An empirically-based criterion for determining the success of an open-source project

Abstract

In order to determine a success criterion for open-source software projects, we analyzed 122,205 projects in the SourceForge database. There were 80,597 projects with no downloads at all. We restricted our analysis to the 41,608 projects that together were downloaded 704,897,520 times. Contrary to what we had expected, the distribution of the number of downloads of each project is not Zipf-like; only a portion of the log-log plot of the number of downloads and their rank appears to be a straight line. We performed least-squares analysis (utilizing the Bayesian information criterion) to divide the plot into three segments. On the basis of the shapes of the corresponding curves and the locations of their boundary points, we categorized the projects as follows: 85 superprojects (highly successful projects with more than 1.1 million downloads); just over 10,000 successful projects (with more than 1680 downloads each); and struggling projects (with 1680 downloads or fewer). In terms of our criterion, only a quarter of the projects that have one or more downloads can be deemed to be successful.