Selectionism and Learning in Complex and Ambiguous Projects

Abstract

Project management literature has increasingly recognized that established project management methods work well for projects with moderate complexity and uncertainty, but have limitations in projects with ambiguity (the inability to recognize the relevant influence variables and their functional relationships; thus, events and actions cannot be planned ahead of time) and high complexity (large number of variables and interactions; this leads to a difficulty of assessing optimal actions beforehand). There are two fundamental strategies to manage projects with ambiguity and complexity: trial & error learning and selectionism. Trial & error learning involves a flexible adjustment of the project approach to new information about the relevant environment, as it emerges. Selectionism involves pursuing several approaches independently of one another and picking the best one ex post. Previous work has combined the two approaches under uncertainty but not under the combined influence of project ambiguity and complexity. We build a model of a complex project with ambiguity, simulating problem-solving as a local search on a rugged landscape. We compare the project payoff performance under trial & error learning and selectionism, based on a priori identifiable project characteristics: whether ambiguity is present, how high the complexity is, and how much trial & error learning and parallel trials cost. We find that if ambiguity is present and the team cannot run trials in a realistic user environment (reflecting the project's true market performance), trial & error learning becomes more attractive relative to selectionism as the project's complexity increases. Moreover, the presence of ambiguity may reverse an established result from computational optimization: without ambiguity, the optimal number of parallel trials increases in complexity. But with ambiguity, the optimal number of trials may decrease because the ambiguous factors make the trials less and less informative as complexity grows.