Will You Come Back to Contribute? Investigating the Inactivity of OSS Core Developers in GitHub

TL;DR

This study introduces a novel method to detect inactivity periods of core OSS developers on GitHub, revealing patterns of breaks and transitions that can inform community management strategies.

Contribution

The paper presents a new approach to identify developer inactivity and provides empirical analysis of core developers' break patterns and transition probabilities.

Findings

94% of surveyed developers agreed with the inactivity model

Approximately 45% of core developers took at least one year-long break

Breaks longer than a year reduce the likelihood of returning to activity to around 21-26%

Abstract

Several Open Source Software (OSS) projects depend on the continuity of their development communities to remain sustainable. Understanding how developers become inactive or why they take breaks can help communities prevent abandonment and incentivize developers to come back. In this paper, we propose a novel method to identify developers' inactive periods by analyzing the individual rhythm of contributions to the projects. Using this method, we quantitatively analyze the inactivity of core developers in 18 OSS organizations hosted on GitHub. We also survey core developers to receive their feedback about the identified breaks and transitions. Our results show that our method was effective for identifying developers' breaks. About 94% of the surveyed core developers agreed with our state model of inactivity; 71% and 79% of them acknowledged their breaks and state transition, respectively. We also show that all core developers take breaks (at least once) and about a half of them (~45%}) have completely disengaged from a project for at least one year. We also analyzed the probability of transitions to/from inactivity and found that developers who pause their activity have a ~35-55\% chance to return to an active state; yet, if the break lasts for a year or longer, then the probability of resuming activities drops to ~21-26%, with a ~54% chance of complete disengagement. These results may support the creation of policies and mechanisms to make OSS community managers aware of breaks and potential project abandonment.