Stars don't review pull requests
The state of open-source black holes: how often a stranger's pull request goes unanswered, in the Porchlamp snapshot of 14 October 2026.
You send a small fix to a project you use every day. Two weeks later, nobody has replied. At more than a quarter of active open-source projects, that is the normal experience.
Key findings
More stars, more silence. The share rises with stars: 21% of projects with 1,000 to 10,000 stars, 31% at 10,000 to 50,000, and 39% at 50,000 or more. #
Fame is no guarantee. 25 of the 100 most-starred projects are black holes, and only 9 of them reach Porchlamp's top 100. #
Stars barely predict an answer. Across 3,833 ranked projects, the rank correlation between stars and the Porchlamp score is 0.13. #
Most of those pull requests waited. Inside the black-hole projects, 71,189 of 91,209 counted outside pull requests got no human reply within 14 days. #
Do popular projects answer fewer pull requests?
They answer less often, on average. The black-hole share climbs at every step up in stars.
The numbers behind the chart
| Stars | Projects judged | Black holes | Share |
|---|---|---|---|
| 1,000 to 10,000 stars | 2,190 | 467 | 21% |
| 10,000 to 50,000 stars | 1,862 | 572 | 31% |
| 50,000 or more stars | 244 | 94 | 39% |
Not charted: 1 judged project with fewer than 1,000 stars.
One likely reason: review time stays roughly fixed while attention grows, so the queue grows faster than volunteers can clear it. The data shows the rise, not its cause.
Which package ecosystems answer outside pull requests?
PyPI and npm projects go quiet more often than Rust, Go, Java or systems projects. Each share is computed within its own ecosystem.
The numbers behind the chart
| Ecosystem | Projects judged | Black holes | Share |
|---|---|---|---|
| PyPI | 968 | 327 | 33.8% |
| npm | 1,283 | 385 | 30.0% |
| Go | 609 | 135 | 22.2% |
| systems (C, C++ and Zig) | 573 | 115 | 20.1% |
| crates.io (Rust) | 586 | 117 | 20.0% |
| Maven (Java) | 278 | 54 | 19.4% |
Size is not controlled for. Bigger projects go quiet more often in general, so part of a gap can come from size.
How we measured
- We followed 212,390 pull requests from 110,605 outside contributors to 11,000 GitHub projects, from 16 April to 13 October 2026.
- A project is judged only when it is licensed, maintained software with at least 10 counted outside pull requests.
- It is a black hole when, with 95% confidence, more than half of those pull requests got no substantive reply, review or merge within 14 days.
- Bots, drafts, pull requests their authors closed, brand-new accounts and the project's own members are not counted.
- Every parameter, gate and weight: Porchlamp methodology.
What this is not
- Not a verdict on the people who maintain these projects. Review time is scarce, and popular projects receive far more pull requests than anyone can read.
- GitHub only. Projects that review on Gerrit or a mailing list are left undecided, not failed.
- One six-month window. The numbers will change at the next snapshot.
- This page names no project. Each project's record, with the pull requests behind it, is on the Porchlamp rankings.
Download the data
porchlamp-2026-10-14-ranked.csv: the 3,833 ranked projects with rank, owner, name, ecosystem, score, tier, coverage, quadrant and the rank in each lens. Released under CC BY 4.0.
The full records, including every project that was not ranked and why, are in the site's JSON data.
Cite this report
Porchlamp (2026). Stars don't review pull requests: the state of open-source black holes. Snapshot 2026-10-14. https://porchlamp.org/black-holes/2026-10-14/
Data licence: CC BY 4.0. Snapshot database SHA-256: bf96e162537333e43e98a59a472755dfee9b20e50da6751c1a1816321e5c237c. Methodology constants hash: 6e4f3e8e1d0c0fcd829d9c8c386a65179eadb5f936595f96e9eba98712d55b94.