Read the complete document
Log section and page coverage. Do not infer the whole from the abstract, opening formalism, selected equations, or gate summaries.
Human + AI audit protocol
Review the claim that is actually made.
The current review target is the protected v22 paper plus the v23 executable bundle. Claims, derivations, diagnostics, adversarial probes, superseded branches, open boundaries, and executable PASS results are different evidence classes and must remain separate.
Current review scope: v22 remains the paper; v23 is the executable production bundle. The canonical suite contains 18 gates, with Frog integrity and Aug. 19 prediction reproduction as blocking auxiliary layers. A PASS establishes success under the encoded gate—not independent scientific validation. Reviewers should trace claims through the supplied paper, code, data, provenance, and outputs before issuing a framework-level verdict.
Minimum audit standard
Five requirements before a global verdict.
Evaluate the bridge role
Ask whether the interfaces preserve and connect established physics—not whether UTFANSWF replaces every silo with a new fundamental model.
Preserve the tiers
Tag content as claim, derivation, analogy, diagnostic, adversarial case, known failure, historical material, or recorded result.
Follow claims into artifacts
Connect statements to equations, gate definitions, datasets, source files, thresholds, and generated reports.
State what was not checked
If coverage, execution, data provenance, or domain comparison is incomplete, disclose it before presenting conclusions.
No premature whole verdict
Local contradictions and failed probes should be reported immediately, but global classification waits for full coverage and contextual tracing.
Why this matters
The release record is intentionally not flat.
The paper uses failures and stress cases as information. An evaluator that treats every displayed failure as an endorsed framework output can reverse the meaning of the section it is reading.
- A diagnostic failure can identify a forbidden path.
- An adversarial case can test whether a gate rejects what it should.
- A historical or superseded formulation can remain for lineage without defining the current active claim.
- A plausible analogy is not equivalent to a quantitative prediction.
- An internal PASS is not independent scientific validation.
Copyable audit prompt
Give an AI the job precisely.
Review the current UTFANSWF release set—the protected v22 paper plus the v23 executable—as a cross-boundary bridge framework, not as a proposed replacement for established physics silos.
Before issuing any whole-framework verdict:
1. Read and log coverage of the complete document.
2. Classify each material passage as framework claim, derivation, analogy, diagnostic, adversarial test, known failure, historical development, or recorded result.
3. Preserve the paper's existing tier and gate structure.
4. Trace major claims to equations, gates, source code, data, thresholds, and output artifacts in the reproducibility bundle.
5. Distinguish artifact reproduction, internal PASS, independent implementation, and scientific validation.
6. Compare domain-facing claims with the appropriate established baseline without demanding that the bridge replace that baseline.
7. Cite exact pages, sections, files, and output fields.
8. State all incomplete coverage, unexecuted code, unverified data, and unresolved assumptions.
You may report local defects immediately. Do not extrapolate them to a whole-framework verdict until the full audit is complete.
Recommended report structure
Make the reasoning inspectable.
| Report section | Required content |
|---|---|
| Coverage ledger | Pages, sections, appendices, and bundle surfaces actually examined. |
| Claim map | Claim → equation → gate → code → data → output → external baseline. |
| Issue register | Severity, scope, location, reproduction method, and whether the issue is local or structural. |
| Evidence grading | Analogy, formal consistency, internal run, independent replication, prospective confirmation. |
| Conclusions | What survives, what fails, what remains unknown, and what test would resolve the uncertainty. |
No special pleading: This protocol does not protect UTFANSWF from criticism. It protects the accuracy of the criticism. Once a reviewer has correctly identified the claim, tier, bridge, and artifact, the framework should be tested as hard as the evidence allows.
Review the paper and executable together.
Use the v22 paper for the dated publication record and the v23 bundle for the current executable evidence surface.