chore(physics): honest-provenance cleanup — demote P5, drop fake-precision score, enforce SPECULATIVE != P0/P1#414
Merged
neuron7xLab merged 1 commit intomainfrom Apr 26, 2026
Conversation
…ision score, enforce SPECULATIVE != P0/P1
Honest self-audit found four lapses in the seven physics-schema PRs from
this session:
(a) INV-OBSERVER-CPT (P5) was a tautological schema with no peer-reviewed
model — shipped as a P2 invariant alongside ANCHORED entries. Demoted.
(b) `TRUTH_COHERENCE_SCORE: float` introduced fake precision (vibes-derived
numbers like 0.9 / 0.6 / 0.55 / 0.3 dressed as quantitative). Dropped.
(c) Validator did not enforce that SPECULATIVE invariants stay below
fail-closed tier. Now it does.
(d) Session-level claim "operationalized 7 fundamental physics problems"
overstated schema work as derivation work. Restated honestly: 6
contract schemas remain — 2 ANCHORED, 4 EXTRAPOLATED, 0 SPECULATIVE
(the seventh, P5, is now a demo not a registered invariant).
Changes:
- Moved core/physics/observer_cpt_asymmetry.py →
examples/observer_cpt_schema_demo.py with explicit DEMO ONLY header.
Removed tests/unit/physics/test_observer_cpt_asymmetry.py.
Removed pncc.observer_cpt block from .claude/physics/INVARIANTS.yaml,
with a note pointing at the demo file.
- In core/physics/{arrow_of_time, observer_bandwidth,
cosmological_compute_bound, jacobson_observer_coherence}.py:
replaced
PROVENANCE_LEVEL: str = "..."
TRUTH_COHERENCE_SCORE: float = 0.X
with
PROVENANCE_TIER: Literal["ANCHORED", "EXTRAPOLATED", "SPECULATIVE"] = "..."
Discrete tier; no float score. Updated test files to assert
PROVENANCE_TIER and dropped the floor/ceiling float assertions.
- In .claude/physics/INVARIANTS.yaml: dropped all four
`truth_coherence_score:` lines. Kept `provenance:` field as the
canonical tier indicator.
- In .claude/physics/validate_tests.py::_self_check: added check #5 —
SPECULATIVE invariants cannot carry P0 or P1 priority. Validator
reports FAIL with the offending IDs if found.
Quality gates:
pytest (4 modules) 59/59 PASS
ruff check clean
ruff format --check clean
black --check clean
mypy --strict (5 modules + tests) clean
validate_tests.py --self-check PASSED — 80 invariants loaded,
check #5 confirms 0 SPECULATIVE
at P0/P1.
Honest stack as of this PR:
ANCHORED: INV-ARROW-OF-TIME (P0)
INV-BEKENSTEIN-COGNITIVE (P0)
EXTRAPOLATED: INV-OBSERVER-BANDWIDTH (P1)
INV-COSMOLOGICAL-COMPUTE (P1)
INV-JACOBSON-OBSERVER (P1)
INV-SIMULATION-FALSIFICATION (P1, registry/statistical)
Demoted: INV-OBSERVER-CPT → demo, no registry weight.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
neuron7xLab
added a commit
that referenced
this pull request
Apr 26, 2026
…416) Critical-cognitive-engineering-perfectionism encoded in the validator: the same failure modes the kernel just cleaned up should not be reintroducible without the validator catching them. Two new checks. Check 6 — fake-precision regression guard: Scans .claude/physics/INVARIANTS.yaml AND core/**/*.py for `truth_coherence_score:` / `TRUTH_COHERENCE_SCORE`. Either reintroduces the float-as-confidence pattern that PR #414 removed in favor of discrete PROVENANCE_TIER. FAIL if any hit. Verified by injection test: appending "truth_coherence_score: 0.42" to the YAML triggers FAIL with exit code 1; reverting restores PASS with exit code 0. Check 7 — source/tests path integrity: Each invariant declares `source:` and `tests:` paths. Validator resolves them on disk; missing paths FAIL. Strips optional `::symbol` anchor (function/class inside a file) before existence check. This check found a real deviation immediately: INV-DRO1's source field is `core/dro_ara/engine.py::derive_gamma`, which the strip handles correctly. Going forward, file moves or typos surface at self-check time instead of at consumer time. Both new checks integrate with the existing failure-aggregation pattern and contribute to the FAIL exit code when violated. The previous five checks (load count, dispatcher coverage, regex consistency, theory cross-ref, SPECULATIVE-misranked guard) remain unchanged. Quality gates: ruff check / ruff format / black / mypy --strict clean validate_tests.py --self-check PASSED, 7/7 injection regression test (truth_coherence_score) detected, exit 1 Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What and why
Self-audit found four lapses in the seven schema PRs from the previous session. This PR addresses all four.
(a) INV-OBSERVER-CPT (P5) demoted. Tautological schema, no peer-reviewed model cited. Moved to `examples/observer_cpt_schema_demo.py` with explicit DEMO ONLY header. Removed from registry; removed its tests.
(b) Dropped `TRUTH_COHERENCE_SCORE: float`. Vibes-derived numbers (0.9 / 0.6 / 0.55 / 0.3) presented as quantitative. Replaced with discrete `PROVENANCE_TIER: Literal["ANCHORED", "EXTRAPOLATED", "SPECULATIVE"]`. No fake precision.
(c) Validator enforces tier vs priority. `validate_tests.py --self-check` check #5: SPECULATIVE invariants cannot carry P0 or P1 priority. P2 is informational; that is the highest tier a SPECULATIVE invariant may carry. Now active in CI self-check path.
(d) Session claim restated. Was: "operationalized 7 fundamental physics problems". Honest: 6 contract schemas (2 ANCHORED, 4 EXTRAPOLATED), 1 demoted to demo.
Quality gate
Honest stack post-merge