An Empirical Investigation of Code Contribution, Communication Participation and Release Strategies in Open Source Software Development: A Conditional Hazard Model Approach

This study uses data from over 200 open source software (OSS) projects hosted at Sourceforge to empirically examine how OSS development characteristics affect project progress. Consistent with prior studies, we find that core developers (i.e., top 20 % of the most contributing developers) develop about 80% of the code. We also find that a group of developers that is about 50% larger than the core group is responsible for 80% of the communication. On average, the top 30% developers contribute about 81% of the messages. It is observed that projects with a clearly identified core group of developers conducting most of the coding are more efficient for project progress. We also find that highly participative peripheral group in communication positively affects project progress. In addition, high level of communication interactivity, measured by the number of email threads and average thread depth, enables a project to progress well. Our results also show that projects that draw upon the larger OSS community for beta testing progress faster. We find that the projects that do not progress well are associated with the following major factors: (i) lack of a core developer group, (ii) dormant peripheral group in communication, (iii) low communication interactivity exemplified by the lack of depth in threaded communications, and (iv) over-dependence on internal community for beta testing.