Public audit method · draft 0.1 · 2026-09-06
Framework comparison protocol
Same questions. Same evidence burden. No aggregate winner score.
The protocol compares individual claims, not reputations or whole theories. It records what each framework assumes, derives, reconstructs, tests, leaves open, and rules out. CPT receives exactly the same classifications and no exemption for being the host framework.
Two axes, kept separate
A theorem can be empirically untested; an effective model can be extraordinarily well tested. Claim status describes the inference. Evidence status describes contact with observations.
A. Claim status
Identity
True by definition or by an already fixed mathematical identification.
Check: Does it add physical dynamics, or only rename an object already present?
Theorem
Follows deductively from stated premises within the formal system.
Check: Are every premise and regularity condition declared before the conclusion?
Native derivation
Follows from the framework's own action or equations without importing the target law.
Check: Would the result survive if the target equation were removed from the inputs?
Conditional reconstruction
Recovers a result after loading named assumptions, equations, or boundary data.
Check: Which part is supplied by the framework, and which part is imported?
Effective model
Controlled description in a declared regime, often with measured or matched coefficients.
Check: Is the validity range and truncation or matching error stated?
Package / dictionary
A consistent interpretation or map between descriptions, not itself a dynamical derivation.
Check: Does the dictionary predict anything that was not used to construct it?
Input
An empirical constant, law, state, boundary condition, or calibration loaded into the calculation.
Check: Is the input visibly named at every downstream claim that depends on it?
Open residual
A result required by the framework's declared ambition but not currently supplied.
Check: Is there a concrete pass condition rather than an indefinitely movable goal?
Closed-negative route
A named mechanism has failed a declared discriminator; the parent problem may remain open.
Check: Are the failed route and the still-open residual kept distinct?
Outside scope
The framework does not claim to explain this item and should not be penalised as though it did.
Check: Is the exclusion consistent with the framework's own stated aims?
B. Evidence status
Formal only
Mathematical consistency or an internal calculation with no empirical comparison yet.
Check: Is there an observable and a measurement protocol?
Retrospective fit
Parameters or choices were selected using the same data later quoted as agreement.
Check: How many choices were made after seeing the target?
Benchmark recovery
A known result is reproduced with inputs and assumptions declared.
Check: Is this a useful consistency check, or an independent prediction?
Constrained postdiction
Existing data not used to set the relevant structure are recovered within a frozen model.
Check: Was the model genuinely frozen before this comparison?
Independent test
A preregistered or otherwise independent prediction is compared with new or withheld data.
Check: Could the result have failed without changing the framework after the fact?
Falsified in domain
A prediction fails a valid test inside the framework's declared regime.
Check: Was the tested claim actually entailed, with no scope or implementation mismatch?
Common audit card
Every framework dossier must answer the same ten questions before comparative conclusions are written.
- 01
Declared scope
What does the framework claim to explain, and what is outside scope?
- 02
Primitives
Which fields, state spaces, symmetries, constants, and background structures are assumed?
- 03
Native dynamics
Is there an action or evolution law, and which results follow from it?
- 04
Loaded inputs
Which measured constants, target equations, initial states, and boundary conditions enter?
- 05
Inference chain
Where are the transitions from identity to theorem, reconstruction, and measurement?
- 06
Validity regime
What approximation, scale, truncation, or matching domain controls the result?
- 07
Empirical contact
Was the comparison fitted, recovered, withheld, independently predicted, or falsified?
- 08
Residuals and negatives
What remains open, and which specific routes are already closed-negative?
- 09
Falsifier
What observation or calculation would make the claim fail without moving the goalposts?
- 10
Provenance
Can each classification be traced to a primary source, calculation, data release, and version?
Pilot calibration matrix
A first-pass calibration of the protocol, not a completed literature audit or ranking.
| Framework | Declared scope | Native core | Loaded inputs | Strongest fair claim | Evidence standing | Boundary |
|---|---|---|---|---|---|---|
| Classical electrodynamics | Classical fields, charges, radiation, and forces in its domain. | Maxwell equations/action plus Lorentz force or a specified matter action. | Charges, masses, currents or matter model, constitutive relations, and boundary/initial data. | Native field propagation and force/radiation consequences once sources and matter coupling are specified. | Extensively independently tested in its domain. | Does not by itself derive the numerical charge spectrum, particle masses, or quantum measurement statistics. |
| Standard Model / QED | Quantum particle interactions for the stated gauge fields and matter content; gravity excluded. | Renormalised gauge-field and matter action with quantum calculation rules. | Measured couplings, masses, mixing parameters, state preparation, and experimental conditions. | Native quantum amplitudes and radiative corrections conditional on the measured parameter set. | Exceptional independent precision and collider validation. | Empirical success does not turn fitted parameters or selected field content into derived constants. |
| General relativity | Classical spacetime geometry and its coupling to specified stress-energy. | Einstein-Hilbert dynamics or equivalent field equations plus a matter model. | G, possible Lambda, matter content/equation of state, and initial/boundary data. | Native spacetime dynamics and derived weak-field, orbital, lensing, and wave consequences. | Broad independent tests from laboratory to strong-field astronomy. | A chosen matter model or cosmological initial condition is not derived by the geometric field equation alone. |
| Effective field theory | A controlled low-energy expansion, not one universal ontology. | Symmetry-allowed operators organised by scale and power counting. | Degrees of freedom, cutoff/matching scale, Wilson coefficients, and renormalisation prescription. | Systematic predictions with quantified truncation inside the declared regime. | Domain-dependent; often strongly validated through its host theories. | Matched coefficients and UV completion are inputs unless separately derived. |
| CPT current skeleton | A one-force geometric programme aiming at measurement equivalence rather than SM/QED ontology. | Geometric supports and labels, typed SAP requirements, and the gated BQ face-current map. | e, alpha, one length/scale access, occupation, and the standard relativistic ED class for the present two-seat response. | Conditional BQ point-current placement followed by Coulomb/Darwin reconstruction; several route classes closed-negative. | Early research programme: benchmark and compatibility work, no discriminating independent prediction yet. | Native action, formation, hold, hard join, absolute scale/R ontology, and an exposed internal-geometry observable remain open. |
Evidence and correction policy
- Prefer the framework's own canonical action, specification, or status document for declared scope and primitives.
- Use primary derivations for theorem and native-derivation claims; record every premise and approximation.
- Use experiment-collaboration releases, metrology institutes, or peer-reviewed data for evidence status.
- Separate a framework advocate's interpretation from the result actually established by the cited source.
- Version every dossier and retain corrections; disputed rows stay visibly disputed rather than silently rewritten.
Non-negotiables
- No aggregate winner score or percentage-complete leaderboard.
- No penalty for an item consistently declared outside scope.
- No crediting an imported target equation as a native derivation.
- No treating empirical success as proof that all ontology or constants were derived.
- No treating a closed-negative mechanism as closure of the parent residual.
- No special vocabulary or relaxed evidence burden for CPT.
CPT self-application
CPT self-application
CPT is specification-complete on its present skeleton, conditionally reconstructive in selected sectors, and generatively incomplete. BQ places input e; the Coulomb/Darwin step imports the standard ED class; the native action, formation, hold, hard join, scale ontology, and a discriminating observable remain open.
Inspect the full CPT status audit →