C001 — OSS CI Adoption Rate Around 40% — Assessment¶
Contents¶
The claim that approximately 40% of open source projects use CI is well-supported by multiple independent empirical studies. The exact figure varies by study and sample composition (40% in 2016, 43.9% in 2022, 50%+ in npm-specific repos), but the approximate 40% figure serves as a reasonable baseline. The claim masks significant stratification: CI adoption is much higher among actively maintained, popular projects.
Evidence Synthesis¶
Evidence quality: Robust — Two peer-reviewed academic studies (Hilton et al. 2016, Decan et al. 2022 via the Mons/Radboud paper) provide independent empirical measurements of CI adoption across large GitHub samples (34,544 and 68,000 repositories respectively). A third study (Golzadeh et al. 2021) adds npm ecosystem data. The JetBrains survey provides developer-level adoption data from a different methodology.
Source agreement: High — All sources converge on a CI adoption rate in the 40-50% range: Hilton et al. found 40%, Decan et al. found 43.9% for GHA specifically, and Golzadeh et al. found 50%+ for npm-related repos. The JetBrains survey (62% personal, 41% organizational for GHA) is consistent when accounting for self-selection bias. No source contradicts the approximate 40% figure.
Independence: Largely independent. Hilton et al. (2016), Decan et al. (2022), and Golzadeh et al. (2021) used different sampling strategies, time periods, and CI measurement approaches. The Mons/Radboud paper cites but does not replicate Hilton's methodology. The JetBrains survey uses a completely different approach (developer survey vs. repository mining).
Outliers¶
- https://blog.jetbrains.com/teamcity/2025/10/the-state-of-cicd/: 62% GHA usage for personal projects appears higher than the repository-mining studies — This is a self-selected developer survey (805 respondents), not a repository-mining study. Developers who respond to CI surveys are more likely to use CI tools, creating upward bias. The 41% organizational figure aligns better with the mining studies.
Probability Assessment¶
- C001-H1: Very likely (80-95%)
- The 40% figure is directly confirmed by Hilton et al. (2016) and closely corroborated by Decan et al. (43.9% in 2022). While the exact percentage varies by study, the claim that 'approximately 40%' of OSS projects use CI is well-supported by multiple independent empirical studies.
- C001-H2: Roughly even chance (45-55%)
- Evidence supports the direction — Golzadeh et al. found 50%+ in npm repos, Decan et al. found 43.9% which is already above 40%. However, these studies still hover around 40-50%, not 'substantially higher.' The claim of outdatedness is partially supported but the figure has not moved dramatically.
- C001-H3: Very likely (80-95%)
- The Mons/Radboud paper confirms that 'CI/CD is widely adopted by the most popular projects' and that adoption correlates with project activity. The Chainguard Scorecard study shows a one-point score increase per 100x increase in GitHub stars. Stratification by project maturity is consistently reported across studies.
- C001-H4: Unlikely (20-45%)
- No direct evidence was found measuring the gap between CI configuration presence and functional CI execution. The Mons/Radboud paper reports 7.3% weekly modification of workflow files, suggesting active maintenance rather than rot. However, this hypothesis cannot be definitively eliminated due to the absence of studies specifically measuring CI configuration rot. Verdict: The claim that approximately 40% of open source projects use CI is Likely to Very likely (75-90%). The 40% figure is empirically grounded, but evidence suggests it may be conservative for actively maintained projects and slightly outdated given GitHub Actions adoption growth.
Evidence Gaps¶
Expected but not found: - No study measuring CI adoption across all GitHub repositories with a publication date of 2024 or 2025 was found. - No study distinguishing between 'has CI config file' and 'has functional, regularly-executing CI pipeline' was found. - No GitHub Octoverse data specifically reporting CI adoption rates was discovered despite searching for it.
Unanswered questions: - What is the current (2025) CI adoption rate when measured by actual build execution rather than config file presence? - Has GitHub Actions' zero-friction integration meaningfully increased overall CI adoption since 2022?
Impact on confidence: The absence of very recent (2024-2025) primary data slightly reduces confidence, but the convergence of multiple independent studies on similar figures provides a strong foundation. The biggest gap — the distinction between nominal and functional CI adoption — could materially affect the assessment if the gap is large.