Assess remaining research value with reproducible controls

.github/workflows/value-assessment-probes.yml

@@ -0,0 +1,26 @@
+name: Research value assessment probes
+on:
+  pull_request:
+    paths:
+      - 'research/value-assessment-20261112/**'
+      - '.github/workflows/value-assessment-probes.yml'
+  push:
+    paths:
+      - 'research/value-assessment-20261112/**'
+      - '.github/workflows/value-assessment-probes.yml'
+  workflow_dispatch:
+permissions:
+  contents: read
+jobs:
+  reproduce:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: '3.14'
+          cache: pip
+          cache-dependency-path: research/value-assessment-20261112/requirements.txt
+      - run: python -m pip install -r research/value-assessment-20261112/requirements.txt
+      - run: python -B -m unittest discover -s research/value-assessment-20261112 -p test_check.py
+      - run: python -B research/value-assessment-20261112/check.py

README.md

@@ -20,7 +20,7 @@ This repository is **not** a derivation of a new fundamental theory of gravity,
 
 ## Repository Layout & Verification Commands
 
-All Python scripts use only the Python 3 standard library. Each certificate file in `results/*.json` embeds a `sha256_canonical` field equal to the `sha256` checksum of `json.dumps(payload_without_sha_field, sort_keys=True, separators=(",", ":"))` computed by the generator script that writes it (this is a deterministic content checksum for reproducibility, not a cryptographic signature). Raw file digests are recorded in `SHA256SUMS`.
+Root-level Wave scripts use only the Python 3 standard library. Each certificate file in `results/*.json` embeds a `sha256_canonical` field equal to the `sha256` checksum of `json.dumps(payload_without_sha_field, sort_keys=True, separators=(",", ":"))` computed by the generator script that writes it (this is a deterministic content checksum for reproducibility, not a cryptographic signature). Raw file digests are recorded in `SHA256SUMS`.
 
 | Script / Config | Output / Target | Verification Command |
 |---|---|---|
@@ -46,3 +46,7 @@ All Python scripts use only the Python 3 standard library. Each certificate file
 - `CHANGELOG.md` -- Reconciled chronological history across Waves 1-12 matching `CLAIMS.md` and code identifiers.- 
 - mic-Dawn quasar confrontation, and next-agent handoff queue.
 - `CHANGELOG.md` -- Reconciled chronological history across Waves 1-12 matching `CLAIMS.md` and code identifiers.
+
+## Research value assessment
+
+The [12 November assessment](research/value-assessment-20261112/) records the Wave 11/12 audit, bounded research directions, reproducible controls and stopping gates. Its separate NumPy/SciPy dependencies are pinned in [the bundle requirements](research/value-assessment-20261112/requirements.txt).

research/value-assessment-20261112/README.md

@@ -0,0 +1,118 @@
+# Which threads merit further research?
+
+**Yes: there are specific questions worth pursuing. The evidence supports two bounded research directions and one preliminary theory test. It does not establish a common mechanism behind gravity, cosmic creation, fractals and quantum measurement.**
+
+This assessment follows the [nine-route exploration](https://github.com/Foundation-0/gravity/pull/8) and [cosmic-dawn audit](https://github.com/Foundation-0/gravity/pull/9). It refreshes all four repositories, tests the strongest remaining connections and records where a promising analogy stops making a physical prediction.
+
+Research was performed by agents using gpt-6-astra with the requested ultra reasoning. Publication preparation uses gpt-6.1-sol. Calculations and independent agent review are recorded with the artifacts. Human scientific review remains pending. No new astronomical observations or individual-galaxy likelihood fit are claimed.
+
+## Decisions supported by this pass
+
+| Question | Decision | What was established | Next result that would justify continuation |
+| --- | --- | --- | --- |
+| Does the galactic acceleration scale evolve with cosmic expansion? | Pursue calibration and inference design. Hold a new observational claim | Exact calibration symmetries and synthetic controls identify what the data must constrain independently. Published work already reports evolution and compares simple scalings. | Recover the selected galaxies and joint uncertainties, independently constrain calibration drift, and recover injected evolution under the actual selection. |
+| Does the lattice fold produce an unusual mechanical response? | Pursue one specified interaction model | The declared ideal spring network changes from three nonzero relaxed homogeneous moduli to one. A weak-spring completion changes the limiting response. | Establish the crossover under a physically chosen force law, including finite-strain and stability limits, and settle relevant prior art. |
+| Can dark-field collapse generate a useful effective pressure? | Pursue a finite AeST action, constraint and response test | A cold-wave control develops a rising pressure proxy without mode scattering. The proxy alone cannot establish thermalization. | Fix one complete branch, construct admissible states, and predict responses across permitted preparations before a nonlinear collapse study. |
+| Does lost interference determine readable environmental records? | Correct the interpretation. Park a new research programme | An established photon-scattering model separates decoherence from record production. The radiation's angular state matters. | A specified illumination or detector problem with a prediction absent from existing work. |
+| Does the quasar pair validate the proposed heavy-seed mechanism? | Reject the current inference | Wave 12 reproduces the quoted numerical scale, but that quantity is total halo mass. The trapping fraction is prescribed and no seed-formation solution is calculated. | A dynamical formation model and comparative likelihood against ordinary growth alternatives. |
+| Do future recollapse or a fifth dimension follow from the current evidence? | Park. Reject those deductions | Past expansion does not select a unique future continuation. Four-dimensional gravity already predicts black-hole clock effects. | A specified model with an observable that separates it from an adequate baseline. |
+
+“Pursue” means a defined next test is worth doing. It does not mean that the preferred explanation is likely true. “Reject the current inference” identifies a failed logical step without excluding every possible future completion of the idea.
+
+## The observational thread now has a precise obstacle
+
+The [acceleration study](observations/research.md) uses the empirical radial acceleration relation, with its characteristic scale written as `a0(z) = A E(z)^q`, where `E = H/H0`. A constant scale has `q=0`. A Hubble-tracking scale has `q=1`. These are testable phenomenological alternatives, not complete gravity theories.
+
+Let `b = B E^s` multiply the nominal baryonic acceleration and `k = C E^u` convert true total acceleration to measured acceleration. In the low-acceleration limit, the inferred slope combines `q + s + 2u`. Evolution of the physical scale and evolution of calibration therefore cannot be separated from that limit alone.
+
+The full interpolation has an exact symmetry as well: shifting `(q,s,u)` to `(q+d,s+d,u-d)` leaves every predicted measured acceleration unchanged. Even adding a local, zero-redshift normalization leaves this drift ambiguity. This is a property of the declared inference model, not evidence that a particular survey has errors of the required size.
+
+Independent calibration can break the ambiguity. The calculation gives conditional precision requirements and tests recovery with known calibration, acceleration-range holdouts, shared radial errors and a designed selection counterexample. It also shows why recasting the same masses and velocities as a Tully-Fisher relation does not supply independent evidence.
+
+The literature and data-access review distinguish published fit summaries, parent survey cubes, best-fit catalogues and the selected sample's joint posteriors. A fresh individual-galaxy fit is not yet supported by the retrieved inputs. The next useful contribution is a reproducible, independently calibrated inference, rather than another curve through published bins.
+
+## The geometry thread produces a narrower, tangible result
+
+The [mechanics report](convergences/research.md) and [derivation](convergences/derivation.md) hold the original fifteen bars fixed as unstressed central springs. After internal relaxation, the open geometry supports three independent homogeneous stress combinations. A nonzero fold supports one. The surviving stress has mixed-sign principal values, so it is not an ordinary fluid-like pressure response.
+
+Adding twenty-one explicitly declared weak springs removes the singular response at fixed positive added stiffness. Taking the fold to zero before removing those springs gives a different tensor from removing them first. Both the formula and the numerical controls are included.
+
+The added springs have their rest lengths reset at each reference configuration. This is a comparison family, not a manufactured material following one unchanged interaction law. The result concerns homogeneous elasticity. The completed model's full spectrum and nonlinear stability remain untested.
+
+Rank-one elasticity, framework flexibility and soft modes are established subjects. The project increment is the calculation for this particular fold and its stated weak-interaction completion. The [prior-art comparison](convergences/research.md) includes pentamode and unimode distinctions.
+
+Historical priority of that result is unresolved. A physically specified crossover or load response would be worth testing. A claim of a new universality class would be premature.
+
+## The collapse thread needs a better discriminator
+
+The [caustic control](closure/research.md) begins with coherent cold infall. A pressure-over-density proxy rises by 5.08 times through the first caustic, even though every Fourier occupation remains conserved and no modes scatter into each other. Filtered density and stress approach the corresponding collisionless multistream solution as physical dispersion decreases.
+
+This does not show that interacting AeST cannot develop an approximate pressure law. It shows that finite stress, smooth collapse and a rising proxy would not demonstrate such a law. Three admissible cold reference preparations also have identical density and zero local mean velocity while their stresses span a factor of four.
+
+The finite next test in the report specifies the full action and branch, checks the canonical constraints, then measures density, current and stress responses. A large three-dimensional simulation is justified only after that test identifies physical states, retained variables and a controlled range below the effective-theory cutoff.
+
+This connects the original question about what survives annihilation or collapse to a practical distinction: conserved charge, bound energy, collisionless stress and thermal pressure are different forms of persistence. The existence of one does not establish the others.
+
+## What the new gravity work changes
+
+The [Wave 11/12 audit](wave12/research.md) now traces the quoted `83,913.96` solar masses and `f_trap = 0.9999699` to explicit Wave 12 inputs. The extended spectrum converges for that calculation. This resolves the earlier missing provenance and high-wavenumber concern for the new result.
+
+The result remains a total halo mass, with an adopted baryon allotment of about 13,126 solar masses and virial compactness around `4.19e-11`. It supplies no black-hole formation condition. This does not rule out later collapse under a different, specified mechanism.
+
+The claimed absolute light-seed clock barrier also depends on opacity, efficiency, duty and accretion history. The audit gives counterexamples to the universal exclusion while retaining those physical qualifications. Direct radial integration changes the quoted infall times. An arithmetic error in the Lean chirp-mass bound has a separate [scoped correction](https://github.com/Foundation-0/gravity/pull/10).
+
+The Wave 11 future-collapse and condensate-origin proposals remain conditional. An alternative future equation of state matches the whole observed past and avoids the proposed turnaround. The minimal condensate construction still needs its stability, source or boundary mechanism, and allowed interactions specified. Algebraic symmetry alone does not establish a creation event or forbid all pair conversion.
+
+## How the repositories connect
+
+| Repository | Concrete role in the next work |
+| --- | --- |
+| [gravity](https://github.com/Foundation-0/gravity) | Keep physical assumptions, calibrated observational tests, action-derived dynamics and arithmetic proofs distinct. |
+| [fractals](https://github.com/Foundation-0/fractals) | Test which repeated structures retain a response after interactions, resolution and weak forces change. |
+| [physics0](https://github.com/Foundation-0/physics0) | Track the variables and preparations needed for prediction. Distinguish geometry, dynamics and readable records. |
+| [astr0](https://github.com/Foundation-0/astr0) | Find the independent measurements, catalogue joins and informative non-events that distinguish competing explanations. |
+
+The useful convergence is methodological and physical: a description can hide a degree of freedom, an interaction or a calibration that changes the apparent conclusion. Joining descriptions becomes productive when it yields a measurable distinction. These calculations identify several such distinctions without turning their resemblance into a common law.
+
+Further work should stop or change direction when independent calibration cannot distinguish the proposed laws, a mechanical effect has no useful range under a specified interaction, or an AeST branch lacks admissible states and a controlled approximation. Those are reviewable outcomes, including when the answer is negative.
+
+## Reproduction and provenance
+
+Use Python 3.14 in a disposable virtual environment, then run from the repository root:
+
+```sh
+python -m venv /tmp/value-assessment-env
+/tmp/value-assessment-env/bin/python -m pip install -r research/value-assessment-20261112/requirements.txt
+/tmp/value-assessment-env/bin/python -B -m unittest discover -s research/value-assessment-20261112 -p test_check.py
+/tmp/value-assessment-env/bin/python -B research/value-assessment-20261112/check.py
+```
+
+The runner regenerates all four probes in a temporary copy and compares saved scientific results. Floats use relative tolerance `1e-7` and absolute tolerance `1e-20`. The listed conservation and roundoff diagnostics use explicit upper bounds in `check.py`. Integers, booleans, structure and text must match. Version labels are excluded. Each probe also checks its own analytic and numerical assertions.
+
+The six files under `inputs/` are historical Wave 11/12 sources, proof statements and certificates from gravity `70a1204a536b3dca0b3053d1a056daf22601ac00`. The manifest records their SHA-256 digests and upstream URLs. The runner checks every digest before execution. The Wave 12 loader also refuses a changed Python source before executing it.
+
+These inputs deliberately retain the historical C33 error so the audit remains reproducible.
+
+The other assessed snapshots are fractals `48b923bd25dbcfef7fa8e8399d4144592145f0ef`, physics0 `913468f093f82d869dc58b404039ea0a416cfb80`, and astr0 `67cafe2defc44d5a4e17815fc7329f65eeef45eb`. Source registries beside each report identify the primary literature and additional pinned inputs.
+
+Publication preparation refreshed gravity main to `364fa3e`, which adds C31-C33 descriptions to `CLAIMS.md` without changing the audited Python inputs. That index repeats the historical C33 exponent error. The scoped correction covers the proof and index. The conditional scientific assessment remains applicable.
+
+The focused workflow runs this bundle's checks. The repository's broad `lake build` has separate Wave 5/6/7 syntax and duplicate-declaration failures. A passing assessment runner is not a full repository build or a proof of the physical claims.
+
+Companion notes: [fractals](https://github.com/Foundation-0/fractals/tree/aeva/research-value-assessment-20261112/research/value-assessment-20261112), [physics0](https://github.com/Foundation-0/physics0/tree/aeva/research-value-assessment-20261112/research/value-assessment-20261112), and [astr0](https://github.com/Foundation-0/astr0/tree/aeva/research-value-assessment-20261112/research/value-assessment-20261112).
+
+The earlier companions are [fractals PR 52](https://github.com/Foundation-0/fractals/pull/52), [physics0 PR 16](https://github.com/Foundation-0/physics0/pull/16), and [astr0 PR 5](https://github.com/Foundation-0/astr0/pull/5).
+
+## Independent review and publication changes
+
+The [independent scientific review](closure/independent-review.md) checks the Wave 12 arithmetic, future continuation, folded tensor and photon kernel. Its author wrote the separate closure route and did not author the reviewed Wave 12 or convergence probes. Its reviewed hashes identify the original research artifacts.
+
+The publication copy uses the halo-to-seed label “unsupported identification”, and the mechanics report cites pentamode prior art and distinguishes mixed-sign stress from isotropic pressure. The review retains its original pending status as a historical record.
+
+These publication edits are recorded by the final manifest. Human scientific review remains pending.
+
+The optional independent reviewer script requires SymPy in addition to the main dependencies. It is included with its saved checks for inspection and is separate from the four-probe CI job. Its optional versions are pinned in `requirements-review.txt`. Running `python -B research/value-assessment-20261112/closure/independent_checks.py` regenerates reviewer results and current artifact hashes, so use a disposable copy to preserve the historical review record.
+
+The [second independent review](second-review.md) passes the closure and observational-identifiability reports at their stated scope, with no blocking scientific findings. It supersedes the historical separate-agent-review entry in the closure source ledger. Its original reviewed hashes remain intact. Human scientific review and the proposed physical/data gates remain pending.
+
+Its [independent checks](second-review-checks.py) and [saved results](second-review-checks.json) use the optional reviewer dependencies. Run the script in a disposable copy to preserve the historical review hashes and evidence.