Robert K. Moniot
Department of Computer and Information Sciences
Prior to about 1980, almost all computers belonged to organizations, not individuals, and were operated by trained professionals. Then the personal computer (PC) made its debut. The first PCs required extensive learning on the part of the users. In an effort to make the machines more accessible to unsophisticated users, PC vendors developed a graphical user interface (GUI) based on an analogy to an office desktop. This “desktop metaphor,” as it was called, was the guiding principle behind the very successful Apple Macintosh and, soon after, the Microsoft Windows operating systems. It is characterized by certain key elements: windows representing sheets of paper on which to work, icons representing documents, folders, or applications, and menus providing choices of actions or documents. The desktop metaphor GUI has become almost universal as the interface between humans and computers.
By means of the desktop metaphor, PC vendors greatly reduced the steepness of the learning curve for their products and made computer technology available to a much larger class of users than before. This development has clearly had many benefits. It increases productivity in many office tasks and places users in direct control of the computer, instead of making them use older, less efficient technology or else rely on intermediary technical staff.
However, the desktop metaphor also introduces a number of thorny problems, raising ethical issues concerning the manner in which this development has taken and is taking place. The drawbacks of the desktop metaphor were recognized early on: it tends to limit the functionality of the software to that of the physical analog; the implementation can never be completely faithful to the metaphor and so will sometimes behave in unexpected ways; and whereas the interface is easy to learn, it is not optimal for many tasks and impedes maximum utilization of the capabilities of the computer system [Gentner and Nielsen 1996, Halasz and Moran 1981, Johnson et al 1985].
Users are being put in control of a system they do not really understand, but the desktop metaphor tends to give them the impression that they do understand it. Thus users are often unaware of the risks they are exposed to when they rely on these systems for important tasks. For instance, The desktop metaphor suggests that computer documents have the same permanence as physical documents. Hard disks are so reliable that this is almost true, and so, to economize, most PC systems are sold without a practical device for doing routine backups. Users do not realize, until too late, that the hard disk can crash at any time, making the user’s files unrecoverable. If they were more informed about the risks they would be more willing to pay for (and use) a backup device.
Similarly, when the system fails in some way not foreseen by the designers, users may be unable to fix it since they lack a real knowledge of its inner workings. Customer support for helping users solve such problems is notoriously inadequate. This problem is becoming increasingly common as the complexity of systems increases, causing conflicts or unanticipated interactions between different system components.
Looking beyond these practical issues, more fundamental concerns can be identified. A metaphor is useful because it makes an analogy between an unfamiliar new situation and a familiar old one. It thus provides a bridge that facilitates the learning of the new system. But the desktop metaphor is of no help to someone who doesn’t have a desktop [Brock 1996]. Young people in particular have the time and willingness to learn an interface based on a new, more efficient paradigm. But it is very difficult for an alternative interface paradigm to get a foothold, because of the immense inertia of the existing installed software base built around the desktop metaphor.
An ethical analysis of this situation must of course consider the many benefits resulting from the wider diffusion of computing technology throughout society. The desktop metaphor approach has been quite effective in enabling this very positive development. It is hard to see a practical alternative to the problem of making a highly complex technological system accessible to untrained individuals. The “blinking twelve” syndrome typical of VCRs shows that when faced with a non-intuitive user interface, users will often fail to learn how to use a complex system effectively rather than devote the time necessary to read and master the instructions.
Nonetheless, the creators of complex systems should be considered responsible for the adverse consequences of their design decisions, even if some of these consequences were not directly foreseen. Different kinds of responsibility can be distinguished: here we are dealing with causal responsibility, since the system designers are responsible for creating the conditions that enabled these consequences to occur. Causal responsibility does not necessarily imply blame, but it does suggest at least some degree of liability [Johnson 1994, Ladd, in Dunlop and Kling 1991].
Ultimately, perhaps the solution will be provided by more powerful software incorporating artificial intelligence, allowing an interface between human and computer that is both natural and efficient. Until the technology reaches this stage, computer vendors must recognize the limitations of the desktop metaphor and seek ways to move beyond it, even if this requires devoting more resources to user education and support.
Brock J. F., “Whose Metaphor?” interactions 3:4, pp. 24-29 (Jul. 1996).
Dunlop C. and R. Kling, eds., Computerization and Controversy: Value Conflicts and Social Choices, Academic Press (1991).
Gentner D. and J. Nielsen, “The Anti-Mac Interface,” Commun. ACM 39:8, pp. 70-82 (Aug. 1996).
Halasz F. and T. P. Moran, “Analogy Considered Harmful,” Proc. ACM Conf. on Human Factors in Computer Systems (March 15-17, 1982), pp. 383-386.
Johnson J. A., D. C. Smith, F. E. Ludolph and C. H. Irby, “The Desktop Metaphor as an Approach to User Interface Design,” Proc. 1985 ACM Ann. Conf. on the Range of Computing, pp. 548-549 (1985).
Johnson D. G., Computer Ethics, 2nd ed., Prentice-Hall (1994).