Sufficiency of the SEED knowledge-level representation for grammatical design

One view of design is that it is a phenomenon only partially comprehended by any model of it. Under this view, the sufficiency of any knowledge representation for design is necessarily an informal question. Answers to it can be sought through experience of a knowledge representation in use, but not by strict appeal to formal argument. What then would allow us to declare a knowledge representation for design sufficient for its purposes? At a minimum, a thorough description of purposes; of the knowledge representation; and of experience in using the knowledge representation for the declared purposes seems to be in order. The methodology developed within the SEED project (a Software Environment to Support the Early Phases in Building Design) develops each of these in parallel, starting from the outset of the project. It turns out that no aspect of the problem can be fixed at the start-as the knowledge representation develops, so do both purposes and use. From the experience on the team developing and applying the “SEED methodology” in four instances, we believe we can offer some insight into the sufficiency of the SEED knowledge representation for its intended tasks