Public Replication And Review Surface¶
Status snapshot: June 2026 public rerun and review route.
This page is the public route for artifacts that can be inspected or rerun now. It sits between the compact evidence summary in What CCT Already Demonstrates, the evaluation rules in CCT Review Protocol, and the detailed package README inside the repository.
The main package is the repo-root cct-public-replication/ folder. Its first reader-facing file is cct-public-replication/README.md; its fuller reviewer guide is cct-public-replication/docs/public_replication_readme.md.
What Can Be Checked Now¶
The public package exposes method and branch artifacts as a reviewable route surface:
| Surface | What it lets a reviewer check | Current claim class |
|---|---|---|
| Baby-theorem verifiers | Command-attribution checks (BT3/BT5), scalar declared-envelope accounting (OP4/BT4), basin/path-measure diagnostics (BT6), and passive aperture/operator-norm routes (BT7b). | Bounded-model and proof-review hygiene. |
| OP3 command-attribution route classifier | Symbolic route grammar for memoryless and history-conditioned command attribution, hidden predictors, shared clocks, state sufficiency, joint capacity, side-channel ledgers, denominator scope, task-functional promotion, and future-leakage guards. | Formal command-attribution theorem companion. |
| Formal discriminators | Scalar multiwell anti-uniqueness / OP0a, QFT-data specificity-filter scaffold / OP0b, and regime-local RFH metrology envelope / OP1, including the accepted algebraic finite-window envelope and a machine-auditable Bridge V1 Fisher-certificate route companion. | Specificity-filter, metrology-envelope, and proof-review route behavior. |
| Basin and observation-control estimators | Basin/path-measure finite-state route tables (BT6), BT6 finite-support terminal/coarse KL concentration/perturbation routes, OP2 observation-to-control randomized holdout routing, OP2 exact conditional tau-grid randomization-inference routes, and shared BT6/OP2 method-ledger checks. | Narrow V1 theorem/method language and formal-review handoff. |
| Vector OP4 tools | Vector OP4 multi-resource accounting: scalarization, Pareto/baseline routing, finite resource-front diagnostics, supported/unsupported front membership, baseline win regions, tau support/win intersection checks, hidden-resource and denominator diagnostics. | Accepted finite/discrete resource-front discriminator in the V1 scope. |
| Reference Stack v1 | Schema validation, manifests, nulls, denominators, baselines, and promotion gates. | Public method grammar. |
| Synthetic capsules | Observer-mode, measurement-band, field-control basin, material-control, timing, calibration, environmental-handle, effective-adjacency, and passive-boundary support-reducer objects. | Branch narrowing, preregistration support, and route-surface confirmation. |
| Tau-X templates | Payload cards, demand ledgers, reconstruction ledgers, fallback routes, effective-adjacency rows, H2B-C2 broad-group support-reducer route cases, and the C3 evidence-bearing heldout-witness next object. | Mission architecture and resource-ledger translation. |
These artifacts make claims checkable, routable, and reviewable. Engineering exposure, CCT interpretation, observer-conditioned interpretation, Tau-X architecture interpretation, and independent replication use later promotion gates.
Suggested First Reruns¶
From the repository root, start with the small public-safe checks:
python cct-public-replication/theorem_verifiers/verify_baby_theorems.py --suite batch1a
python cct-public-replication/theorem_verifiers/verify_baby_theorems.py --suite bt6
python cct-public-replication/estimators/op3_route_classifier.py --cases cct-public-replication/examples/op3_route_classifier_cases.csv
python cct-public-replication/estimators/bt6_finite_state_discriminator.py cct-public-replication/examples/bt6_finite_state_discriminator_cases.csv
python cct-public-replication/estimators/bt6_op2_finite_sample_intervals.py bt6 cct-public-replication/examples/bt6_finite_sample_interval_cases.csv
python cct-public-replication/estimators/bt6_op2_finite_sample_intervals.py op2 cct-public-replication/examples/op2_holdout_delta_interval_cases.csv
python cct-public-replication/estimators/bt6_op2_method_ledger.py cct-public-replication/examples/bt6_op2_method_ledger_cases.csv
python cct-public-replication/estimators/op2_randomization_inference_interval.py cct-public-replication/examples/op2_randomization_inference_interval_cases.csv
python cct-public-replication/estimators/bt6_kl_perturbation_concentration.py cct-public-replication/examples/bt6_kl_perturbation_concentration_cases.csv
python cct-public-replication/theorem_verifiers/verify_aperture_focusing.py
python cct-public-replication/theorem_verifiers/verify_op0a_multiwell.py
python cct-public-replication/theorem_verifiers/verify_op0b_qftdata_filter_examples.py
python cct-public-replication/theorem_verifiers/verify_op1_rfh_regime.py
python cct-public-replication/theorem_verifiers/verify_op1_rfh_manifest_examples.py
python cct-public-replication/theorem_verifiers/verify_op1_bridge_v1.py
python cct-public-replication/estimators/vector_op4_resource_front.py cct-public-replication/examples/vector_op4_resource_front_cases.csv --check-expected
python cct-public-replication/estimators/vector_op4_resource_front.py cct-public-replication/examples/vector_op4_resource_front_tau_cases.csv --win-threshold 1.2 --check-expected
python cct-public-replication/tools/validate_manifest.py cct-public-replication/examples/reference_stack_cases/*.json
python cct-public-replication/tools/energy_sensitivity.py cct-public-replication/examples/reference_stack_cases/positive_manifest.jsonThe package README contains the fuller run order for synthetic capsules, manifest generation, OP3 command-attribution route classification, the Vector OP4 multi-resource simulator and finite resource-front verifier, OP2 observation-to-control randomized holdout, BT6/OP2 V1 interval-method objects, and toy-world smoke tests.
How Review Records Fit¶
Review packets and external-review files are provenance records. They are useful when a reviewer needs to see why a route label, theorem status, or packet status changed, but they are not the best first public reading surface.
Use this order:
- read What CCT Already Demonstrates;
- run or inspect
cct-public-replication/; - use the packet README and run commands for a focused review packet;
- consult raw Pro, heavy-model, or external-review records only when provenance matters.
When a review packet is included in a zip, the packet's 00_README_FOR_REVIEWER.md, 00_README_FOR_PRO.md, REVIEW_PROMPT.md, or PACKET_RUN_COMMANDS.md is authoritative for that packet. The full public replication README gives context, but the packet controls the included file set.
Public / Protected Layers¶
The public replication surface exposes method behavior, route labels, manifest grammar, baselines, nulls, denominator checks, and promotion gates. That is the audit layer: it lets reviewers inspect how a claim is routed and what would close, narrow, or promote it.
Protected lab protocol records carry build-specific implementation detail: optical layouts, field/control geometries, material recipes, drive searches, calibration tolerances, hardware constants, protected notebooks, internal value registers, and partner-specific build details.
This split keeps public review focused on route discipline while the lab track carries execution-specific protocol detail. Public artifacts show method discipline, route behavior, ledgers, closure paths, and review status; protected annexes carry the implementation details that belong to controlled bench execution.
Failed or negative results use the same route discipline. Public branches report route decisions such as no_go, narrow, baseline_wins, or defer, and publish public-safe negative-result summaries when a public branch changes status. Protected lab records carry raw failed-search maps, failed recipes, near-miss windows, hardware settings, and calibration records.
Hardware Exposure¶
Hardware is the physical exposure layer in the same stack. The current public artifacts prepare hardware work by making the question precise enough to run:
- formal targets define what a result is allowed to mean;
- simulations and route tables identify which branch is ready for physical exposure;
- ledgers and nulls declare the denominator, incumbent, and failure routes;
- preregistration locks the decision rule before bench interpretation;
- physical benches test whether selected regimes survive real instruments, materials, drift, noise, and replication.
That is the role of the public package: it prepares hardware-facing claims to be interpretable, comparable, and narrowable.