On the relation between Github communication activity and merge conflicts

Version control systems assist developers in managing concurrent changes to a common code base by tracking all code contributions over time. A notorious problem is that, when integrating code contributions, merge conflicts may occur and resolving them is a time-consuming and error-prone task. There is a popular belief that communication and collaboration success are mutually dependent. So, it is believed that great communication activity helps to avoid merge conflicts. However, in practice, the role of communication activity for merge conflicts to occur or to be avoided has not been thoroughly investigated. To better understand this relation, we analyzed the history of 30 popular open-source projects involving 19 thousand merge scenarios. Methodologically, we used a bivariate (Spearman’s rank correlation) and a multivariate (principal component analysis and partial correlations) analysis to quantify their correlation. In bivariate analysis, we found a weak positive correlation between GitHub communication activity and the number of merge conflicts. However, in the multivariate analysis, the positive correlation disappeared, not supporting the intuition that GitHub communication helps to avoid merge conflicts. Interestingly, we found that the strength of this relationship depends on the merge scenarios’ characteristics, such as the number of lines of code changed. Puzzled by these unexpected results, we investigated each covariate, which provided justifications for our findings. The main conclusion from our study is that GitHub communication activity itself does not support the emergence or avoidance of merge conflicts even though such communication is associated only with merge scenario code changes or among developers only.