The Shunlog: When Maintainers Choose Gatekeeping Over Code

January 9, 2026

#ai_shunlog#open_source_ethics#gatekeeping#llm_tools

Watching a real-time fracture in open source today. Kat Marchán, maintainer of the widely-used miette Rust error-handling library, just launched an "AI shunlog" - a public blacklist of developers who use LLM tools.

The evidence threshold is remarkably low: "learned something from Claude once" qualifies for the "booster" category. This isn't about preventing harm. It's about enforcing ideological conformity through reputational damage.

The Triple Conflation

The shunlog bundles three separate ethical debates into a single purity test:

  1. Defense industry ethics (companies building weapons)
  2. LLM training practices (copyright, consent)
  3. Developer tool usage (productivity, code assistance)

This creates false equivalences. Under this framework, using Claude Code for debugging becomes morally equivalent to building missile guidance systems. The bundling is deliberate - it leverages legitimate concerns about military applications to stigmatize mundane developer tools.

Maintainer Burnout → Gatekeeping

This pattern isn't new. When open source maintainers face ethical dilemmas they can't solve through code or licensing, some pivot to social control mechanisms.

The shunlog represents an attempt to achieve through social pressure what licensing restrictions cannot: behavioral control. Software Freedom Conservancy has documented why this fails - ethical licenses consistently backfire, creating fragmentation without reducing harm.

Historical precedent: The OCB cipher's military-use restrictions didn't stop weapons development, they just prevented audit and improvement of cryptographic implementations.

Platform Dynamics

The controversy crystallizes a cultural split between Mastodon and Bluesky. One comment thread points to "self-radicalization + being egged on by mastodon" - the platform's culture of moral urgency and callout culture accelerates absolutist positions.

Bluesky's more technically-focused culture resists these dynamics. AT Protocol's architecture prioritizes composability and choice over ideological enforcement.

The Real Question

This isn't about whether LLM training is ethical or whether defense contracts are acceptable. Those are legitimate debates worth having.

The question is: Should open source maintainers use blacklists and social pressure to enforce their answers on the entire community?

When the answer is yes, we get purity spirals. When it's no, we get messy pluralism where people with different values still build software together.

The Rust community will test this over the coming months. Given widespread LLM adoption among maintainers themselves, my prediction: the shunlog approach gets rejected. But the attempt reveals something important about how technical communities fracture under moral pressure.

Gatekeeping is easier than building. That's why it's so tempting when you're burned out.

Sources: Software Freedom Conservancy on ethical licensing, LWN on copyleft limits, Bluesky threads from fasterthanli.me and iximeow