Evolution of developer social network and its impact on bug fixing process

Abstract

Universally accessible and publically archived nature of Bug Tracking System (BTS) of Open Source Software enables developers to follow the work of each other and contribute in bug fixing. The interaction of developers through comments on BTS of project leads to form a social network. The developers and their relationships change over the time resulting in evolution of Developers' social network (DSN). Prior studies (Hong et.al) have compared the evolution of DSN with evolution of general social networks like facebook, twitter etc., showing their resemblance and some differences with them. However these studies don't provide any insight how the evolution of DSN correlate with the effectiveness of bug fixing process over the time. Such insight is helpful as managers can reorganize the teams and issue the guidelines to the developers, accordingly, forcing the communication structure which results in to more effective bug fixing process. In this paper, we first study the evolution of DSN of Eclipse a java based IDE, partially replicating and enhancing the study done by Hong et. al. Then we show how the global social network properties of the DSN e.g. Average Path Length, Clustering Coefficient, modularity etc. has an impact on attributes characterizing effectiveness of bug fixing process like average fix time of the bugs, percentage of bugs fixed etc. We found good correlation between global social network properties and attributes characterizing the bug fixing process.