C001 — CI Adoption Rate: 40% Figure Is Outdated — Assessment¶
Contents¶
The 40% CI adoption figure is likely correct as a historical data point from the seminal 2016 Hilton et al. study, but is very likely outdated as a description of current adoption. GitHub Actions alone matched the entire 40% rate by January 2022. The figure is also sensitive to population definition — active projects show much higher CI adoption than the full repository population including abandoned and hobby projects. The researcher should not present 40% as a current figure.
Evidence Synthesis¶
Evidence quality: Robust — Multiple peer-reviewed academic studies (Hilton et al. 2016, Golzadeh et al. 2022, IST 2026) with large-scale empirical data. JetBrains survey provides complementary industry perspective.
Source agreement: High — All sources agree that CI adoption has grown substantially since 2016. The original 40% figure is confirmed by the primary source but is understood to be outdated. Current adoption rates are higher, with GitHub Actions alone at ~44% of sampled repos.
Independence: Sources are largely independent — different research groups at different universities using different methodologies arrive at consistent findings about CI growth.
Probability Assessment¶
- H1: Likely (55-80%)
- The 40% figure was accurately reported in the original 2016 study of 34,544 GitHub projects. However, the 'approximately' qualifier and the age of the study introduce uncertainty about its current applicability. As a historical data point, the claim is correct; as a description of current state, it is outdated.
- H2: Very likely (80-95%)
- GitHub Actions alone achieved 43.9% adoption in 68K repos by January 2022. JetBrains reports 55% developer-level CI/CD usage in 2025. The CI landscape has shifted dramatically since 2016. Total CI adoption among the same population measured by Hilton et al. is almost certainly above 40% today.
- H3: Very likely (80-95%)
- The original study itself notes CI is 'widely adopted by the most popular projects.' The Golzadeh et al. study explicitly focuses on 'active npm packages' — a subset that shows much higher CI adoption. Activity level is a critical stratification variable that the 40% figure does not capture.
- H4: Unlikely (20-45%)
- No direct evidence was found measuring the rate of non-functional CI configurations. Build failure studies exist but address failures within active CI, not CI that has been configured and abandoned. Verdict: The claim that 'only approximately 40% of open source projects use CI at all' is likely correct as a historical fact from the seminal 2016 Hilton et al. study (55-80%), but is very likely outdated as a description of current adoption (80-95% probability that current rates are substantially higher). The figure is also heavily dependent on how 'open source project' is defined — active projects show much higher CI adoption than the full population of all public repositories.
Evidence Gaps¶
Expected but not found: - No post-2023 comprehensive study measuring overall CI adoption (all tools combined) across a representative sample of all GitHub repositories. - No study distinguishing CI configuration file presence from active CI pipeline execution.
Unanswered questions: - What is the current (2025-2026) overall CI adoption rate across all GitHub repositories? - How does CI adoption rate differ between the full population of public repos and the population of 'active' repos (recent commits, multiple contributors)?
Impact on confidence: The gaps primarily affect our ability to state a precise current CI adoption rate. However, the directional finding — that CI adoption is substantially higher than 40% today — is well-supported by multiple independent studies. The researcher should avoid presenting 40% as a current figure.