Chapter 11. CONTRIBUTIONS

The topic of this dissertation has been the problem of providing computer support for cooperative design given the nature of tacit and explicit understanding. But at a meta-level, an important theme has been the role of theory in software design. Often, work in cognitive science and artificial intelligence proceeds with little reference to philosophy, which is given lip service as one constituent of these interdisciplinary endeavors. Of course, preconceptions

abound in such work, but they are either treated as self-evident common sense or addressed through discussions of individual concepts whose inner coherence remains outside the investigation.
This dissertation is an attempt to take theory seriously in computer science. Rather than first creating a software artifact whose theory is at best only tacitly available retrospectively,[1] and then subjecting the artifact to controlled user testing to determine its effectiveness, the approach followed here is to formulate a set of explicit theoretical principles to motivate an approach to computer support of design and then to present a package of prototyped functionality to illustrate that approach. Together, the theory and the examples are meant to provide cogent arguments for the deliberation of central issues in software design of systems to support innovative, collaborative design work in exploratory domains.

Of course, several preconceptions have been at work here, too. However, the major assumptions have been systematically reflected upon in the process and explicated or modified as need be. It has been assumed, for instance, that software to support professional designers should be based on an understanding of the structure of their work processes. As a guiding idea, the design process was viewed (or pre-viewed) as a process of interpretation (Chapter 1). Two approaches were then taken to explore this work process: one by looking at some of the best available descriptions of the way designers work (by Alexander, Rittel, and Schön in Chapter 2), and the other by looking at a concrete example of designers working (on lunar habitat design in Chapter 3). To make this theory even more explicit and general, it was then put into the framework of a philosophy (Heidegger’s hermeneutics in Chapter 4). An explicit theory of computer support for interpretation in design was built on top of the results of the preceding investigations (Chapter 5 and 6). The theory developed in this way was then used to evaluate related software systems meant to support design (Chapter 7). Finally, the theory served to motivate and justify design decisions in the Hermes software (Chapters 8, 9, and 10).

While this approach stresses theory, it does not ignore the need for empirical grounding or iterative testing. The design methodologies reviewed all grew out of either reflection on professional practice or consideration of experimental findings. The study of lunar habitat design pursued as part of the dissertation took on the flavor of participatory design (Ehn, 1988; Greenbaum & Kyng, 1991) by having the software designers and the design professionals working together on a lunar habitat design, and by involving the two groups in dialogue about the design work and about possibilities for computer-based support of this work. Although it was never reflected in the Chapter 3 transcripts, the lunar habitat designers have been involved in on-going evaluation of the Hermes system and its functionality as part of their role as corporate sponsors of the funded research. In addition, the design of Hermes is a response to empirical experience with the related design environments on which it is based, as well as on a series of programming walkthroughs to evaluate the Hermes language design (reported in Appendix A).

Evaluation and refinement of the Lunar Habitat Design Environment (Lhde) and Phidias II built on top of the Hermes substrate are expected to continue indefinitely. Clearly, the greatest need for future work is to build a robust design environment that exercises all of the functionality of the Hermes substrate and to gain experience in the utility of this functionality through use by professional designers. Unfortunately, that is beyond the scope of the present effort. For one thing, it will involve identifying real-world projects in which a system like Lhde makes commercial sense in order to get professionals to invest significant time in using preliminary versions. The support of lunar habitat design has served as a fruitful application domain in developing Hermes, but a specific project must now provide a practical context for further participatory development and workplace evaluation.

It is useful to view the unfolding of this dissertation as a hermeneutic process, in which a vague preconception of interpretation in design becomes increasingly clearer through precisely the kind of interpretive process that has been analyzed in the dissertation. The concept of interpretation has been elaborated through an investigation of the role of interpretation in design. The guiding perspective was the intuition that interpretation is the central category for founding a theory of computer support. This perspective was tied through a process of reflection to its explicit roots in Heidegger's philosophy, but also to the almost forgotten role of interpretation in the related systems that Hermes grew out of. In a sense, the dissertation embodies a moment of reflection in which the effort to build systems of computer support ran up against the limits of multi-faceted, domain-oriented, knowledge-based systems; made explicit the role of interpretation in design; and then, using this, proposed a system that integrates the facets in a hypermedia substrate, extends the notion of domain-orientation with perspectives, and uncovers the basis of explicit computer knowledge representations in the expressing of tacit human preunderstanding in language.

In looking back over what has been accomplished in this dissertation, it is clear that no final answers have been given. The analysis of interpretation remains unclear and incomplete in many ways. The theory of computer support is no more than a beginning in an attempt to provide rationale for a new direction in artificial intelligence. The design of Hermes is suggestive of promising functionality, but this promise remains largely untested. Nevertheless, whatever the limits of this work, it does seem to have made significant contributions on three primary levels: on a philosophical level (11.1), on a theoretical level (11.2), and on a system building level (11.3).

Go to top of this page

Return to Gerry Stahl's Home Page

Send email to Gerry.Stahl@drexel.edu

This page last modified on January 05, 2004