Watch the tests

Every result is dated and linked to its source, including the cases that fail. If the design can't meet its own standards, this is where you'll read it first.

Milestones

  1. DONE · OCT 2026Accounting model publishedOpen source; anyone can run it.
  2. DONE · NOV 2026Challenge casesAn 18-case first assessment, then 20 held-out cases written and sealed by a separate AI evaluator.
  3. IN PROGRESSRules for the gapsAdopted on paper; being built and checked.
  4. NOT STARTEDEconomic testsPrices, credit and the transition, including cases chosen to make the design look bad.
  5. NOT STARTEDOutside expert reviewMonetary economics, banking and constitutional law.

Latest result

20-case held-out challenge · run November 1, 2026 on locked code · reports on GitHub

15passed fully
2partial
1tests a feature not yet built
2ambiguous or incomplete as written

A separate AI evaluator wrote and sealed the cases before seeing the code. First run: 9 passed. After the evaluator corrected its own test harness (the code was not changed), 15 passed fully. The executable harness was written after the evaluator could see the code. Earlier, an 18-case first assessment found 9 passes, 6 rules not yet decided, 3 features not yet built and 0 accounting failures.

NOV 2, 2026 · DEBT-BUDGET MODEL 0.3 · DRAFT, NOT MERGEDThe example's accounts balance, but its transaction timing violates Treasury's upper cash limits and the monthly issuance schedule, and misses the required cash-remedy deadline. The next step is to test whether realistic obligations can be met under the existing rules, before proposing rule changes. The failing example is kept unchanged. All figures are synthetic.

These are accounting checks, not a score for the proposal, and this was AI-assisted checking, not an outside professional audit. These checks do not establish inflation outcomes, credit costs, or the economic and operational feasibility of the transition.

What we may and may not say May: an early accounting model passed specified checks and exposed gaps; rules for the gaps are adopted on paper. May not: “all tests passed,” “validated,” “proven to stabilize prices,” or “independently audited.”

The book's 22 promises, tracked

Part I of the book makes 22 promises about what the proposal must show. Each one is listed here with the evidence for it.

StatusCountExample
Met1Results anyone can download and rerun
Partly backed by accounting4New money without government borrowing
Rule decided, not yet built3Emergency powers with written limits
Waiting for economic tests5Stable prices; fair-cost credit
Not started9Privacy design; amendment text

Report a correction

Found an error in the book or on this site? Tell us the page and the source. Accepted corrections are published in a dated log, with credit if you want it.

Email corrections to corrections@peoplesmoneyproject.org. Please include the chapter, page or URL and your source.

Methods & AI disclosure

AI tools assisted with drafting, source research, code development, and separate accounting checks. The code, test cases, failures, and revisions are publicly available. These checks are not an independent professional audit. They test the accounting of specified cases, including a synthetic transition; they do not establish the proposal's economic effects, the transition's economic or operational feasibility, or its legal feasibility.

Corrections log: none published yet