Reverse Engineering the Bazaar: Collaboration and Communication in Open Source Development

Open source project data has been a bonanza for software engineering research, much like Affymetrix´s chips and gene sequencers have been for bio-informaticians. We are swimming in an ocean of data, merrily analyzing, clustering, correlating and model-fitting. However, while researchers have been busily mining the wealth of information in open source code repositories to study phenomena such as co-change patterns, evolution, defect occurrences and so on, with a few exceptions, the human side has largely remained unexplored. How do people collaborate? Who do they talk with? What do they talk about? How much do they talk? How do new people join? How do existing project members associate? These are questions are of vital importance to the organization of any software project; successful open source projects have a lot to teach us about such phenomena. Fortunately, there is a wealth of information in mailing list archives, bug report databases, and even code repositories that can shed light upon these vital issues. Indeed, because of their distributed, public nature, there is more information readily available about human interaction in open source development than in traditional organizations. In this talk, the author will discuss some recent research in the area, both at UC Davis and by other colleagues.