Knight and Stiny's paper ‘Classical and non-classical computation’ (arq 5/4, pp 355–372) is a most valuable review of recent work in shape computation but also raises a number of concepts of value in architectural education. One of the problems faced by a ‘learner designer’ is the development of an effective ‘design process’. The ways in which a designer represents form in order to deal with it during design are not necessarily apparent in the final design, or in the resulting building: it is in design development that these representations and operations become explicit and the processes of composition become transparent. Since architectural computing makes explicit the structuring and ordering logic inherent in formal models and makes transparent the operations upon objects which result in designs, the computer provides a unique environment in which to explore the principles of design.