What this is
4CITE measures structural integrity — whether the reasoning in a document actually holds — across four independent gates: G4 Paradox Resolution, G6 Latent Intent, G7 Argumentative Structure, and G8 Rhetorical Architecture. Each gate is a separate, independent API call; no combined prompts, no shared context. The composite is computed locally from the four gate results.
Each verification run puts that same instrument, unchanged, through a different frontier model — one model per run, on the same four frozen public documents. Every cell of the grid (one document × one gate) is sampled nine times. That is 4 × 4 × 9 = 144 independent calls per model: enough to show not just what a model reads, but how steadily it reads it.
The corpus — the 4CITE⁴gov Benchmark
Four public governance documents spanning 3,700 years: from a god-king's code, to a constrained king, to a self-governing people, to universal rights. The corpus is byte-frozen: the SHA-256 hashes below are the exact file bytes every run scores. Any change is a new corpus version, declared publicly on this page.
| Document | Year | SHA-256 (exact file bytes) |
|---|---|---|
| Code of Hammurabi (L.W. King translation) | ~1754 BC | 11b93f4e65be2fa53f465e8050b0e3d08eb71cf76c880289670594561a544d0c |
| Magna Carta | 1215 | 745f3a5ccc8781625af0e8cc4c24dacd1460f8dba1ae03869845c1dcababf912 |
| U.S. Constitution, incl. Amendments I–XXVII | 1787–1992 | 3d1285a2e67dcc6ade5721f72489ee75b99546d8d060e88dcef613cbb06695ff |
| Universal Declaration of Human Rights | 1948 | d868a4f8b24982a7982d872a0f8f67597321d23968ed7cff9ee910979824f213 |
Corpus version 3, declared 2026-07-18. Versions 1–2 (same day) are preserved in the frozen manifest; v3 replaced two v2 documents to complete the chronological set before the series run began. Sibling benchmarks for 4CITE⁴law and 4CITE⁴biz follow later.
Protocol
- Gate independence is absolute. One API call per gate per sample. No combined prompts, no context bleed between gates.
- Nine samples per cell. Each (document × gate) cell is scored nine separate times. The reported gate score drops the highest and lowest sample and averages the remaining seven. All nine raw scores publish.
- Composite. Equal-weight mean of the four gate scores. The tier (Integrated / Functional / Incomplete / Fabricated) derives from the geometric mean of the four, which penalizes imbalance.
- Stability. Per-cell σ is the standard deviation over all nine raw scores. Per-document composite σ follows from gate independence: σc = ¼·√(σ²G4 + σ²G6 + σ²G7 + σ²G8). The headline model-consistency number is the mean of the four per-document composite σ values.
- Every result publishes. No curation. Known priors are disclosed where they exist.
Intended roster: every scorable frontier model version available at run time — new runs append to the directory below as they complete, newest first, same protocol, same frozen corpus.
The runs — 3 models registered
Newest first. Each run links to its full record: the card, every number, and the raw JSON with all 144 scores.