Gerry's Home Page Preliminary Materials Chapter 1 Chapter 2 Chapter 3 Chapter 4 Chapter 5 Chapter 6 Chapter 7 Chapter 8 Chapter 9 Chapter 10 Chapter 11 Bibliography Appendix

Sec 6.1

6.1.      A People-Centered Approach

Hermes is a people-centered computer system, designed with the nature of human understanding foremost in mind. People and computers have different strengths. If one accepts Heidegger’s principles of human understanding and contrasts them with traditional AI analyses of computer computation, it follows that people process information very differently from computers. So if one wishes to use computers to support humans doing difficult cognitive tasks like designing then it is necessary to distinguish the roles of computer and human carefully and to define a “cooperative problem solving system” (Fischer, 1989) in which they can work together most effectively.

While much attention has been paid in computer science to the theory of computation and to the mechanics of making computers perform efficiently, little work has been devoted to a theory of the human understanding that is necessary to make sense of computer output and to extend the domain of computer application beyond routine algorithmic computations. Computer scientists tend to leave the analysis of the human partner in human-computer interactions to cognitive psychology, which generally ignores Heideggerian ideas in favor of functionalist approaches. There have been some notable exceptions to this rule, which have provided much of the inspiration for this dissertation and that have covered much of the argumentative background that therefore does not have to be detailed here—e.g., Winograd & Flores (1986), Suchman (1987), Ehn (1988), Budde & Züllighoven (1990), Dreyfus (1991), Coyne & Snodgrass (1991), Schön (1992). Unfortunately, these exceptions have not included convincing examples of software as models for a people-centered approach.

Hermes is an example of people-centered software. It grew out of a recent tradition of domain-oriented design environments (discussed in Chapter 7) that tends toward a person-centered approach. This tradition was a practical response motivated by breakdowns in autonomous expert system approaches. The present dissertation is a reflection on the theoretical framework implicit in that tradition, aimed at repairing the breakdown at the conceptual as well as the practical level.

The systems that Hermes evolved from are people-centered in various ways. Fischer & Nakakoji (1992), for instance, argue that the Janus system empowers the human designers who use it. Similarly, McCall, et al. (1990) point out that the Phidias system allows the use of an open-ended set of domain categories so that designers are not restricted to a predefined representation of relationships. The previous chapter tried to sketch a philosophical justification for these ideas. It argued that tasks like innovative design require acts of application and judgments of relevance that require interpretive powers and intentionality that come naturally to people but cannot be programmed for computers. A reasonable conclusion to draw from this argument is that computer systems for non-routine design should be people-centered.

People-centered software in the sense proposed here is computer software that provides information to people and then lets the people make the judgments. Rather than incorporating heuristic tricks that allow the computer to make decisions that in most cases look like reasonable human judgments, the software is structured to involve the people using it in a decision-making partnership. The partnership is based on the asymmetry in which computers excel at searching large information spaces and people excel at making judgments of relevancy. In other words, designers interpret and computer systems like Hermes support this interpretation in design.

The term “people-centered” is intended to extend the approach of “user centered system design” (Norman & Draper, 1986). That was an attempt to view interface issues from the user’s perspective, not necessarily to include system users in either the software design process as in participatory design (Ehn, 1988) or in the computational decision-making as in the people-centered approach. User centered system design views people as information processors, not as interpreters; it seeks to adjust the software interface to the parameters of human processing characteristics at the periphery of the computation rather than trying to support human interpretation as the center-piece of the computation.

The first principle for a theory of computer support that overcomes the problem of tacit and explicit understanding is that the software should be people-centered. Three further principles are given in the next section.

Go to top of this page

Return to Gerry Stahl's Home Page

Send email to

This page last modified on January 05, 2004