Evolving a Plan: Design and Planning with Complexity

Abstract

Producing physical plans that manipulate urban form and function to generate optimal designs with respect to an affected community is a many-stage process of resolving inherent conflicts between those who represent the interests of the community. Here we introduce a class of decision models that involve resolving conflicts between a series of opinions that differ from one another and are associated with a set of agents who act as designers. These opinions are expressed as differing interest and control in factors that influence the design and these are articulated as spatial plans based on the suitability or desirability of different map locations for physical development. We define a set of agents who motivate the process and whose interactions which involve resolving their conflicting opinions, are used to pool opinions where, at each stage, some degree of resolution takes place. Ultimately because every opinion relates to every other through the network of relations that bind agents together, a consensus is reached that can be interpreted as a process of weighted averaging whose formal properties mirror the operation of a first-order Markov chain. The elaboration of this process that we invoke here is based on a process of exchange due to Coleman (Foundations of Social Theory. Belknap Press, Cambridge, MA, 1994) in which we characterise the problem as one of resolving conflicts between agents which we call the primal or differences between factors in terms of opinions which we call the dual. We define several variants of this process and then demonstrate this for a semi-real ‘toy’ problem of land development in the heart of London where a small set of stakeholder agents have different degrees of interest and control in a small set of land and building sites (parcels). In terms of the model, we show how the problem is already in equilibrium if interest and control are the same and this provides a benchmark for differences between interest and control which characterise the actual problem. We conclude with proposals for making the model more realistic and extending it to deal with problems where conflicts are only partially resolved or not resolved at all.