Open Source Software Development and the Small World Phenomenon: An Empirical Investigations of Macro Level Colaboration Network Properties on Project Success

In this study, I investigate whether small world properties of developer collaboration networks influence developer productivity. Small world networks are characterized by two macro level network properties: dense clustering and short average path length. Structure of relationships among developers affects how information and knowledge is shared among them. I argue that networks characterized by both high clustering and short average path length would positively influence the productivity of the developers by providing them with speedy and reliable access to more and more varied information. Using hierarchical linear modeling, I analyze extensive data collected from 2013 open source projects and 8512 developers belonging to 15 different networks hosted at Sourceforge.net. I find evidence that ‘small world’ properties of these networks positively affect their members’ productivity. The findings are robust to a number of controls and model specifications.