What this is. A short philosophical treatise on a real engineering problem: a blockchain program can verify signatures but cannot see the world — so how could it ever trust that some off-chain agent truly performed a behaviour, such as checking that a domain’s DNS records prove who owns it? The book argues there is a clean and surprising answer, states it as a conserved law, and derives seven concrete techniques from it — with digressions through Asimov, Star Trek, Tolkien, and the legend of the golem along the way.

Who it is for. Anyone curious. If you are not a cryptographer, begin with An On-Ramp to the Problem, which builds every idea you need from scratch. If you already speak in signatures and TXT records, jump to the treatise proper.

How it came to be. This is a human–AI collaboration: a person posed the question and the seed idea; the treatise — its framing, its laws, its proofs, and its prose — was written by Claude. The full story, and the original prompt, are in The Commission and the Colophon.