PR spam today looks like email spam in the early 2000s
Points and comments are a snapshot, not live.
AI-generated pull request spam on OpenClaw mirrors early 2000s email spam patterns.
OpenClaw grew from 2 PRs/week in December to 3,400/week in February, with merge rates dropping from 48% to 9.3%. Median time between one contributor's 106 PRs was three seconds.
The author identifies three trends: PRs will need sender reputation (like Vouch), AI agents reduce diversity of thought, and refactors merge at 35% vs 9% for features. The piece argues that deep codebase understanding, not novel feature generation, drives accepted PRs.
What commenters are saying
Commenters split into two camps: those who see AI-generated PRs as selfish reputation-building (Goodhart's law applied to open source contributions), and those who believe many contributors genuinely want to help but lack skill. Several maintainers advocate for banning AI contributions entirely or using trust systems like Vouch. One commenter notes that job interviews rarely assess actual ability anyway, so the metric of open source contributions was already flawed before AI slop.