A book in progress · draft in public · 2026
Simulating futures to build the one we want. The story of teaching computers the law, giving them everyone's circumstances, and grading their predictions against reality — and of the fork that decides who such machinery serves.
Every dot is a household. Same average income either way — but only one view shows who falls off a benefit cliff.
In 1957, Guy Orcutt noticed that national averages can be identical while the societies underneath them differ completely — so he proposed simulating the households instead. This book is about what happened when that idea finally got the machinery it needed, and the discipline that keeps the machinery honest.
The discipline the book runs on
A simulation is admissible only where its verification chain terminates in ground truth — a rule checked against the statute, a population checked against the census, a forecast graded when the official number lands.
Chapter 3 states it; chapters 9–13 build the machinery that enforces it — merge-blocking source checks, oracle races against the models governments already trust, and a public scoreboard that waits for reality.
A draft, graded like its subject
The book argues that claims should carry their verification state, so this draft carries its own. These figures are computed from the manuscript at every build; the pending marks are visible in the text where they sit.
occasional updates as chapters land · no spam