Computational Composition in Architecture

The impact of computers on architectural design, manifested through the creation of architectural designs that equal or surpass in quality and sophistication architecture generated using traditional means, has yet to be demonstrated. Computer-aided design is conspicuously absent from theoretical discussions of architectural design per se, and there exist no major built or published buildings that have been designed using the computer in a true design capacity. Most prominent architects continue to ignore computer-aided design. This paper argues that the issues of architectural theory and composition must be addressed directly by the computer-aided design field, and that until this occurs computer-aided architectural design will necessarily be peripheral to architecture because it does not address the central problems of architecture as an expressive medium. To this end, it proposes a shift to the paradigm of computational composition, and discusses recent work in this direction. The paper is divided into four parts. Part I identifies fundamental theoretical problems, contrasts the application of computation to architecture and to music, and draws upon several different areas for insight into the nature of making, Part II reviews particular architectural implications of these considerations, introduces the concept of computational composition in architecture, and presents a brief overview of important precedents, Part III proposes new goals for computer-aided architectural design and presents a framework for computational composition, finally, Part IV presents recent work directly related to the ideas presented in the previous parts and leads to the Conclusion. The appendices contain a pseudo-Prolog expression of Alvar Aalto's architectural language and notes on features of the PADL-2 solid modeler that are architecturally interesting.