Where Can Open Collaboration Thrive? A Model of Performance

Abstract

Patterns of collaboration once observed mainly in open source software have mushroomed beyond it, constituting an important part of the informal economy. As the economic impact of, and the interest in, such open collaboration systems of innovation and production have increased dramatically, we contribute to explaining and predicting their performance. Marrying theory and evidence from the innovation literature with recent empirical findings on human cooperation, we develop an agent-based model to examine the performance determinants of open collaboration as a general system for the innovation and provision of products and services, such as Linux and Wikipedia. We define the phenomenal boundaries and identify three performance-related elements: cooperativeness in the population, need heterogeneity of participants and rivalry (subtractability) of the goods. In three studies, we vary those elements while observing the impact on system performance. Results suggest that performance is surprisingly robust, even in seemingly inhospitable environments with free riders, rival goods and homogeneous participant needs. Specifically, because cooperators stimulate contagious behavior, reasonable performance is achievable even when they constitute just a sliver of the population. In most realistic situations, free riders matter little. While low rivalry improves performance, high performance is possible even with rival goods as long as participant needs are heterogeneous. Participants with diverse needs are an advantage, but even homogeneous needs can often be satisfied. The results imply that open collaboration is likely to continue its expansion into new areas of innovation and production. They also resolve some prior puzzles, illuminate the possibility of cooperation between fleeting strangers and offer policy implications.