Entropy in urban and regional modelling

More than fifty years have elapsed since the publication of ‘A statistical theory of spatial distribution models’. A personal stocktake of subsequent developments is presented here. This begins with identifying what have become key concepts in the development of spatial interaction modelling and dynamics: the way in which knowledge can be represented as constraints, the importance of accounts, and generalised cost (and the attractiveness part of a ‘utility function’). These have enabled the original concept to be used in an expanded way in a wide range of models. In a paper with Britton Harris in 1978, new foundations were laid for dynamic modelling – later seen as adding space to Lotka-Volterra concepts to the Boltzmann core, and hence for these models to be labelled as BLV models. Extensions into history and archaeology and into political science are described, concluding with some speculations on future developments and broader links to the complexity science agenda.