Skip to content

Research: current-main minimal in-band Contract C conformance RC1 - #72

Draft
camerontjs-dot wants to merge 10 commits into
mainfrom
research/contract-c-minimal-in-band-conformance-rc1-20260912
Draft

camerontjs-dot wants to merge 10 commits into
mainfrom
research/contract-c-minimal-in-band-conformance-rc1-20260912

Conversation

@camerontjs-dot

@camerontjs-dot camerontjs-dot commented Sep 12, 2026

Copy link
Copy Markdown
Owner

Classification

Draft Research Infrastructure / current-main supervisor-context conformance. Do not merge. This is not a clean-room independent Consumer B reproduction and does not authorize Contract C/D mutation, production Decision changes, release, tag, promotion, Authorization, or execution.

Terminal status

Scientific disposition: SUPPORTED_CURRENT_MAIN_MINIMAL_IN_BAND_CONFORMANCE

Programme gate: BLOCKED_CONTEXT_FREE_REPRODUCTION_REQUIRED

Terminal record:

  • path: research/contract-c-minimal-in-band-conformance-rc1/TERMINAL_RECORD.md
  • terminal record commit: 3227726e8aee0eefa637fc65a756ca90edd6c844

Exact base and immutable maintained surface

Protected Decision Engine main at launch:
7be709b2141c767c5da89b8b94cf90233c4238fe

Maintained V1 blobs remained unchanged:

  • Contract C ingress: f57a8067dadc04afb459f1d0342b2b786ec775e6
  • supported-claim policy: 2225f73eb6eefd83609f0ba19e4786d1267dd527
  • Decision materializer: 1562fb29da6679a0cf894e478cbdd4ae16e21a18

Upstream frozen research authority

  • Apparatus surface-necessity PR #89 terminal commit: c5fa9be39e6d521fdad8fd4c9cbb1d3cd10cc53b
  • Apparatus decisive run: 34695273270
  • Apparatus artifact: 10298184921
  • digest: sha256:2f3e713524fb978f9ab5b7e51c0402b276f57b56f1c6b5e59ba10b7256d22fae
  • frozen in-band handoff authority: Apparatus PR Research: qualify frozen Decision core against parent-bound Contract C #85 receipt ad1ffbd7906a7cf34cce5afa906a5797cd4a14ff

Exact copied handoff bytes:

  • valid-shadow.json: blob 14e88cbc691f7eba9366b4bf88611ef834637f27
  • contract-b-index.json: blob 4de40713482a1fc5a075a230a61e19acc25afbd9
  • object SHA-256: sha256:325962ebcdbf6af836bb6193a451524ccd40b4d10f2394ff9f703fbfce1ec1e3
  • result-set identity: result-set:4483272c4f6fbd9cb2362be7e3174bbd00aff3cf761d6c374897f3478818c9f0

No old PR #67 consumer implementation, Apparatus producer/sidecar code, CAL semantic code, or hidden evaluator logic was copied into this consumer.

Preregistration and frozen expectation order

  • preregistration commit: 26d096b007f14bdce31793f7d9e186ee4feeaa4b
  • frozen expectations committed before consumer implementation
  • decisive scientific head: 3ad518b9315d4767eae3d2a46f14e09975949479

The research consumer was restricted to existing Contract C structures plus the candidate non_deciding contribution channel. It did not require sidecar-specific member IDs, state IDs, receipts, duplicated carrier binding, or a second attribution object.

Preserved first-run deviation

Initial run 34695672580, job 103558617719 failed closed on the missing-external-profile test with the wrong error class: whole_object_mismatch was emitted before the intended missing-authority-context error.

No object was accepted and no profile self-selection occurred. A narrow correction reordered validation to validate the external profile object before exact-object digest comparison. No handoff, expectation, semantic, policy, hostile mutation, or maintained source changed.

Correction/decisive head:
3ad518b9315d4767eae3d2a46f14e09975949479

Decisive execution

  • run: 34695724968
  • job: 103558752853
  • result: PASS
  • artifact: 10297559181
  • artifact digest: sha256:609f32ce2fdcceae70710322342521a74dcb084d73918bbe3cef7f7e423f4c92

Observed from the exact frozen in-band object:

  • causal channels: [non_deciding, non_deciding]
  • causal evidence: [u-a, u-b]
  • causal form: independent_sufficient_alternatives
  • policy-equivalent outcome: HOLD / contract_c_proposition_not_checkable
  • same-members jointly_sufficient variant remained distinct
  • causal/residual neutral evidence remained distinguishable
  • residual-only neutral evidence remained reconstructable
  • no sidecar-only fields were required
  • no scalar/confidence/winner surface was required or invented

The deliberately unsafe causal non_deciding => CLEAR mutant was exercised and killed by the frozen expected HOLD outcome.

All 15 hostile controls were rejected or exposed as semantic changes, including wrong exact object/profile/Contract-B bindings, proposition/evidence substitution, missing/overlapping/unclassified participation, stale result identity, causal cardinality errors, neutral->support/counterevidence laundering, and dropped neutral evidence.

Maintained controls on the same exact head

  • engine invariant sweep: 81/81
  • output-quality sweep: 56/56
  • combination matrix: 6309/6309
  • Gate head: 14/14
  • Decision materializer: 4/4
  • Contract C seam fixtures: 165 checks across 8 fixtures
  • maintained released Contract C 1.0 integration: PASS
  • released supported control: CLEAR / contract_c_supported
  • non-supported control: HOLD
  • wrong version: rejected

The maintained released-1.0 ingress still rejects the research successor, preserving the version boundary.

Bounded inference

The minimized in-band invariant set survives current protected Decision V1 conformance while ordinary released behavior remains unchanged. This strengthens the evidence that a Contract C successor can reuse existing contribution/evidence/basis/residual/causal-form machinery and add the missing neutral semantic category in-band, rather than making the richer sidecar canonical.

Remaining gate

This implementation is supervisor-context conformance, not the project’s required context-free independent Consumer B reproduction.

The next justified action is one narrow context-free reproduction using only a frozen handoff/specification/expected-behavior aperture, without prior implementation/evaluator/research history.

No Contract C successor version, promotion, merge, release, tag, Authorization, or execution is authorized. Keep Draft.

Copy link
Copy Markdown
Owner Author

Remaining gate satisfied by fresh context-free reproduction

The BLOCKED_CONTEXT_FREE_REPRODUCTION_REQUIRED programme gate recorded in this PR has now been exercised on an isolated sparse aperture.

Fresh evidence:

  • research-scaffold-harness Draft PR Promotion: bounded Contract C 1.0.0 → Decision → Contract D 1.0.0 #29
  • aperture commit cb1d27ff00ae030093c5e97d78477cec7f20c6f4
  • fresh implementation/test freeze c1c31885f5802d4ea710dea7bcec214e977b9f03
  • receipt commit 07e60447fae17369be5676f013c187b9fd0bfd98
  • clean aperture receipt, network OFF, zero forbidden reads
  • prereveal tests 21/21 PASS

Frozen post-freeze evaluator:

  • evaluator commit 97cb968ac8793adb575640404730399a8798ad98
  • decisive run 34717144897
  • job 103616318296
  • artifact 10305635450
  • digest sha256:e0b1e6427f935e801ae41566907580fb0ef58fc7b32642783b61030b70e8fd47
  • disposition SUPPORTED_CONTEXT_FREE_CONSUMER_B_REPRODUCTION
  • failures []
  • baseline remained HOLD
  • weak non_deciding => CLEAR mutant killed

Therefore the specific context-free reproduction blocker recorded by this PR is satisfied within the bounded tested fragment.

This does not itself authorize a Contract C successor version, merge, promotion, release, tag, production Decision mutation, Authorization, or execution.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant