Browser-level proof of determinism

Tests reproducibility,
proof of tampering

These tests are intentionally simple. They demonstrate a core RPO property: defensible determinism.
Same input → same output, SHA-256 is re-computable, tampering becomes detectable.
No server, no external service: everything runs locally.

Open the sandbox Read the RPO spec

Important: passing these tests does not “prove truth”. It proves the pipeline is auditable, and transformations become explainable.

Test matrix
Test What it verifies Status Notes
Determinism The same input produces the same canonical bundle and the same hash. Not run Runs extraction twice, compares canonical payload + SHA-256.
SHA-256 The hash is reproducible for the same canonical string. Not run Uses local WebCrypto SHA-256 (no backend).
Anchors Base anchors are extracted consistently (dates/places): no silent rewrite. Not run Conservative heuristic extraction: stability over “smartness”.
Output (proof)
Click “Run tests…” to get local proof: results + hash + canonical payload.
Key idea: these tests don’t validate an “app”. They validate a trust boundary: the system cannot “change its mind” silently.
Why it matters (one sentence)

You’re not challenged for your decisions. You’re challenged for what you cannot reconstruct — and therefore cannot defend.

Want the full pilot (CNRS × TruthX): interpretive coherence, structured narrative layers, advanced signal fusion, and a human-readable proof report designed for professionals.
Request pilot access

This page is informational. The test environment does not store case data and does not determine truth. It demonstrates a deterministic structure and verifiable integrity anchors.

Gersende Ryard de Parcey is the editor of the OpenProof RPO v0.1 specification (Rapport Probatoire Ouvert), a proof standard based on signed JSON, a human-readable PDF, and a public SHA-256 hash.