Use and adaptation of precedents in architectural design. Toward an evolutionary design model

For centuries, architects have re-used design precedents in the conception of new design solutions. Whether explicitly - as in the case of Le Corbusier, James Stirling and Jo Coenen - or implicitly - as with J.J.P. Oud, Aldo van Eyk, and Santiago Calatrava - this practice has led to very advantageous efficient, effective, and innovative results. The goal of this research project is to contribute to the construction of computational tools to facilitate this practice by developing a model that grasps significant characteristics of the design process as it employs precedents. The model is built drawing an analogy from the natural evolution. The intention is not to represent the processes that take place in the architects minds but rather their behavior as this is manifested in their design products. The project draws from the multidisciplinary methodology of design tool development of the Design Knowledge Systems Research Center. It employs an analogy with Darwinian evolutionary theory in combination with recent theories of genetics and embryology. The criteria of usefulness in picturing the phenomenon in architecture determines the focus on particular aspects of the analogy. The research also uses three case studies from the architectural domain: J.J.P. Oud, to identify adequacy criteria for the model; Le Corbusier, to illustrate the components and conduct of the model under development; and Santiago Calatrava, to test the model. The research develops a pre-computational qualitative model that provides insights into the process of re-use and adequacy criteria for an analytical model to succeed in architectural practice. Given the notorious history of misuse of analogies from the Darwinian model and biological models in general to other fields, special attention was paid to circumscribe the limits of the analogy. There are basic differences between design and evolutionary models, the most important being the process of selection, natural versus artificial. As "breeders", designers recall from memory and/or from archives through "artificial selection" - this is not the case in natural selection. In natural evolution, mutations are "random", and natural selection gives the direction. In the human design process, mutations and selection are mostly intentional. Many analogies suffer from confusion between the natural science notion of evolution and the cultural notion of progress; they are also highly reductive in their representation of cognitive processes of design, misrepresenting the design process. The evolutionary and genetics analogy serve as heuristic devices to represent the mechanisms in the process of use and adaptation of design precedents and the elements of such precedents accumulated over the years that are adapted and recombined during the design process often leading to design innovations. The model employs the notion of "design feature", a precedent component, as the most important unit of selection. Drawing from developmental genetics, and the idea of regulatory genes, each feature is derived from two interlinked kind of instructions where the "regulatory d-gene" deals with the configurational instructions and the "structural d-gene" deals with the technique and materials used. In the design model, just as in evolution, the notion of fitting environmental constraints in the generation of form is essential. Fitness relates to both internal and external constraints; it is multi-dimensional in a multi-criteria ecological environment.Architectur