Characterization of Successful Open-Source Projects
Dr. Dror Feitelson, Hebrew University
This study focuses on the characterization of successful open-source projects. It attempts to do so based on the simplest data available, namely the number of downloads.
The first phase of the study classified projects based on their total number of downloads. It was shown that the distribution of the number of downloads can be divided into three segments. Based on this we classified the projects as superprojects (85 projects with more than 1.1 million downloads each), successful projects (about 10,000 projects with at least 1680 downloads each), and struggling projects (some 30,000 projects with up to 1680 downloads).
In the second phase of the study we refined the classification to consider how the pattern of downloads changed with time. We identified six basic patterns, including increasing downloads, decreasing downloads, peaks in downloads related to releases, and a high download rate during a limited time span. Interpreting these patterns allows us to identify successful projects even if their total downloads are low; for example, this may be the case for projects that cater for a small community of potential users, who nevertheless continue to download each new version.
In future work we plan to automate the classification procedure and try to apply it to all SourceForge projects.
- D. G. Feitelson, G. Heller, and S. R. Schach, "An empirically-based criterion for determining the success of an open-source project". In Australian Software Engineering Conference, pp. 363-368, Apr 2006. Digital Object Identifier: 10.1109/ASWEC.2006.12
- A. Israeli and D. G. Feitelson, "Success of open source projects: patterns of downloads and releases with time". In IEEE Intl. Conf. Software Science, Technology, & Engineering, Oct 2007. Digital Object Identifier: 10.1109/swste.2007.4384088