{"slug": "the-median-open-source-repo-pays-no-ai-review-tax-data-from-5388-repos", "title": "The median open-source repo pays no \"AI review tax\" (data from 5,388 repos)", "summary": "An analysis of 5,388 open-source repositories and 444,225 merged pull requests by The Review Tax Index found that the median repository pays no 'AI review tax,' with 56% showing no detectable agent authorship and a median time-to-merge ratio of 0.5x for attributed work versus the rest of the same repository. Only 29% of comparable repositories pay more review time on attributed work, while 71% pay the same or less, and the median detected AI-attributed share across attributed repositories is 3%.", "body_md": "The Index\n\n# The Review Tax Index\n\nThe claim you will hear is that AI-written code costs more review everywhere. The data says something more useful: the cost is a split — it lands hard on some repositories and not at all on others. This page measures where.\n\nRepositories read\n\n5,388\n\nMerged PRs analyzed\n\n444,225\n\nZero attribution\n\n56%\n\nno detectable agent authorship at all\n\nAgent-native repos\n\n15%\n\nof attributed repos, 20%+ of merged work carries attribution\n\n## The split\n\nThe measures run over the 2,353 repositories with detectable attribution and at least 10 analyzed PRs — a review-cost ratio needs an attributed side to exist. Within each, attributed work is measured against the rest of the same repository — same reviewers, same conventions, same era. A multiple above 1x means the attributed side took more; below means it took less.\n\nTime to merge, attributed / rest\n\n354\n\nrepos above 1x\n\n874\n\nat or below 1x\n\n0.5x\n\nmedian\n\n29% of comparable repos pay more review on attributed work; 71% pay the same or less. Middle half of the population: 0.2x to 1.2x.\n\nReview rounds, attributed / rest\n\n6\n\nrepos above 1x\n\n13\n\nat or below 1x\n\n0.6x\n\nmedian\n\n32% of comparable repos pay more review on attributed work; 68% pay the same or less. Middle half of the population: 0.2x to 1.2x.\n\n## Where the tax lands\n\nMedian detected share across the attributed repositories is 3% — and that is a floor, not a share: attribution comes only from marks a tool leaves on a commit, so inline-completion work is invisible by construction. Across the full population the median is zero — most repositories read carry no attribution at all.\n\n### By primary language\n\n| Segment | Repos | Median detected share | Time to merge | Review rounds |\n|---|---|---|---|---|\n| TypeScript | 442 | 3% | 0.4x | — |\n| Python | 361 | 4% | 0.5x | — |\n| Go | 302 | 3% | 0.5x | — |\n| Rust | 237 | 2% | 0.7x | — |\n| JavaScript | 217 | 2% | 0.4x | — |\n| C++ | 170 | 3% | 0.4x | — |\n| Java | 131 | 2% | 0.3x | — |\n| C# | 129 | 6% | 0.3x | — |\n| C | 98 | 2% | 0.4x | — |\n| PHP | 87 | 2% | 0.3x | — |\n| Swift | 68 | 3% | 0.5x | — |\n| Kotlin | 53 | 2% | 0.4x | — |\n| Ruby | 49 | 3% | 0.5x | — |\n\nA dash means the segment has too few comparable repos for that measure to publish — a median over a handful of ratios is noise wearing a number.\n\n### By project size\n\n| Segment | Repos | Median detected share | Time to merge | Review rounds |\n|---|---|---|---|---|\n| 50k+ stars | 238 | 4% | 0.4x | — |\n| 20k-50k stars | 624 | 3% | 0.5x | — |\n| 10k-20k stars | 688 | 3% | 0.4x | — |\n| 5k-10k stars | 775 | 3% | 0.5x | — |\n| 2k-5k stars | 26 | 1% | — | — |\n\nA dash means the segment has too few comparable repos for that measure to publish — a median over a handful of ratios is noise wearing a number.\n\n## How to read this\n\n- Attributed vs rest, never AI vs human. The comparison group contains AI-assisted work that leaves no trace, which narrows every gap shown here rather than widening it.\n- The sample self-selects twice. These are repositories someone chose to scan plus a curated backfill of well-known projects — a population of repos people care about, not a random sample of software.\n- Withheld beats invented. Repos below 10 analyzed PRs do not count; segments below 12 comparable repos publish no median.\n\nThe Index republishes quarterly. Repos already read keep their history, so future editions carry trend lines — not just where the tax lands, but where it is moving.\n\nFull method and every limitation: [/methodology](/methodology). Place your own repo against this population: [/scan](/scan).", "url": "https://wpnews.pro/news/the-median-open-source-repo-pays-no-ai-review-tax-data-from-5388-repos", "canonical_source": "https://forge.ambera.app/review-tax-index", "published_at": "2026-08-13 10:44:44+00:00", "updated_at": "2026-08-13 11:11:16.058129+00:00", "lang": "en", "topics": ["ai-tools", "developer-tools"], "entities": ["The Review Tax Index"], "alternates": {"html": "https://wpnews.pro/news/the-median-open-source-repo-pays-no-ai-review-tax-data-from-5388-repos", "markdown": "https://wpnews.pro/news/the-median-open-source-repo-pays-no-ai-review-tax-data-from-5388-repos.md", "text": "https://wpnews.pro/news/the-median-open-source-repo-pays-no-ai-review-tax-data-from-5388-repos.txt", "jsonld": "https://wpnews.pro/news/the-median-open-source-repo-pays-no-ai-review-tax-data-from-5388-repos.jsonld"}}