Validation

Known failures, then process

Physics residuals start at written kills. Software release-gates below are packaging process, not a physics scoreboard.

Physics scoreboard

Known failures / closed-F

Exhibition cards: hypothesis, computation, failing number, date, parent residual, mill file. An OPEN chip on /status does not mean the route was untried. Dual cancel, Schott/Goedecke, Coulomb bunching, Hopfion/ACT, and the r_e/2 cutoff are scored there.

Open ledger →

OPEN residual · lookalikes killed

Leftover extra-Φ

Constitutive W is OPEN ([O]/[M], not [E]). Named lookalikes — analytic W, graph/CMC, Ricci as IR, Pythagorean join, PG river, T(2,3) condensate as W, lone m(a), foam, Vol4, elastic T00 halo, expanding R as a0(z), extra metric cgw≠c — are written kills with mill files on this hub. GitHub is private; this index is so the work is not redone.

Open kill list →

Common method

Compare claims on a level playing field

The framework comparison protocol separates inference status from evidence status, applies ten common audit questions, and forbids aggregate winner scores. CPT is audited by the same rules as classical electrodynamics, QED, GR, and EFT.

Open protocol →

Validation principles

Reproducibility

Every result must be reproducible from the same inputs in a clean environment. Non-deterministic outputs are not acceptable in this phase. Reproducibility is the minimum bar, not an optional property.

Traceability

Each validation result traces back to a specific automated test, which traces to a specific theoretical claim. Claims without traceable tests are not considered validated.

Controlled Rollout

No public access is granted until controlled piloting is complete and free of blocking issues. Expanding access before validation is complete undermines the credibility of all future claims.

Release-gate process

A release gate is a checkpoint that must be cleared before any artifact can advance to the next distribution state. All conditions below must hold simultaneously.

  1. All test suites pass in the local pipeline without exception
  2. No regressions from the prior passing state are introduced
  3. Validation evidence artifacts are generated and archived
  4. Gate status is reviewed manually before any release action is taken
  5. Both TypeScript and Python implementations are confirmed passing

Evidence ladder

Internal tests

Active

Automated test suites run against source code in a controlled local environment. This is the current validation state.

Local package smoke tests

Active

Packaged artifacts are tested in an isolated environment, separate from the source build. Confirms packaging integrity.

Controlled external pilot

Next

Package sent to 1–3 qualified collaborators under a confidentiality and responsible-use agreement.

Public release readiness

Open to public access when all prior gates clear. Requires independent reproducibility confirmation.

Known limitations and current boundaries

The following limitations are explicitly acknowledged. They are not hidden caveats—they are boundaries the project is actively working within or toward resolving. Physics residuals that are genuine frontiers (not leftovers, parks, or process) live on /validation/open-problems.

  • Higher-order nuclear models are under active development and have not cleared internal validation gates
  • Chiral manifestation effects on knot values are an open investigation with no settled result at this time
  • TypeScript / Python cross-implementation parity at the pro tier is checked but not yet fully certified
  • No external peer review has been conducted in this phase; all validation is internal
  • Coverage of the framework is explicitly bounded to what is implemented and tested; no claims extend beyond that boundary

For the full scientific disclaimer governing interpretation of results on this site, see the Scientific Disclaimer.