Thanks for opening this discussion Nic! 3 seems like a good number to me. A thing I've noticed is that an issue opened by a regular contributor with the intent of immediately working on it will sometimes get a PR opened against it within an hour or two. The newly opened PR will typically come from a new contributor and is likely AI-generated. This will likely be resolved with the limited number of PRs.
Rok On Thu, Aug 13, 2026 at 4:25 AM Jeffrey Vo <[email protected]> wrote: > For reference, I've seen polars has a similar policy, though theirs is only > for first time contributors: > > You may not have more than one open PR at a time. > https://docs.pola.rs/development/contributing/#first-time-contributions > > Cheers, Jeffrey > > On Tue, 11 Aug 2026 at 18:48, Nic Crane <[email protected]> wrote: > > > After a brief reprieve, we've had an uptick in AI contributions, some of > > which are causing issues like people not responding to feedback and > leaving > > open PRs which get stale but appear that someone is working on them, > > preventing other people from taking them up. > > > > There is currently a PR [1] open on the ASF infrastructure repo to > enabled > > a GitHub setting [2] whereby contributors without write access can only > > have a limited number of PRs open at the same time. > > > > I think it'd be worth pushing for that to get merged, and in the meantime > > agreeing on our own policy in terms of the number of concurrent PRs. > > > > We need to strike a balance between preventing excess low-value PRs but > > allowing non-committers to continue to be productive. I did a quick > > analysis, and non-committers have a mean of 1.43 and median of 1 > concurrent > > open PRs, with the mean being skewed by a long tail with some folks with > > lots of open PRs, so a fairly low number like 3 could work quite well. > > > > What are peoples thoughts? > > > > Nic > > > > > > [1] https://github.com/apache/infrastructure-asfyaml/pull/111 > > [2] https://github.com/community/maintainers/discussions/840 > > >
