AI

Who Checks a Proof No Human Can Read? — Leo de Moura

Read full article at source ↗

Based on the episode description, not a full transcript. Source evidence retrieved 30/09/2026, 03:58:24 UTC.

Briefings

Leo de Moura discusses Lean's kernel architecture, explaining how independent checkers like Nanoda and Z3 verify proofs without relying on a single trusted kernel. He details the Collatz incident where two different bugs allowed a proof to pass both verification systems, highlighting the risks of formal verification leaving the lab. The episode covers Lean's evolution from a small trusted kernel to a larger ecosystem, emphasizing the importance of independent verification and transparency.

Key points

• Lean's core kernel remains small and protected, with independent checkers like Nanoda and Z3 verifying proofs without relying on a single trusted kernel.
• The Collatz incident demonstrated that a proof could pass verification in both Lean's official kernel and Nanoda by exploiting different bugs in each system.

Context and limitations

Coverage is limited to the podcast episode discussing Lean's architecture, verification methods, and the Collatz incident. Specific details about the exact nature of the bugs exploited in the Collatz incident are not fully detailed in the provided evidence.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 30/09/2026, 04:00:19 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeia4oag6pbn7sh2jhb5obu2jojzgbpjakacqhh2ivkshgn7r74juii