The desire to benefit from voluntary external support is one reason why individuals or firms would engage in open source software (OSS) development, notably by founding their own custom-tailored OSS projects. We study how contributors to OSS communities self-select into projects, and how this affects project performance.

Bridging the theory from human resources and innovation management, we suggest that extrinsic incentives should attract higher-skilled workers more the better such incentives allow contributors to leverage their abilities when attaining their personal goals and minimizing related opportunity costs. For our tests, we draw on an originally compiled and custom-tailored dataset merging information from two major software community archives—Stack Overflow and GitHub. Our data contains information on both contributors’ programming activity—as measured in their pull requests with proposed code sent to other repositories—and contributor skills—as measured by the expert answers they provide on Stack Overflow. Our results, obtained on almost 50,000 contributor-project-month observations, are threefold. First, we show that community contributors select projects that match their skills when the projects provide incentives to gain recognition from potential employers or peers. Notably, more visible projects (as measured by the number of affiliated members or number of received downloads) lead to the attraction of higher-skilled contributors, all else being equal. Second, coders also sort on skill when community projects are managed so that contributors get a chance to document their individual skills. This is the case when coders receive fast feedback on their pull requests, and when they may expect their actions to get visible traction—notably when contribution acceptance (= pull request merging) rates at the repository level are high. This latter finding suggests that sending status signals in a traditional sense would matter less to skilled contributors than seeing their work disseminate.

Our findings bear several contributions for management scholars. Aside from corroborating earlier empirical work relying on stated preference data, we can show that contributors consciously sort on skill when engaging in OSS communities. Most importantly, we suggest that community designers may leverage this effect to attract top engineering talents to their projects and potentially approach them for hire. Notably, by ensuring both skill-based sorting and incentive alignment, using a community to screen for talent may appear superior to traditional approaches of setting steep financial incentives for performance only.