The open-source movement provides a way to observe software engineering in the wild. This paper studies 352 SourceForge projects to see if their organization and process can be correlated with the speed with which features are added and bugs removed. The paper uses social science techniques to propose and test four hypotheses. Three seem to have significant evidence supporting them. The data also shows that sponsorship has a positive relation to maintenance speed. The paper gives an excellent summary of the state of the programming art, with copious references. The authors note that most software work is maintenance and that open-source projects have a structure that accommodates both registered developers and a larger number of lower-status contributors. The authors distinguish the internal from the external developers, and correlate the proportion of internal participants with the rate at which maintenance is done. There is a significant negative relation with changes that enhance the software, but no significant association with correcting errors in the code.

The authors also examined the way the project determines who does what. As the authors expected, enhancements are done faster when people can volunteer to do them, but corrections are done faster when they are delegated to people. The paper is dense with jargon and multisyllabic words. It equates correlation with causality. It does not mention null hypotheses. I am not sure that the results are reliable, but it still helps to fill a gap in our understanding of software development processes.