Study: 52.5% of dormant open-source projects have unknown cause of inactivity

A new empirical study from researchers Addi Malviya Thakur, Bogdan Vasilescu, and Audris Mockus explores the phenomenon of dormancy and revival in scientific open-source software. Analyzing 18,247 repositories from the SciCat corpus, they identified 2,984 candidates that exhibited periods of inactivity followed by activity. Through a rigorous coding process involving 75 analysts on a stratified sample of 750 projects, the team classified five dimensions: dormancy cause, revival mechanism, nature of revival work, revival sustainability, and lifecycle archetype. A striking finding: simply changing the inactivity threshold from 1 month to 36 months shifts the count of 'abandoned' projects from 18,030 to 8,010, showing these thresholds are unreliable. Among dormant projects, 52.5% had no clear cause of inactivity from repository evidence alone. When causes were identifiable, feature or milestone freezes outnumbered research-output completion by a ratio of 5.4:1, suggesting many projects pause for planning rather than permanent abandonment.

The study also reveals that revival is often fragile. Non-sustained recovery (where projects become dormant again after revival) outnumbers sustained recovery by 2.14 to 1, and 11.5% of apparent revivals are actually bot-only or single-spike artifacts, not genuine human activity. The researchers developed a rule-based classifier to map each project's lifecycle archetype, finding that archetype is more strongly associated with sustainability than the revival mechanism or nature of work. Their conclusion: gap duration, lifecycle archetype, and contributor continuity together provide far more discriminating information for classifying abandonment than any single inactivity threshold. For the open-source community, this means that assuming a project is dead based on a fixed time window is flawed; instead, evaluators should consider the project's history of pauses, the roles of contributors, and the underlying reasons for dormancy.