be able to play more responsible professional roles by adopting a more fundamentally social view of computerization. REFERENCES Berleur, Jacques, Andrew Clement, Richard Sizer, Diane Whitehouse (eds.). 1991. The Information Society: Evolving Landscapes. Springer Verlag. Clement, Andrew. (1994, in press) Computing at Work: Empowerment for whom? Communications of the ACM. 37(1) (Jaunary). Denning, Peter J. et. al. 1989. Computing as a Discipline. CACM 31(1)(January):9-23. Denning, Peter. 1992. "Educating a New Engineer" Communications of the ACM. (December) 35(12):83-97 Dunlop, Charles & Kling, Rob (eds.). 1991. Computerization and Controversy: Value Conflicts and Social Choices. Boston, Ma. Academic Press, Boston. Ehn, Pelle. 1988. Work-Oriented Design of Computer Artifacts. Stockholm, Arbetslivcentrum. Ehn, Pelle. 1989. The Art and Science of Designing Computer Artifacts." Scandinavian Journal of Information Systems, 1 (August), pp. 21-42. Reprinted in Dunlop, Charles & Kling, Rob (eds.). 1991 Feigenbaum, Edward and Pamela McCorduck. 1984. Fifth Generation: Artificial Intelligence and Japan's Computer Challenge to the World. New York: New American Library. Kling, Rob. 1983. "Value conflicts in the deployment of computing applications: Cases in developed and developing countries." Telecommunications Policy, (March), 12-34. Reprinted in Dunlop, Charles & Kling, Rob (eds.). 1991 Kling, Rob. 1987. "Defining the boundaries of computing across complex organizations." Critical Issues in Information Systems. Richard Boland and Rudy Hirschheim (eds.) London, John Wiley. Kling, Rob. 1992. "Behind the Terminal: The Critical Role of Computing Infrastructure In Effective Information Systems' Development and Use." in Challenges and Strategies for Research in Systems Development. William Cotterman and James Senn (Eds.) New York, John Wiley. Kling, Rob. 1993a. "Computing for our Future in a Social World" Communications of the ACM, 36(2)(February 1993):15-17. Kling, Rob. 1993b. "Organizational Analysis in Computer Science." The Information Society. 9(2) (Mar-May):71-87 Kling, Rob and Suzanne Iacono. 1988. "The mobilization of support for computerization: The role of computerization movements." Social Problems 35 (3), 236-243. Kling, Rob & Suzanne Iacono. 1990. "Making the Computer Revolution" Journal of Computing and Society. 1(1). Reprinted in Dunlop, Charles & Kling, Rob (eds.). 1991. Kling, Rob & Tom Jewett. (in press). "The Social Design of Worklife With Computers and Networks: A Natural Systems Perspective." Rob Kling and Tom Jewett. Advances in Computers. Orlando, Fl: Academic Press. Kling, Rob and Scacchi, Walt. 1982. "The web of computing: Computer technology as social organization." Advances in Computers, 21, New York, Academic Press. Kumar, Kuldeep and Niels Bjorn-Anderson. 1990. A cross-cultural comparison of IS designer values. Communications of the ACM. 33(5):528-539. Papert, Seymour. 1980. Mindstorms: Children, Computers and Powerful Ideas. New York: Basic Books. Shneiderman, Ben. 1987. Designing the user interface: strategies for effective human-computer interaction. Reading, Mass.: Addison-Wesley. Winograd, Terry. 1988. A Language/action Perspective on the Design of Cooperative Work,'' Human-Computer Interaction 3:1 (1987--88), 3--30. Reprinted in Greif, Irene (Ed.), Computer-Supported Cooperative Work: A Book of Readings, San Mateo, California: Morgan-Kaufmann, 1988, 623--653. Yourdon, Edward. 1986. Nations at Risk: The Impact of The Computer Revolution. New York, Yourdon Press. ENDNOTES: 1. Based on a presentation at the August 1989 IFIP Triennialme very helpful comments by Werner Beuschel, Deborah Estrin, Dan Hirschberg, Nancy Leveson, Tom Jewett, Peter J. Neumann, Leigh Star, Karen Wieckert, and Terry Winograd. 2. There are well over 1,000,000 computer specialists worldwide. They have various jobs, including computer scientists, system designers, programmers, systems analysts, trainers, database administrators, computing facility managers, chief information officers, educators, and consultants of various kinds. Since computer specialists have diverse occupations, political and religious orientations, educational backgrounds, and career experiences, any claims about their shared perspective is highly simplified, and cannot capture the diversity of views represented in this large occupational community. 3. I have developed this set of five perspectives informally. It is not the result of systematic reading of a specific body of literature or an empirical analysis of beliefs held by a particu- lar group of computer specialists. One could, for example, examine some body of texts to identify key organizing ideas. Plausible groups of texts would include: the ACM's Turing Award Lectures from 1970-1993; talks given at all the major ACM and IFIP Conferences held in the 1970s, 1980s, 1990s, etc. Kuldeep Kumar and Niels Bjorn Anderson (1990), for example, have empirically studied the values of systems analysts in a small number of American, European and Canadian firms. Even so, I have some confidence that these perspectives characterize impor- tant assumptions and beliefs of many computer specialists. But the list and characterizations should be read as provisional. 4. The Computer Professionals for Social Responsibility may be reached at PO Box 717, Palo Alto, Ca 94301. 5. The computer science programs at a few universities, such as the University of California, Irvine, the Arhus University in Denmark, and the University of Dortmund in the Federal Republic of Germany attempt to integrate serious social analysis into more traditional "technical" computer science. See also Denning's (1992) proposal for a Computer Science curriculum which emphasizes studying technology in meaningful organizational contexts.