# When Patches Cost Nothing the Described Problem Is the Contribution ## Claim When an agent can generate a patch for almost no effort, a pull request stops proving that someone worked on the problem, so the part of a contribution that still costs effort is describing the problem well. <!-- atomicity: 1 idea ✓ --> ## Scope & conditions - Holds for projects where maintainers can run their own agents against a well-described issue and review the result with more confidence than a drive-by patch. - Fails where a fix needs domain knowledge only the contributor has, or where low-effort issues pile up as fast as low-effort PRs did. The hosts raise that second case themselves ("agents talking to each other") and answer it with discussion first, then implementation. - The evidence is host commentary on maintainers' announcements. The hosts relay that four named maintainers closed or plan to close external PRs and add two anecdotes of their own. That supports "some maintainers have turned external PRs off," not that pull requests are dead, and the transcript gives no data on issue quality. - This note states the mechanism. It doesn't recommend turning PRs off. ## Evidence One host's premise for why a PR used to matter: > a pull request was golden because somebody would put some effort into solving that issue, figuring out the issue and then sending you the code. ## Links - [[3. Resources/MOCs/Open Source Governance MOC]] — a contribution-policy shift: the unit maintainers accept moves from code to discussion. - [[3. Resources/Permanent Notes/Cheap Generation Raises the Total Maintenance Bill]] — because the maintainer pays the review cost of every cheap patch, which is why the intake point moves. - [[3. Resources/Permanent Notes/Code Review Is a Human Knowledge-Transfer Ritual Not Just a Quality Gate]] — contrasts: a PR also carried knowledge transfer, which an issue-first flow has to rebuild in the discussion thread. - [[3. Resources/Permanent Notes/Human Review Capacity Sets the Agent WIP Limit]] — extends: the same review ceiling, applied to inbound contributions instead of internal agent loops. - [[3. Resources/Readwise/Articles/The Pi Updates Are Insane ⟡ Pull Requests Are Dead ⟡ Turso Joining Supabase ⟌ Syntax Weekly ⟌]] — source episode. ## Tests / falsifiers - Compare merged-fix quality in projects that moved to issue-first intake against similar projects that kept open PRs. If the issue-first group sees more duplicate or empty issues, the claim's scope shrinks.