Loris Cro’s latest post, “Contributor Poker and Zig’s AI Ban”, lays out a view of open source that centers less on individual pull requests and more on the long-term relationship between maintainers and contributors. Cro argues that early PRs should be treated as a kind of bet on a person’s future value, rather than judged only on first-pass code quality.
He describes this as “contributor poker,” a dynamic the Zig uses to rely on as it works on a large compiler toolchain with help from outside contributors. According to Cro, that approach has already paid off in concrete ways, including support from contributors such as Ryan Liptak and Frank Denis on parts of Zig that would have been difficult to cover internally.
The post also points to the downside of growth. Cro mentions that the volume of incoming PRs has now outpaced the time available to core reviewers, creating delays that can leave promising work unreviewed for too long. He notes that Zig has already acknowledged this pressure in its financial reports and is still looking for ways to reduce the strain.
From there, Cro turns to Zig’s ban on AI-generated contributions and explains why the policy fits the project’s current review model. He claims that most LLM-based PRs have brought “background noise,” hallucinations, and oversized first-time submissions, while also making it harder to sustain the trust that contributor poker depends on.
Cro does not present the policy as final, though. He indicates that Zig may adjust it as the project gains more insight, and he closes by inviting support for the Zig Software Foundation. The full post is worth a read for anyone interested in how open source projects balance quality, reviewer bandwidth, and contributor trust.
Source: Loris Cro
