Closed-loop funded audit toy
Read funded-audits.md.
python3 -m unittest discover -s papers/07-release-packet/models/funded_audits -p 'test_*.py' -v
python3 papers/07-release-packet/models/funded_audits/run_experiments.pyPython3.9+ standard library only. Eighteen tests,11 scenarios,20 seeds each, 220 run rows and11 complete illustrative traces. The existing math sampler is imported; its digest is recorded alongside local source hashes.
Fixed ex-ante offers, concealed hard quota, all decisions committed before audit, actual detection and fully collateralized toy losses. Refused selected slots are unused; no silent quota redistribution. Separate token audit/reward budgets and one person-hour ledger across review/audit roles. Shared roles audit another person’s work, never their own. Preflight capacity/independence/funding failures return unavailable before offering; broken promises appear only under an explicit negative-control flag.
Selection inclusion probability is k/N; executed-audit fraction among participants is random and can differ per run. Its conditional marginal requires behavior independent of concealed selection. No actual funds, sanctions, authenticated identities or deployment authority exist. This is a standalone behavior/resource model, not a replacement for the protocol runtime or a cooperation equilibrium.
Red review found that the first floating-hour ledger allowed positive
work within a
tolerance above zero capacity. The repaired ledger uses exact
decimal-rational reservation/debit arithmetic with no overspending
tolerance. Regression tests reject positive tiny work against zero
capacity and verify that .1 plus .2 exactly fills .3. The original
source is preserved in
models/funded_audits/results/model-before-exact-ledger.txt.
The red review also exposed binary-floating tie reversals between honest/shallow and honest/refusal utilities. Preferences now use exact decimal-rational arithmetic internally; output includes the exact advertised utilities. Regression tests preserve the specified refusal-first, then honest, then shallow tie rule. Token rewards and losses enter utility at a stipulated one-to-one marginal conversion, not an inferred welfare measure.