C.23 - MethodFamily Evidence & Maturity (Method‑SoS‑LOG)
Status: Stable
LOG (logic) for deductive shells for admissibility
SoS‑LOG means Science‑of‑Science LOG.
Registration boundary. A MethodFamily is registered by one exact G.5 registry row and registry edition. That record names the family, its admitted A.3.1 Method members and grouping basis, and intended selector use. Establish evidence, claim scope, validity, and the decision result under their respective rules.
C.23:1 - Problem frame
Use this when a G.5 selector must admit, restrict or refuse a registered MethodFamily for a declared TaskSignature and use. Apply executable rules to typed descriptors and the applicable evidence, retaining the branch and its grounds. Family maturity is evidence about that bounded admission use; an admission threshold belongs to a G.4 AcceptanceClause.
An ordinary application of a sufficient domain Method needs no MethodFamily admission record. Use C.11 for a local choice when no selector-facing family admission is required.
C.23:2 - Problem
Unstructured “readiness” stories and undisciplined evidence lead to:
- (i) Illicit scalarisation across mixed scale types,
- (ii) ambiguous gating conditions that a dispatcher cannot apply,
- (iii) reuse after the family, evidence profile, claim scope, qualification window, or comparison basis changed, or reliance on an unstated source-local, kind, or plane relation, and
- (iv) Immature families entering production.
We need a notation‑independent LOG layer whose executable rules use TaskSignature (S2) + EvidenceProfiles to return admit / degrade / abstain, retaining relation losses and any justified assurance consequences under R4.
C.23:3 - Forces
- Pluralism vs. dispatchability. Competing Traditions expose different invariants; selection must compare without semantic flattening.
- Maturity vs. opportunity. Open‑ended exploration (E/E‑LOG) must coexist with run‑safe exploitation; immature ≠ forbidden → provide safe degrade paths.
- Unknowns (tri‑state). Missing or
unknownvalues in live S2 fields must propagate to Degrade(mode) (including sandbox) or Abstain; no silent coercions. - Meaning and implementation. Keep the family, its description, rule and evaluation result distinct. A particular notation or software implementation must preserve the same admission conditions.
C.23:4 - Solution — Method‑SoS‑LOG: deductive shells over Eligibility & Evidence
C.23:4.1 - Objects & heads (LEX/I‑D‑S)
MethodFamily(registered in G.5) carries Eligibility and artefact identity.MaturityCard(this pattern) carries evidence‑aware maturity.SoS‑LOG.Rule(this pattern) is an executable rule schema.
Evaluating a rule returns one of {Admit | Degrade(mode) | Abstain} for a (TaskSignature, MethodFamily) pair.
A qualifying description episteme uses …Description; …Spec names that same episteme only after the E.10.D2 specification-use gate grants the named use.
C.23:4.2 - Rule schema (normative)
For each MethodFamily f, author an executable rule set:
LOG.Deduce_f(TaskSignature S2) → {Admit | Degrade(mode) | Abstain}
with the following branch obligations:
What the family claim covers. State which tasks the claim covers and what it promises about the registered members. Evidence that one member works on one task can support that use. A claim that every task has a suitable member allows the member to vary with the task. Promising one fixed member throughout requires evidence for that member throughout. Permitting arbitrary substitution requires the promised performance from each permitted member on each task where it may be used.
Identify the member, task conditions and performance criterion behind each result. Cover the combinations the claim needs through applicable trials, a warranted account of why the result carries over, or both. For a fixed set of tasks, report the observed proportion or the bound the outcomes establish. Eight confirmed successes among ten eligible tasks establish at least 80% for that set even while two outcomes are unknown. To estimate the rate for a future population, identify the population and explain why the available results carry over. For a sample estimate, give the selection method and its uncertainty. Even if every task in a sample succeeded, a guarantee for every future task needs grounds covering those future tasks. If successful members differ by task, an agent can be assigned to choose among them when the information and a supported way to choose will be available in time. It can still be known that a suitable member exists before it can be identified.
R0 — CG-Spec gate (precondition). For the exact G.5 registry row and MethodFamily, verify the cited CG-Spec.MinimalEvidence and EvidenceProfile for every CHR characteristic used by the family’s acceptance clauses and flows, under the declared claim scope and selected slices, qualification window, and intended selector use. Failure ⇒ Abstain with reasons. Publish the consulted CG-Spec, EvidenceProfile, registry, and policy editions.
Rationale: selector legality requires an explicit CG‑Spec gate. Publish associated ReferencePlane notes alongside the consulted ids.
R0.QD — QD pre-gates (when QD is used). Apply these gates when the TaskSignature enables a QD search or QD illumination use, including an open-ended generator that uses QD. Resolve that use and its fields under C.22:5.1. Retaining evidence under PortfolioMode=Archive or emitting a set of alternatives does not activate this branch. For the active QD use, verify:
- (i) CharacteristicSpaceRef resolves to the CHR-typed coordinates and dimension required by the declared descriptor and comparison rules, with ReferencePlane wherever interpretation or the receiving contract requires it;
- (ii) ArchiveConfig is lawful (topology, resolution, K>0,
InsertionPolicyRef,DistanceDefwith edition id and declared metric/pseudometric status); - (iii) EmitterPolicyRef present (with edition id);
- (iv) resolve DominanceRegime; if absent, use default= ParetoOnly.
Failure of any ⇒ Abstain with reasons.
R1 — Admit. Admit IFF
- (a) S2 satisfies Eligibility predicates of f (tri‑state aware),
- (b) the exact EvidenceProfile minima referenced by Acceptance/Flows for f are met for the declared claim scope and selected slices, qualification window, and intended selector use (post R0),
- (c) all relevant CAL.AcceptanceClauses (G.4) evaluate to true under lawful CHR comparisons,
- (d) any maturity gating (e.g., a floor on Maturity rungs) is expressed as an AcceptanceClause and referenced here by id (no acceptance thresholds inside LOG).
LOG never sets acceptance thresholds; its rules use and cite Acceptance verdicts.
R2 — Degrade. Apply R0 and, when applicable, R0.QD before any degrade branch, including U2, U3, and R7. Failure returns Abstain for the attempted use.
After those gates pass, return Degrade(mode) only through a declared family branch when an admitted unknown or an unmet acceptance condition permits a narrower scope or execution mode under the applicable CAL failure behavior; mode ∈ {scope-narrow | sandbox | probe-only}. An eligibility violation still requires R3 abstention.
Record the exact S2 unknowns or unmet conditions, narrowed claim scope or execution mode, qualification window, governing policy edition, and result. If the branch changes the intended use, re-evaluate R0 and eligibility for that bounded use before relying on it. LOG-Degrade never changes CHR scales or planes or turns a failed CAL verdict into a pass.
Note (CAL vs LOG). CAL‑level degrade.order (fall‑back to order‑only comparisons) is governed by G.4/CG‑Spec and is not a LOG mode. SoS‑LOG never overrides CAL outcomes; a LOG branch only narrows Scope(G) or execution mode (e.g., sandbox, probe‑only), it does not alter CHR scales or admissible orders.
probe-only MUST cite the applicable probe-budget and Acceptance-bound guards. Cite an E/E-LOG policy when that policy governs exploration over a still-live pool.
R3 — Abstain. If S2 violates Eligibility or R0 fails, return Abstain with the failed rule, reasons, and available policy, evidence-profile, scope, and qualification-window basis. Report missing references and unperformed judgements under §4.2.1. Abstain is mandatory for illegal CHR operations and when a conclusion depends on an F.9 Bridge, kind relation, or plane relation that has not been established.
R4 — Relation and loss routing. Cite an F.9 Bridge, kind relation, or plane relation only when the admission decision relies on that obtaining relation. Record its participants, direction, what meaning is preserved and what is lost, receiving use, and applicable policy edition.
When the admission use makes a separate named assurance claim, identify its exact target claim and receiving use under B.3.
Assess loss under the receiving-use rule. A numerical assurance penalty requires an applicable domain model under B.3:4.3–4.4 and affects that model’s declared assurance result. Expression properties, claim scope and admission conditions retain their own governing rules.
A changed registry row, evidence profile, claim scope, qualification window, or intended use is not by itself a crossing.
R5 — Proof hooks. Every branch MUST retain its A.10 evidence/source basis for the conclusion it makes. For an early Abstain, cite the available basis for the failed prerequisite and name the required evidence that could not be recovered.
Evidence used by a branch retains the lane tags (TA/VA/LA) and freshness windows required by its CG-Spec.MinimalEvidence and EvidenceProfile; missing requirements remain explicit gaps under §4.2.1.
Cite Bridge ids + loss notes when the branch relies on a Bridge; the decision is SCR‑visible. When G.6 EvidenceGraph is present, also publish EvidenceGraph path id(s) for the branch (admit/degrade/abstain). A branch verdict is not its own evidence basis.
R6 — Retention, source and selector outcome. Resolve PortfolioMode under G.5 CC-G5.23; it governs retention of exploration evidence. After Admit, a QD archive may supply candidates to G.5, but the archive remains a source. The emitted outcome follows G.5:4.4b–4.4e, with its result family when applicable, exact members, ordering and basis. Retaining an archive supplies neither a new result family nor a QD requirement.
When illumination telemetry is used, it remains report-only unless CAL authorises DominanceRegime=ParetoPlusIllumination and the policy id is recorded in SCR; coverage and regret likewise remain reported telemetry unless their governing policy promotes them.
R7 — GeneratorFamily branches (open‑ended). If S2 includes GeneratorIntent, SoS‑LOG MUST:
- (i) verify
EnvironmentValidityRegionis declared and lawful; - (ii) verify
TransferRulesRefexists; ifunknown⇒Degrade(scope‑narrow)orAbstainper family policy; - (iii) treat the selection surface as pairs
{environment, method}; publish coverage and regret under their declared telemetry policies. When the active branch or receiving use requires IlluminationSummary, obtain that retained-set report under C.17:5.1 with itsDiversity_Pmeasurement policy, Scale, descriptor or feature basis, and distance or covering definition. Generation alone requires no illumination report. These results remain report-only unless their governing policy promotes them under R6.
R8 — Telemetry and refresh hooks. When an active QD or open-ended generator branch records an illumination increase or changes its search archive, publish the current editions and actual edition increments of the definitions on which the changed use depends.
For QD, these include its characteristic space, descriptor map, distance, insertion and emitter policies under R0.QD and C.22.
For open-ended generation, retain TransferRulesRef.edition and the applicable validity and telemetry policy ids; add QD definitions only when that branch uses QD.
A separate illumination report retains the measurement basis required by C.17:5.1.
Ordinary retention of selection evidence follows its own retention and source references; it does not require QD definitions. Expose PathSliceId for refresh or decay in SCR only when an E.18 path slice is current.
Admit on the complete R1 basis. Degrade only through a declared narrower-use branch after its prerequisites pass. Abstain when a prerequisite fails or the policy supplies no permitted branch.
C.23:4.2.1 - Report the branch actually reached
An early Abstain completes the admission result for the attempted use. Keep the failed rule, reasons, known family and registry edition, TaskSignature, intended use, claim scope, qualification window, and available source and policy references. Name any required reference that could not be resolved and the available basis for that finding. The report can stop there without completing later judgements.
Retain each premise or result that was established or validly reused for this branch, with its source edition and the scope, window, evidence profile, and use that make it applicable. Reuse does not require recomputing the result. An earlier result whose applicability is unresolved remains unavailable as a premise for this use.
Distinguish these situations for each affected entry:
| Situation | What the report says | Result value |
|---|---|---|
| A required basis is missing or unavailable | Name the missing profile, reference, or evidence and why it could not support this use. | Leave the dependent judgement value unestablished. |
| A judgement was not evaluated | Name the judgement and the reason; use not reached after R0 or not reached after R3 when an earlier rule stopped evaluation. | No judgement result was established or reused for this entry. |
A live S2 value is the admitted unknown | Retain that value under its C.22 value rule and cite the family branch that handles it. | unknown remains the supplied value; U2/R2 govern the branch. |
An evaluated predicate or AcceptanceClause returned false | Cite the predicate or clause, its result, and the basis of that evaluation. | Retain the computed false; apply R3 or the declared CAL failure behavior as appropriate. |
These descriptions qualify report entries; they do not extend the S2 value sets, the closed maturity rungs, or the Acceptance verdict domain. An applicable earlier judgement used by the branch is reported as reused, not as unevaluated merely because it was not recomputed. A known result not used by this branch may be cited separately with that limited purpose.
First use. A registered family resolves, but the evidence profile required by R0 cannot be recovered. Report Abstain, the missing profile and available source basis, maturity not evaluated when no applicable judgement is available, and Acceptance not reached after R0. Do not insert L0 or false to fill those result positions. If R0 passes but Eligibility evaluates to false, report that predicate result and R3 Abstain; later Acceptance may remain not reached after R3. Keep a previously established maturity result if it is applicable and used.
Admit still requires the complete R1 evidence, Eligibility, and Acceptance basis. A declared Degrade(mode) retains the premises that selected its branch and any consumed maturity results. An unmet Acceptance condition stays unmet when its failure behavior permits a narrower use. R2 still requires R0 and eligibility to be checked for that changed use before reliance; a failed R0 or an Eligibility violation for the original attempted use remains Abstain.
C.23:4.3 - Maturity ladder (poset, not a scalar; Description, not Spec)
When a maturity judgement is established for an admission use, publish or cite its editioned MaturityCardDescription for the exact evaluated MethodFamily, G.5 registry edition, evidence profile, claim scope and selected slices, qualification window, and intended admission use (UTS enum ids; scale kind = ordinal; reference plane declared).
Here L0–L4 name maturity rungs for the evidence about a registered family. For example, maturity L2 (Replicated) identifies independent replication under the stated conditions. A model abstraction level L2 identifies what a disciplinary model specifies (B.5.MR:4.3). A replication can test a claim at any model abstraction level.
Cite an existing card when its judgement remains applicable. If no applicable judgement is available at an early stop, report that state under §4.2.1 without creating a card or assigning a rung.
Do not embed acceptance thresholds here; an admission floor remains a G.4 AcceptanceClause cited by R1.
- L0 — Anecdotal. Claims exist; lanes sparse; examples ad‑hoc.
- L1 — Worked‑Examples. Multiple worked examples with lane tags and Scope slices declared; no replication yet.
- L2 — Replicated. Independent replications identify their distinct bearers or operating conditions and declare the claim scope and selected slices, source and method editions, and qualification windows used; lane separation is observed and decay windows are explicit.
- L3 — Benchmark‑Severe. Repeated wins or parity on community baselines or severe tests. Declare any F.9 Bridge on which this comparison relies, with its loss notes.
Optional rung (for QD/OEE‑heavy families; ordinal, closed enum):
- L4 — QD‑Hardened. For an active QD use, establish archive stability under the declared InsertionPolicy/DistanceDef editions and reproducible IlluminationSummary improvements under controlled budgets. For an OEE generator, establish severe tests of its EnvironmentValidityRegion. An OEE use without QD does not acquire QD archive or illumination requirements; retain the exact branch and evidence behind the rung.
Norms.
M1. The ladder is lane‑aware (TA/VA/LA) and freshness‑aware; it is not a global numeric score. Declare Scale kind=ordinal and the closed enumeration of rungs; register the enum at UTS (twin labels; editioned).
M2. Transitions MUST be justified by EvidenceGraph paths (once G.6 is available) and published at UTS; missing anchors ⇒ no advance.
M3. A maturity floor used for admission belongs in a G.4 AcceptanceClause. Cite its policy edition, scope, qualification window and verdict. For a floor of “at least L2”, apply the order declared for those rungs under G.8:4.5. A larger numeral alone cannot establish precedence. If the policy requires independent replication, check the evidence for replication or an applicable rule that entails it. Repeated wins by one group on severe benchmarks can support a benchmark claim while leaving independent replication unestablished.
M4. Declare the MaturityCard reference plane. If an admission decision relies on a relation to another plane, cite that exact obtaining plane relation, its direction and loss, and the applicable policy edition; any loss assessment or assurance penalty follows R4.
Rationale note. Treating maturity as a poset aligns with B.3’s requirement for lawful comparisons and avoids scalarisation across ordinal/ratio scales; assurance calculations follow R4; maturity and admission retain their separate rules.
C.23:4.4 - Unknowns & Shift classes (tri‑state discipline)
U1. (LEX). Enumerations for Degrade(mode) and Maturity rungs MUST be declared as closed value sets and registered at UTS (twin labels). Lexical SD (E.10) applies.
U2. A live S2 characteristic or predicate admits unknown only when its C.22 value rule permits it; unknown MUST map to a branch (Degrade or Abstain) declared on the family (no coercions). Each branch publishes a branch‑id and (where used) a mode from a closed enum registered at UTS.
U3. ShiftClass semantics follow C.22. For a known ShiftClass ∈ {covariate‑shift, concept‑drift, adversarial}, the default is Degrade(scope‑narrow) unless a CAL.AcceptanceClause explicitly guards the regime. A live unknown follows U2 and its declared Degrade or Abstain branch. All these branches remain subject to R0, applicable R0.QD and R2/R3.
C.23:4.5 - Publication & wiring
W1. Register the SoS-LOG rule ids. Publish or cite a MaturityCardDescription for an established maturity judgement under §4.3; use §4.2.1 when an early stop leaves that judgement unavailable. RSCR tests cover Admit, Degrade, Abstain, and unknown paths, including early reports with missing bases or unperformed judgements. Relation and loss-policy ids appear only where a branch relies on them.
W2. Admissibility Ledger. Publish an editioned AdmissibilityLedger. Each selector-facing row identifies the exact MethodFamilyId, G.5 registry edition, TaskSignature, RuleId and rule edition, intended admission use, claim scope, qualification window, decisive branch, and decision result.
Record its MaturityRung, EvidenceProfile, AcceptanceClause and policy references, verdicts, evidence paths, DominanceRegime, and PortfolioMode according to §4.2.1: retain established or applicable reused values and explain any required but unresolved reference or unperformed judgement. An explanation of a missing or unperformed result accompanies its unfilled value position; it is not a substitute verdict or rung.
Include obtaining relation and loss-policy ids only when used, and G.6 path ids under R5’s condition. UTS registers the row vocabulary; the ledger records the admission result and its basis.
W3. Strategy composition. For a selection composition called a strategy, cite its governing G.5 rule. Cite an E/E-LOG policy when the composition uses that policy over a still-live pool.
W4. The dispatcher (G.5) consumes the rules and reports each decision with its conditions and grounds. To assign a job, identify an applicable member or an authorized way for the performer to choose one in time. Check the required result, means, authority and execution conditions under A.15. A probe-only decision supports the specified probe within its budget and guards.
C.23:5 - Archetypal Grounding
The following are constructed branch examples. For each later branch, stipulate that the named family and policy editions resolve and R0 passes; the QD case also passes R0.QD. Every Admit below additionally requires the complete R1 Eligibility, evidence and Acceptance basis. If a stipulated prerequisite is missing, use the early-Abstain result in §4.2.1 instead.
Continuous dynamics (ODE task).
S2 excerpt. DataShape=ODE; stiff?=unknown; Hamiltonian=unknown; Size≈10^3; Objective={↓error@ratio, ↑throughput@ratio}; Jacobian_sparsity=high; Missingness=MAR. A declared safety predicate tests the required ordinal safety level.
Families. Implicit‑BDF vs Explicit‑RK vs Symplectic.
Rules.
Implicit‑BDF: withJacobian_sparsity=high, its declared unknown-handling branch permitsDegrade(probe-only)under exploration policyEE-ODE-1and Acceptance-bound guardsProbe-ODE-1. An applicable L3 maturity judgement remains cited. If later observations establishstiff?=trueandHamiltonian=false, and the complete R1 basis for the intended use holds, that new input can receiveAdmit; the original unknown is not silently converted to true.Explicit‑RK: requiresstiff?=falsefor ordinary use. Its declared unknown branch permitsDegrade(sandbox)after R0 and eligibility are checked for that bounded use.Symplectic: requiresHamiltonian=true; its declared branch for the liveunknownreturnsAbstain, without reporting a false Hamiltonian judgement.
Didactic anchor. This mirrors C.22’s typed‑signature discipline and CHR legality (no ordinal means; unit alignment for ratio).
Rackauckas and Nie (2017) describe DifferentialEquations.jl as a common problem-and-solve interface for different differential-equation methods. That interface illustrates how task description can remain stable while solver choice varies. The C.23 admission conditions above are FPF’s conditions. (Journal of Open Research Software)
Planning/scheduling (MIP task).
S2 excerpt. DataShape=MIP; NoiseModel=deterministic; Objective={↓cost@ratio, ↑service_level@ordinal}; Size≈10^5 vars; convex_relaxation=available.
Families. MILP (branch‑and‑bound), Constraint‑Programming, Heuristic meta‑search.
Rules.
MILP: Eligibility requiresconvex_relaxation=available; the cited L3 MaturityCard edition names the registered family, evidence profile, benchmark basis, claim scope, qualification window, and intended selector use ⇒Admit.Constraint‑Programming: MaturityCard=L2; the ordinalservice_level≥BAcceptance predicate is met. Baseline parity is an additional CAL condition, not a missing R0 minimum in this example. Its declared failure behavior permitsDegrade(scope-narrow)while parity is unresolved, after R0 and eligibility pass for the narrowed use. If that parity basis were required by R0, the result would beAbstain.Heuristic meta‑search: MaturityCard=L1. The applicable AcceptanceClause and family failure policy decide between a permittedDegrade(sandbox)branch andAbstain; L1 alone supplies neither outcome.
Using these results. Hand the dispatcher the admission decision and permitted conditions for each family. A family admitted for a sandbox trial can enter that trial. Before selecting it for ordinary work, establish its eligibility for that use. A shortlist can retain alternatives eligible for the same intended use. To claim that a member is non-dominated, compare performance on cost and service under the stated order and common conditions (C.18:4.4). The admission results above provide no such comparison.
Dunning, Huchette and Lubin (2017) describe JuMP as a modeling language that separates model expression from optimization-solver choice. This supplies a concrete software example of preserving a problem description while considering different solvers. (SIAM Review)
QD archive (policy search).
S2 excerpt. PortfolioMode=Archive; CharacteristicSpaceRef(d=2); ArchiveConfig(CVT, res=1k cells, K=1, DistanceDefRef.edition=v2, InsertionPolicyRef=dyn‑elite); EmitterPolicyRef=v3; DominanceRegime=ParetoOnly.
Rules. After Admit, G.5 declares a Shortlist drawn from the QD archive, with exact members, ordering and basis under G.5:4.4b–4.4e. The archive remains its source. Illumination is reported separately; changes to DistanceDef or the emitter are editioned in SCR; dominance remains ParetoOnly.
Open‑ended GeneratorFamily (POET‑class).
S2 excerpt. GeneratorIntent{GeneratorFamilyRef=GF‑01, EnvironmentValidityRegion=EVR‑A, TransferRulesRef=TR‑A, CoverageMetric=…}; PortfolioMode=Archive.
Rules. After Admit, G.5 declares a Shortlist over eligible {environment, method} pairs with its source and selection basis; Degrade(scope-narrow) if TransferRules=unknown, as permitted by the family policy under R7. Telemetry publishes the declared coverage and regret with their edition and policy basis. This use requires no QD search or illumination report. If a later receiving use needs IlluminationSummary, obtain its C.17:5.1 measurement basis and retain the governing policy id on improvements.
C.23:6 - Bias‑Annotation
Maturity is evidence-relative: a rung established for one family, scope and use does not establish suitability for another. A declared bounded exploration branch can remain useful below an exploitation floor, while keeping its own prerequisites and limits.
C.23:7 - Conformance Checklist (normative)
| ID | Requirement | Purpose |
|---|---|---|
| CC-C23.1 | An established or reused maturity judgement SHALL be cited through an editioned MaturityCard naming the exact family and registry edition, evidence profile, claim scope, qualification window, intended use, rung justification, A.10 anchors, and freshness windows. An early stop with no applicable judgement SHALL report that state under §4.2.1 without inventing a card or rung. Cite a relation and loss note only when the admission claim relies on it. | Makes established maturity auditable and keeps an unperformed judgement distinct from L0. |
| CC-C23.2 | Each AdmissibilityLedger row MUST identify the exact MethodFamilyId and registry edition, S2, rule edition, intended use, claim scope, qualification window, decisive branch, and decision result. It MUST retain the consulted policy editions, established or applicable reused Eligibility, CG-Spec, evidence-profile and Acceptance results, and Γ-fold contributors where used; required missing bases and unperformed judgements are reported under §4.2.1. The A.10 source basis and any G.6 path ids follow R5; relation and loss-policy ids appear only when relied on. | Keeps the premises of each actual branch reconstructable, including early refusal. |
| CC‑C23.3 | Enumerations used by the rules (Degrade(mode); Maturity rungs) SHALL be closed and UTS‑registered (twin labels). | |
| CC‑C23.4 | Unknowns in live S2 fields SHALL map to Degrade(mode) (including sandbox) or Abstain with explicit branch‑ids; no unknown→0/false coercions. | Tri‑state discipline. |
| CC-C23.5 | If a branch relies on an F.9 Bridge, kind relation, or plane relation, it MUST cite that exact obtaining relation, direction, what meaning is preserved and what is lost, receiving use, and applicable loss policy; loss assessment and any assurance penalty follow R4. A changed family, evidence profile, claim scope, qualification window, or use is not by itself a crossing. | Keeps relation, loss, assurance and admission claims separate. |
| CC‑C23.6 | No acceptance thresholds in CHR or Maturity; acceptance thresholds live only in AcceptanceClauses (G.4). | Separation of concerns. |
| CC‑C23.7 | MaturityCard SHALL NOT be turned into a global scalar; treat as poset; any ordering MUST be lawful over CHR types. | Forbids cross‑scale scalarisation. |
| CC‑C23.8 | Publish to UTS with twin labels. Run GateCrossing visibility checks only for a cited crossing of a selected E.18 transformation-flow structure. Require CrossingBundle attestation only when the named receiving use needs it; apply E.18/F.9/F.17/E.17 under their respective current uses. Apply LanePurity and Lexical SD (E.10); use GateChecks/GateProfile (A.21) only when a named gate decision is current. | Publication & crossing visibility. |
| CC‑C23.9 | All enumerations (e.g., Degrade(mode), Maturity rungs) SHALL declare a closed value set and Scale kind, and be registered at UTS. | Avoids lexical drift; lawful typing. |
| CC‑C23.10 | RSCR tests cover negative/refusal paths (illegal CHR ops; CG‑Spec gate fail; Bridge missing when relied on; Φ table/policy‑id missing when that penalty policy is used; Lexical SD violations (E.10)); ensure branch coverage (Admit/Degrade/Abstain, unknown). | |
| CC‑C23.11 | For an active QD use, R0.QD MUST pass: lawful CharacteristicSpaceRef (coordinates typed, dimension sufficient for the declared descriptor and comparison rules, planes declared where interpretation or the receiving contract requires them), ArchiveConfig (topology, resolution, K, InsertionPolicyRef, editioned DistanceDef) and EmitterPolicyRef. Evidence retention or set-valued selection alone does not activate this gate. | QD requirements apply to the actual QD use. |
| CC‑C23.12 | DominanceRegime SHALL default to ParetoOnly; switching to ParetoPlusIllumination MUST be authorised by CAL and cited by id in SCR. | Prevents implicit scalarisation. |
| CC‑C23.13 | Resolve PortfolioMode through G.5 CC-G5.23. An archive remains a source; G.5 outcomes follow G.5:4.4b–4.4e after the applicable admission result. When illumination telemetry is used, it remains report-only unless CAL authorises its dominance contribution. | Separates retention, QD construction and selector outcome. |
| CC‑C23.14 | If GeneratorIntent is present, R7 MUST verify EnvironmentValidityRegion and TransferRulesRef. G.5 declares its outcome over {environment, method} pairs under G.5:4.4b–4.4e; coverage and regret remain telemetry. | OEE legality and telemetry. |
| CC‑C23.15 | For the active QD or generator branch, illumination increases or search-archive changes retain the current definitions, actual edition increments and applicable policy ids under R8. Ordinary evidence retention does not activate QD telemetry fields. | Reproducibility and refresh of the actual use. |
C.23:8 - Consequences
- Explainable admission. Every Admit/Degrade/Abstain is backed by anchored evidence and explicit unknown handling (selector reports are SCR‑linked).
- Run‑safe pluralism. Multiple families can co‑exist with policy‑governed exploration (E/E‑LOG) and maturity‑aware gating.
- Cross-Tradition reuse. The account identifies the obtaining relation, its limitations, and any justified assurance penalty under R4.
C.23:9 - Rationale
For a named admission-assurance claim, B.3 requires each of F, G, and R that the use consumes to have a declared bearer, meaning, and scale. Claim scope and selected USM slices remain explicit; WLNK and Φ(CL) are used only under an applicable, calibrated aggregation or loss rule. Evidence-typed maturity rungs preserve the declared ordinal distinctions and admissible comparisons.
Separately, locally declared DesignRunTag values retain their temporal predicates and C.27 adequacy conditions; cite a GateCrossing only when the selected E.18 transformation-flow structure contains it.
C.23:10 - Relations
Builds on: G.5 (selector consumes these rules), G.4 (Acceptance & EvidenceProfiles), C.22 (S2 typing), C.18 NQD‑CAL, C.19 E/E‑LOG, B.3 (named assurance claims and declared aggregation, including WLNK only when applicable), A.10 (evidence basis) and E.10 (naming). G.6 supplies EvidenceGraph paths when used; E.18 applies when a selected structural crossing is current.
Publishes to: UTS (MaturityCards, rule ids), SCR/RSCR (branch coverage; parity hooks).
Constrains: G.8 (LOG Bundling must cite MaturityCards), G.9 (parity harness draws baselines per rung), G.11 (refresh windows per rung & decay), G.5 (Open‑Ended Family mode for GeneratorFamily).
Outcome. Admissibility logic for MethodFamilies combines LOG shells, the maturity poset, degrade modes, and publication requirements with CG‑Spec legality rules, CHR guard‑macros, and CAL acceptance mechanics.