Generation and Search Methods in Design: Discussion

Abstract

Design generation is the process of formulating design solutions within a design space. An early view of design as a search process was discussed by Simon (1969), in which knowledge can be modeled as a set of operators, and design problems modeled as goals; design is then the process of searching for sequences of operators which produce solutions to those goals. Depending on the nature of the operators, this process may involve the refinement and transformation of incomplete designs until a complete solution is obtained, or a traversal through a space of possible solutions occurs, or a combination of both. The proce ss may be deterministic, in which the space is searched systematically, according to fixed procedures, or stochastic, in which probabilistic methods are used to sample the space. Design as search has been critici sed as being too restrictive to capture the real nature of design. Smithers et al. (1994) argue that a large part of design is concerned with formulating the problem and discovering the relationship between the design goals and design knowledge, and thus that design is better captured by the term exploration. The session summarized here considered methods of formally describing and searching spaces in order to generate design solutions, and discussed the adequacy of search as a characterization of design .