Tools vs. Treaties - by Dave Griffith

The gas gauge arrow that tells you which side your fuel cap is on was invented in 1986 by a Ford designer who got soaked switching sides at a gas station. He wrote a one-page memo, and three years later it was in every Ford. Now it's in every car worldwide. That's epistemic infrastructure - the right information, at the right moment, to the person who needs it. No coordination required, no international standards, no training. Just works.

This week two pieces crossed my desk about AI governance that represent completely different theories of change. One proposes an international treaty restricting AI compute above certain thresholds, requiring global coordination on strategically vital technology. The other sketches ten practical AI tools anyone could build - community notes for everything, rhetoric-highlighting that flags manipulative framing, reflection-scaffolding that asks if you're sure before you send that angry email. One requires convincing every major government to agree and comply. The other requires a developer and a weekend.

The pattern matters because the constraint that made Moylan arrows rare is dissolving. In 1986, you needed to work at a car company to fix the gas pump problem. Now anyone can build browser extensions, AI-powered tools, epistemic infrastructure. People are already doing it - Future Tokens for argument analysis, specialized agents for codebase learning. The Forethought sketches describe tools small teams could prototype in weeks, not moonshots requiring institutional backing.

Governance that actually works tends to be invisible - circuit breakers that halt trading automatically, unemployment insurance that scales with recessions, type systems that make wrong code harder to write. The Moylan Arrow is governance. It just governs so quietly it doesn't look like governance at all. The changes coming won't be governed through treaties because there isn't time. They'll be engineered by people who notice confusion and build something that alleviates it. Not because they filed patents or wrote position papers, but because useful things get adopted.