Using Social Network Analysis to Inform Management of Open Source Software Development

Abstract

The community-based open source software (OSS) development model has emerged as a viable alternative to firm-based traditional software development. The naturally evolving structure of collaborative relationships among software developers is a major distinction between the OSS development model and the traditional development model. Conventional statistical methods that focus on individual cases and their attributes cannot properly inform the management of the naturally evolving collaborative relationships in open source project. We emphasize social network analysis as a method especially suitable for management of open source development projects, because it focuses on relations among individuals rather than attributes of individual cases. We show how open source development can be represented as a collaboration network graph and how the network can be characterized by various network structure metrics. We present four metrics as a starting point -- size, centralization, density and clusterness, that are most useful in revealing collaborative relationships in OSS development process. We discuss how to generate collaboration network for OSS projects and how to calculate the metrics. We further describe how these metrics can assist in effective management of open source software development process. We conclude by presenting preliminary empirical evidence in support of the metrics.