Turnover in Open-Source Projects: The Case of Core Developers

• Fabio Ferreira IF Sudeste MG
• Luciana Lourdes Silva IFMG
• Marco Tulio Valente UFMG

Resumo

In order to know whether FLOSS projects are performing well, researchers have studied the ability of these projects to both attract and retain contributors, and consequently, their turnover, that is, the exit of some contributors and the entry of others. However, most contributors accounted for turnover in FLOSS projects are not active and make one or two contributions per year, which can lead to inaccurate results. In this paper, we compute the turnover of 174 FLOSS projects for each year between 2015 and 2018, considering only core developers. We analyze how the Core Developers' Newcomers and Leavers rates can influence the actionability of these projects, considering the time to fix issues and bugs, and the time to implement enhancements. We found out that 104 (59.7%) out of 174 projects have at least 30% turnover per year, 46 (26.4%) projects exceed 50%, and only 10 (5.7%) projects have less than 10% of annual turnover on average. We also found that projects owned by organizations have a higher Core Developer Turnover (CDT) rate than projects owned by individual users. The results show that the turnover of core developers increases with the size of teams, and it is notably higher in Ruby projects. Finally, we use the Core Developer Newcomers (CDNRate) and Leavers (CDLRate) rates to classify FLOSS projects as Attractive, Unattractive, Stable, Unstable. The results show that projects classified as Unstable (High Number of Core Developers Leavers and Newcomers) take a longer time to fix issues and bugs and to implement enhancements than other groups.

Palavras-chave: GitHub, Turnover, FLOSS, Core Developers, Free/Libre and Open Source Software