Every encoding, cross-checked
Open isn't enough — an encoding you can read but can't test is still a claim. We run every published rule against independent engines we don't control, and publish the comparisons so anyone can re-run them.
The oracle dashboard
Open full sizeCross-engine comparisons as they stand — every divergence is a question about the law or its encoding.
The oracles
PolicyEngine
Open-source microsimulation of US and UK tax and benefit policy — the broadest-coverage oracle we compare against.
Visit the oracleTAXSIM
NBER's tax calculator, the reference standard in economics research for US income tax liabilities.
Visit the oracleUKMOD / EUROMOD / SOUTHMOD
The EUROMOD family of tax-benefit microsimulation models, maintained by university and EC research teams.
Visit the oracleSPSD/M
Statistics Canada's Social Policy Simulation Database and Model — the reference microsimulation of Canadian federal and provincial taxes and transfers.
Visit the oraclePSL Tax-Calculator
The Policy Simulation Library's open-source US federal tax model, developed in the open with public revision history.
Visit the oracleACCESS NYC
New York City's public benefits screener — its published Drools rules and Screening API make it a checkable oracle for city-level eligibility.
Visit the oracleSNAP quality-control data
USDA's case-level QC microdata — real adjudicated cases we replay against the encodings to catch divergence from practice, not just from text.
Visit the oracleHow a comparison runs
Same case, every engine
Concept-keyed test cases — a household, an income, a date — run through each oracle behind a thin adapter, so no engine's quirks leak into the comparison.
Normalized outputs
Engine results map onto shared program concepts before comparing, and every mismatch lands in a taxonomy: encoding bug, oracle bug, or genuine ambiguity in the law.
Disagreements explained
Reviewer agents explain each discrepancy. Disagreements get explained, not erased — the audit reports ship with the release.
The comparison harness is open, and every engine behind a thin adapter makes the whole ecosystem harder to fool. If you maintain a calculator, a screener, or an eligibility system, plug it in and put us to the test.
Looking for the engineering contracts behind this? Documentation →