Integrating haman and computer activity in semi-automated timetabling is a difficult problem. This paper attributes this dimculty to the choice of computational paradigm, and introduces an alternative framework for computer-based modelling in which to create an instrument for timetabling. The potential use of such an instrument (the 'Temposcope') is illustrated in connection with work-in-progress on a modest practical case study in timetabling (the scheduling of project oral presentations). The adoption of this framework is associated with a radical shift in perspective on the user-computer relationship, from realist to idealist users, and from the computer as an abstract machine to the computer as a physical device. This shift suggests new modes of human-computer co-operation with potentially broad implications for computer-supported timetabling.