New competence model uses code metrics to predict OSS contributor skills
Core contributors are leaving OSS projects—code metrics could find their replacements.
A new paper from Sabahat Younas, Márcia Moraes, and Fabio Santos tackles a growing crisis in open source software (OSS): the disengagement of core contributors, which increasingly threatens project survival. The research, submitted to arXiv and accepted for the ACM/IEEE International Symposium on Empirical Software Engineering and Measurement (ESEM 2026), argues that finding replacements is slow and unreliable because existing methods for evaluating candidates rely on interviews or exams that lack precision.
To close this gap, the authors propose a competence model that automatically assesses contributors' abilities by collecting source code metrics from their contributions. This allows the model to predict the competence level required to solve specific tasks, enabling maintainers to identify and train promising candidates before core contributors leave. The paper includes 14 pages, one figure, and a public replication package with code, data, and documentation, giving other researchers a concrete foundation to build on.
- Uses source code metrics to assess contributor competence levels, moving beyond interviews and exams.
- Predicts the competence required for specific OSS tasks, aiding replacement planning and training.
- Replication package with code, data, and docs available; paper appears at ESEM 2026.
Why It Matters
This model could prevent OSS project collapse by systematically identifying and training successors before core contributors depart.