Provides a longitudinal analysis of library abandonment in the Maven ecosystem, identifying release speed slowdowns as a key leading indicator of failure.
A 10-year longitudinal study of 403,048 Maven libraries revealed consistent annual abandonment rates ranging from 20.9% to 24.4%. Approximately one in four libraries fail to survive their first year, and the two-year cumulative abandonment rate approaches 50%. While 59.1% of abandoned libraries are short-lived (<1 year), 22.8% are long-lived, indicating that even mature digital infrastructure faces substantial sustainability risk. The research identifies 'slow release speed' as a primary precursor to abandonment—exhibited by 20.3% of abandoned libraries compared to 8.4% of active ones—but warns that 21.3% of libraries show a burst of high release activity late in their lifecycle before failing, making recent activity an unreliable standalone signal of long-term survival.
The study confirms that 15-24% of Maven libraries are abandoned annually, with 50% failing within two years of creation.
dependency on digital infrastructure that is no longer being actively maintained or secured
identifying abandoned repositories is critical for managing the high 'infant mortality' rate (25% in year one) identified in the Maven ecosystem.
Slow release speed and long periods of inactivity (2+ years) were identified as the most common precursors to formal abandonment.
reliance on decaying libraries that have lost their maintainer momentum
release staleness is an empirically validated proxy for the final stages of the library abandonment lifecycle.
Abandoned libraries exhibit a slowdown in their release speed prior to abandonment... 21.3% abandoned libraries exhibit high release speeds... high release speed alone may not reliably signal the risk.
Risk Guard uses binary staleness thresholds but does not analyze 'Velocity Trends' (e.g., a drop from high-speed to low-speed) as a leading indicator.
Risk Guard would be better if it calculated a 'Maintenance Velocity Momentum' score to detect projects that are trending toward abandonment before they reach a binary staleness threshold.