A.19.SelectorMechanism - Selection Kernel
Type: Architectural (A) Status: Stable Normativity: Normative (unless marked informative) This pattern declares Select and SelectEligibility under A.6.1. Applications and any separately asserted Work use the conditions in §4.1.
A.19.SelectorMechanism:0 - At a glance — didactic, informative
- What it is: a universal set-returning selection kernel: it takes candidates, admissible comparison outcomes, and explicit criteria, and returns a selected set, not a forced single winner.
- Why it exists: to prevent three recurring failure modes: hidden thresholds, silent scalarization, and winner‑take‑all defaults under partial orders and uncertain evidence.
- Use this when: admitted candidates must be selected under explicit criteria after comparison. Identify the candidate universe and choice rule, recover the required comparisons and their conditions, then return the justified subset. Stop with comparison when the receiving question needs no selection; a selection result alone supplies no authorization to act.
- What this buys: the practitioner gets one selected-set value whose criteria, finite basis of upstream binary CPM applications, required comparison coverage, token provenance, scope, any governing A.19 predicate, plane, window, and policy bindings are explicit.
degradeandabstainremain eligibility values, not selected-set members or alternative result kinds. - First output: read the by-value candidate set bound to
SelectionSlot. Read the candidate universe, finite upstream CPM application basis, required pair coverage, derived comparison-token union, selection conditions, claim scope and context slices, reference plane, evaluation window, eligibility value, and evidence use from theSelectapplication and neighboring relations; they are not fields inside the selected set. - How it evolves: method semantics and SoTA algorithm families connect via
G.2packs and wiring modules; the kernel signature stays stable and teachable. - Suite stage:
select(ordering lives only inA.19.CHR:4.5andsuite_protocols; suite membership is a set inA.19.CHR:4.2). - Inputs (conceptual): admitted candidates; a finite by-value basis of upstream binary CPM applications, each with its pair, realized
GuardDecision, and ownComparisonResultSlotbinding when produced; the union of justified relation or poset tokens from those bindings; explicitCriteriaSlot,CNSpecSlot,CGSpecSlot; oneU.ClaimScopewith selected A.2.6U.ContextSlicemembers; the same A.19 predicate basis when one governs the comparisons or selection criteria; effective reference plane; explicit evaluation window; and optional TaskSignature and MinimalEvidence policy refs. - Output (conceptual): the by-value
SelectionSlotcandidate set. A singleton is allowed when the explicit criteria determine one candidate; a total order supplies comparison information but needs a choice rule. The output is not a decision log, guard value, result episteme, generic result relation, publication, or replay record. - Planned use: on a P2W path, an A.15.2 baseline selects editions and policies. A.15.3 and SlotFillingsPlanItem apply only to independently declared receiving positions under A.19.CHR:4.7.2. The Select application carries its effective arguments and SelectionSlot binding under §4.1. Dated selection Work and an A.10 provenance account retain their independent grounds.
- Transformation-flow use: when selection is represented by an E.18 node, that node cites this declaration. Planned refs use A.15.3 typed filling only for independently declared receiving positions; Select bindings remain governed by §4.1.
- Failure mode: tri‑state guard (
pass|degrade|abstain); missing or unknown evidence never coerces topass. - Mental model:
SelectEligibilitygates the step;Selectapplies explicit criteria to set‑valued comparison outcomes; the result is a selected set whose “single winner” behavior must be explicit.
A.19.SelectorMechanism:1 - Problem frame
FPF’s Characterization (CHR) suite treats selection as a mechanism boundary within the suite (authoritative membership: A.19.CHR:4.2).
Suite membership is a set; order has no semantics. Any intended ordering is expressed only via suite_protocols (A.19.CHR:4.5), under suite obligations (A.19.CHR:4.3).
Within the suite‑closed protocol, SelectorMechanism appears as the select stage (after admissible comparison; optional stages remain optional per suite_protocols). The kernel’s role is concept‑level and governed by CN‑Spec and CG‑Spec:
- consume admissible comparison outcomes without collapsing them into a hidden scalar,
- apply explicit criteria and policy references, and
- return a selected-set result under policy-bound defaults, with the
Selectapplication and its argument and result bindings recoverable; dated Work and evidence-provenance grounds enter when the account asserts them or the receiving use consumes them.
The kernel uses the CHR suite SlotKind lexicon (A.19.CHR:4.2.1) to prevent SlotKind drift across specializations and across SoTA wiring layers.
A.19.SelectorMechanism:2 - Problem
Engineering teams regularly need to make “a selection decision” under conditions that are normal in projects:
- comparisons are partial, multi‑criteria, or set‑valued,
- evidence is incomplete or policy‑gated, and
- different stakeholders ask for different “best” notions.
If selection is not a first‑class mechanism boundary with stable semantics, the same high‑risk drift happens repeatedly:
- Silent winner forcing: partial orders get collapsed to a single winner by ad‑hoc tie‑breakers or hidden weights.
- Hidden thresholds and constants: thresholds, weights, dominance regimes, and default
PortfolioModefields get smuggled into implementations and become invisible in discussion and audit. - Scalarization by convenience: set‑valued comparison outcomes get replaced by a scalar “score summary” that is treated as decision‑relevant without being declared as such.
- Evidence coercion: missing or unknown evidence gets treated as “good enough” (implicit pass) rather than yielding explicit
degradeorabstain. - Boundary erosion: selection performs comparison, scoring, aggregation, or publishing.
- Selection-boundary drift: a selected-set label is reused after candidate universe, finite upstream comparison-application basis or its required coverage, selection conditions, A.19 predicate, claim scope, selected context slices, reference plane, or evaluation window changed.
- Guard-output collapse:
degradeorabstainis treated as a selected-set member or as a generic selection result.
A.19.SelectorMechanism:3 - Forces
-
Set‑valued reality vs single‑winner convenience. Many admissible comparisons are partial orders. The kernel must preserve set‑valued semantics while still allowing single‑winner outcomes when requested by criteria.
-
Policy primacy vs method freedom. Criteria and defaults must be explicit and policy‑bound, while multiple method families and decision styles must remain add‑able without mutating the kernel.
-
No hidden thresholds vs usability pressure. Engineers often want “just pick one.” If the spec does not constrain this, hidden thresholds and tie‑breakers become de facto policy.
-
Evidence discipline vs delivery pressure. Under uncertainty, teams default to coercion (unknown → pass). The kernel must enforce tri‑state eligibility and fail‑closed discipline.
-
Replayability vs conceptual minimalism. The mechanism declaration stays small. Replay recovers the
Selectapplication, its argument andSelectionSlotbindings, effective editions, policies, candidates, and selected set. Dated selection Work and evidence-provenance retain their independent grounds when asserted by the account or consumed by the receiving use. -
Evolvability vs didactic usability. The kernel must be stable enough to support SoTA wiring and specialisation chains, but also teachable: one place states the mechanism boundary, laws, eligibility behavior, and the neighboring replay basis for realized use.
-
Planned use and gate/guard separation. On a P2W path, A.15.2 carries the intended baseline; A.15.3 typed filling is conditional on independently declared positions. Selection must not mutate into a gate pattern: no GateDecision or decision logs inside the mechanism boundary.
-
No competing defaults. If defaults exist for
PortfolioMode, dominance regime, or archive policy, cite their declared sources rather than re-declaring them in the kernel. -
Scope continuity vs legitimate reselection. Selection may narrow candidates or apply explicit policy, but it may not change the finite upstream comparison-application basis, its required pair coverage, any member’s predicate basis, claim scope, selected context slices, reference plane, or evaluation window. A justified change is a new selection application and may require new binary comparisons.
A.19.SelectorMechanism:4 - Solution
SelectorMechanism is the selection kernel for CHR and for selector specializations. It provides:
- a stable mechanism boundary for
select, - a stable SlotKind field set (via the CHR lexicon),
- a minimum law set that preserves set‑valued semantics and forbids hidden thresholds and hidden scalarization,
- a tri‑state admissibility guard that is fail‑closed under missing admissibility or evidence,
- a replay basis that separates effective occurrence bindings, the selected-set result, and supporting evidence from reusable selector semantics;
- a selection-use boundary that keeps candidate universe, the finite upstream comparison-application basis and required coverage, the derived token union, selection conditions, scope, predicate basis, plane, and window distinct; and
- output discipline:
SelectionSlotcontains only the selected candidate set, while eligibility, evidence use, provenance, currentness, result epistemes, and publications remain separate.
Method semantics and SoTA algorithm families do not live inside the kernel: they connect via G.2 SoTA packs and wiring modules, and via declared specializations subject to the local restrictions in CC‑A19SelectorMechanism‑10. A claimed refinement, conservative extension or equivalence uses its own comparison test in A.6.1 §4.8; the labels ⊑ and ⊑⁺ do not supply that test.
A.19.SelectorMechanism:4.1 - Operation declaration — normative core
-
Declaration boundary: this A.6.1 intension declares
SelectandSelectEligibility; it does not bind project-specific pins or create selection scope, dated work, an operation application, gate decision, selected-set episteme, evidence use, provenance path, currentness relation, or publication relation. Each neighboring object or relation uses its governor. -
Declaration identity: SelectorMechanism.IntensionRef cites this A.6.1 U.Mechanism episteme. The CHR select stage resolves to its local Select operation. A changed argument, selection law or guard requires explicit selection of the changed declaration; another realizer of the same declaration changes no suite member.
-
IntensionHeader:
id = SelectorMechanism,version = 1.0.0,status = stable. -
IntensionRef:
SelectorMechanism.IntensionRefdesignates thisU.Mechanismepisteme as the suite member named inA.19.CHR:4.2; it is not theEntityOfConcernRefof the declared operation family. -
Purpose: universal set‑returning selection kernel over candidates and criteria; defaults remain policy‑bound; no hidden thresholds.
-
Imports:
A.6.1:4.8 (declaration comparisons),A.6.1 (operation-local declarations and bindings),A.19.CN (CN-Spec comparison basis and use conditions),C.22 (TaskSignature as a policy-reference artifact when used),G.5 (selector conformance and default selection policy),G.0 (CG‑Spec admissibility and evidence gates),A.19.CHR:4.2.1 (CHR SlotKind Lexicon). -
EntityOfConcernRef: the selection operation family declared by
SelectandSelectEligibilityin this section. -
Effective reference scheme: the CHR suite reference scheme in which the A.19.CHR SlotKind lexicon, CN-Spec, CG-Spec, and any current TaskSignature tokens are interpreted.
-
Signature components:
- SubjectKind: candidate set, supplied through
CandidateSetSlotin bothSelectandSelectEligibility. An incomplete eligibility proposal can leave this argument unbound under the rule below. - RangedValueKind: pair of values
<admitted candidate set, relation or poset token set over the same candidate universe>, carried respectively byCandidateSetSlotandComparisonResultSlotin both operations. - Results are operation-local: the selected candidate set and the separate guard judgment.
- Input qualification: selection ranges over one admitted candidate set and the union of justified relation or poset tokens from a finite basis of binary CPM applications whose pair endpoints lie in that candidate set and whose coverage satisfies the explicit selection conditions, all in one
U.ClaimScope; selectedU.ContextSlicevalues are members of that scope under A.2.6 and do not create duplicate membership.
These are A.6.0 declaration components. They do not form another selector-content container, and they do not absorb candidate admission, comparison work, dated selection work, result, evidence-provenance, or replay relations.
- SubjectKind: candidate set, supplied through
Operation-local argument and result declarations
Each argument below is declared separately for Select and SelectEligibility, except the realized eligibility argument of Select itself. ByValue carries the stated value; a named Ref or ByRef resolves one value and edition. Cardinality is per application. The guard can assess an incomplete proposal with 0..1 of each required argument; missing inputs have no binding and prevent pass.
| Direction | Local designator | Meaning and ValueKind | Designation; cardinality |
|---|---|---|---|
| Argument | CandidateSetSlot | Set of admitted candidate values, each retaining its governing kind and identity | ByValue; 1 set |
| Argument | ComparisonResultSlot | Set of relation/poset tokens, the union of the comparisonBasis members’ own returned values | ByValue; 1 set |
| Argument | CriteriaSlot | Set of selection clauses and explicit tie-breakers used to determine the selected set; acceptance thresholds remain in their own declarations | ByValue; 1 set |
| Argument | TaskSignatureSlot | TaskSignature supplying selector policy defaults when used; it does not replace CN-Spec or CG-Spec | TaskSignatureRef; 0..1 |
| Argument | CNSpecSlot | CN-Spec supplying admission/acceptance and candidate-use conditions | CNSpecRef; 1 |
| Argument | CGSpecSlot | CG-Spec supplying selector admissibility and default evidence conditions | CGSpecRef; 1 |
| Argument | MinimalEvidenceSlot | MinimalEvidence override used instead of CGSpecSlot.MinimalEvidence | MinimalEvidenceRef; 0..1 |
| Argument | comparisonBasis | Finite set of binary CPM Compare application references, each with its left/right pair, realized eligibility and own output binding or explicit absence | ByValue set; 1 |
| Argument | requiredComparisons | Finite set of required binary comparisons derived from the candidates, criteria and effective policy, including direction/comparator distinctions that affect selection | ByValue set; 1 |
| Argument | tokenProvenance | Mapping from every consumed token to at least one producing CPM output binding in comparisonBasis | ByValue; 1 mapping |
| Argument | claimScope | U.ClaimScope for claims about comparability and selection under the specified conditions. CandidateSetSlot binds the candidates considered in this application. | ByRef; 1 |
| Argument | selectedSlices | Set of selected U.ContextSlice members of claimScope under A.2.6 | ByValue set of references; 1 set |
| Argument | characteristicPredicate | A.19 CharacteristicSpacePredicate basis shared as required with the relevant CPM members | ByValue; 0..1, explicitly absent when none governs the use |
| Argument | referenceScheme | EpistemeInterpretationScheme (C.2.1:4.1.2), governing this selection | ByRef through C.2.1 SchemeDesignation; 1 |
| Argument | referencePlane | CHR:ReferencePlane value qualifying selection | ByValue; 1 |
| Argument | evaluationTime | Selection-evaluation point or interval in the declared time basis | ByValue; 1 |
| Argument | selectorPolicy | Effective selection policy and resolved defaults, including candidate-level failure behavior when degrade is used | ByRef to the declared policy; 1, resolved through TaskSignature when it supplies that policy |
| Select argument | eligibility | GuardDecision determined by SelectEligibility for these same proposal arguments, used to admit this selection | ByValue; 1, with its producing guard-result binding recoverable |
| Select result | SelectionSlot | Set of candidate values selected under those criteria and policy | ByValue; 1 set on completed admitted selection, 0 on abstain |
| SelectEligibility result | GuardDecision | Judgment determined by the eligibility predicates: pass, degrade or abstain | ByValue; 1 on completed evaluation |
Resolve the scheme designation under C.2.1:4.2.6 to one rule collection, then check its C.2.1:4.1.2 range and the rules needed by these selection inputs. The reference token is not the bound scheme. An unresolved required interpretation cannot support a completed admitted selection; apply the guard and abstention rules below.
Membership alone supplies neither successful selection nor a cross-scheme correspondence. The application and result bindings below remain independently required.
For each argument row, its bindingPredicate holds when that application uses the resolved value for its stated purpose: candidates supply the selection universe, criteria/policy supply the choice rule, specifications/evidence requirements govern admission, and the basis, coverage and provenance supply the comparisons on which the choice relies. Scope, slices, predicate, scheme, plane and time bind only when they qualify that application. A nearby policy or copied comparison record does not supply those bindings.
The SelectionSlot bindingPredicate holds when that Select application returns the candidate subset determined by its bound criteria/policy from the justified comparison basis, under its declared pass/degrade conditions. The GuardDecision bindingPredicate holds when that SelectEligibility application returns its assessed judgment. An equal selected set or guard value stored elsewhere is insufficient.
A.6.1 governs each binding’s identity and continuous extent within the application; result binding begins at its return.
SlotIndex (derived projection). Project the designators, ValueKinds, designation and cardinalities above. Historical Slot names support CHR lookup. The other local names declare operation arguments, not new U-kinds or CHR SlotKinds; A.6.5 relation SlotSpecs supply none of their semantics.
OperationAlgebra. The select stage resolves to Select with the declared candidate/token, criteria/specification and selection-use arguments, returning SelectionSlot. SelectEligibility evaluates that proposal under the guard below. ComparisonResultSlot contains only the token union; application references, coverage, scope and provenance remain separate arguments. A CPM abstain with no output contributes no token. Every required comparison must be discharged by a basis member; degraded selection names excluded candidates under the bound failure behavior and retains complete coverage for its reduced use.
ApplicationPredicate. Select obtains when a selection act applies the bound criteria and effective policy to the candidate universe using the bound justified CPM outputs and use restrictions, with pass or an explicitly permitted degrade outcome under SelectEligibility. It returns the resulting set; on abstain no Select act proceeds. SelectEligibility obtains when an evaluation checks that proposal’s coverage, provenance, admission and evidence conditions and returns the corresponding judgment. A valid proposal or an equal saved selected set does not establish that selection act.
ApplicationIdentityRule. One application is one selection invocation or guard-evaluation invocation at its calculation locus, from taking up its declared arguments until return or termination. References to that same episode reidentify it. Two independently begun invocations remain distinct even when the complete CPM basis, criteria, policies, qualification window and results are equal. Law 6 additionally makes a changed binding a new selection application; a completed prior application cannot acquire the replacement basis or result.
ApplicationExtentRule. Select extends from use of its candidates, CPM basis and choice rules through selected-set return or termination; SelectEligibility extends from proposal assessment through judgment or termination. An unfinished invocation has an open extent and no unreturned result binding.
The selection-evaluation window qualifies the use and does not determine these calculation intervals. A trace designates an established episode; ordinary set-selection mathematics asserts no dated U.Work.
For example, use the admitted incomparable supplier pair from A.19.CPM §4.1 and the declared criterion “retain every nondominated candidate.” With complete required coverage and the CPM token binding, two separately performed selections can each return {A,B}. Their Select occurrences and result bindings differ even when they reuse the same CPM producer application. A copied {A,B} without a corresponding selection return establishes no fresh binding. A copied comparison token whose producing CPM return is unestablished fails tokenProvenance and causes abstain; it cannot enter the basis merely because its text matches.
-
LawSet (minimum): the selection kernel is set-returning and policy-bound
- Set‑returning by default: a conformant
SelectMUST return a declared selected set by default. It MUST NOT silently collapse partial orders or incomparabilities to a single winner; a singleton outcome MUST follow from the explicit criteria applied to the admitted comparison outcomes. - No hidden thresholds or constants: a conformant publication MUST NOT smuggle thresholds, weights, dominance rules, or tie‑breakers. Selection‑level commitments MUST be explicit in
CriteriaSlotand, where needed, in policy defaults exposed throughTaskSignatureSlot. Admissibility and acceptance thresholds are applied only viaSelectEligibilityusingCNSpecSlot.acceptanceand the effective evidence policy (MinimalEvidenceSlot?orCGSpecSlot.MinimalEvidence). - No hidden scalarization or token aggregation by assertion: a conformant publication MUST consume
ComparisonResultSlotas the union of the finite basis members’ justified set-valued or partial outputs. Every consumed token MUST be traceable to at least one producing CPM application. Scalar summaries or relation tokens inferred from a missing pair, empty output,degrade, orabstainare forbidden; scalar summaries, if produced at all, are report-only unless explicitly promoted by policy outside suite closure. - Evidence gating is explicit: when selection depends on evidence, it MUST cite either
MinimalEvidenceSlotor the effectiveCGSpecSlot.MinimalEvidencepolicy and evaluate selection with the tri-state predicate. Candidate-level ineligibility handling MUST be explicit in current criteria or upstream results and bound by the selection application; the kernel MUST NOT invent evidence thresholds. - No competing defaults: effective
PortfolioMode, dominance regime, and other defaults come from declared policy refs and are bound by the application. - Binding changes require a new selection:
Selectdoes not change candidate universe, comparison-application basis membership, required comparison coverage, any member’s pair, eligibility or output binding, selection conditions, A.19 predicate basis, claim scope, selected context slices, reference scheme or plane, or evaluation window. A changed binding is another selection application and may require new binary comparisons. - Guard-output separation:
GuardDecisionis not a selected-set member. Onabstain, noSelectionSlotvalue is fabricated. Adegradeeligibility value permits a reduced set only under the bound failure behavior and criteria.
- Set‑returning by default: a conformant
-
AdmissibilityConditions (tri-state guard; fail-closed on missing admissibility, comparison coverage, token provenance, or evidence)
SelectEligibility(CandidateSetSlot, ComparisonResultSlot, CriteriaSlot, CNSpecSlot, CGSpecSlot, TaskSignatureSlot?, MinimalEvidenceSlot?; selection-use bindings) → GuardDecision ∈ {pass|degrade|abstain}.passrequires: (i) every basis member’s pair lies insideCandidateSetSlot; (ii) the basis covers every binary comparison required by the candidate universe and explicit selection conditions; (iii) every consumed relation token traces to a member’s own output binding; (iv) explicit selection conditions and tie-breakers; (v) compatible claim scope, selected A.2.6 context slices, reference plane, and evaluation window across the basis and selection, together with any A.19 predicate governing the comparisons or selection criteria; (vi) coherent CN-Spec and CG-Spec editions; and (vii) satisfied admission, acceptance, and effective MinimalEvidence predicates under their owners.- If
MinimalEvidenceSlotis absent,SelectEligibilityMUST evaluate evidence againstCGSpecSlot.MinimalEvidence, and missing or unknown evidence MUST NOT yieldpass. - A basis member with
GuardDecision = degrademay support a reduced set only when a current selector policy names the candidate-level failure behavior and the remaining basis still covers the comparisons required for that reduced use. The selection application binds that policy and its own realized eligibility value. - A missing required comparison, untraceable token, or required basis member with
GuardDecision = abstainmakesSelectEligibility = abstain; selection does not proceed and no selected-set output is created.
-
Applicability:
- Intended for the CHR
selectstage after the required finite set of admissible binary comparisons and produces a selected-set value. Selection remains distinct from comparison, acceptance, gate decision, publication, and telemetry. - Applicable only when
CNSpecSlot,CGSpecSlot, explicit criteria, the effective evidence policy, and a finite comparison-application basis with complete required coverage and token provenance are current for the candidate universe. Missing declarations or coverage fail closed. - Inside the CHR suite,
A.19.CHR:4.5alone determines stage ordering and optionality. - Every selection application binds one
U.ClaimScope, selected A.2.6U.ContextSlicemembers, the finite basis of binary CPM applications and their pair, eligibility, and output bindings, the derived token union, any A.19 predicate governing the comparisons or selection criteria, effective reference plane, selection conditions, and explicit evaluation point or interval. There is no implicit latest value and no default window inherited from the predicate or comparison label. - A selection across reference schemes or planes follows the relations the case needs. When it relates two F.17
SchemeSenseCellvalues from different semantic contexts, test the F.9BridgePredicateProfileand cite the Bridge only when its predicate obtains; state suitability for the named selection use in a separate C.2.1 claim. If that predicate is false or unresolved and the semantic crossing is required by the selection conditions,SelectEligibilitycannot bepass; follow the already declareddegradepolicy when applicable, otherwiseabstain.
When the selection crosses ReferencePlanes, cite the plane relation and policy. If both facts are current, state both under their own predicates. A cell or plane difference alone establishes neither relation, and one branch never fabricates the other. Neither relation supplies candidate universe, comparison-application basis or coverage, relation tokens, selection conditions, scope, predicate, or time.
- Intended for the CHR
-
Neighboring semantic-Bridge, bounded-use, and reference-plane relations:
When candidates or comparison tokens require interpretation across different semantic contexts, resolve the two F.17
SchemeSenseCellendpoints and test one F.9BridgePredicateProfile. For an obtaining Bridge, state its endpoints and profile separately. State suitability for the named selection use in a C.2.1 assertion whose EntityOfConcern is that Bridge and whose ClaimGraph designates<u,d,r,t>and polarity. IncludeCLor an observed-loss note only when the receiving use consumes it; permitted loss remainstin the bounded-use claim.For a ReferencePlane crossing, cite the plane relation and policy separately.
Open A.10 only when bounded reliance is current and B.3 only when a named assurance claim is current. If that assurance argument consumes a locally declared
R_effcalculation, cite its domain model and calculation; neither the Bridge norCLcreates a penalty.Adding or changing any of these neighboring objects does not by itself change the selector declaration.
-
Neighboring dated work, operation application, result binding, and evidence relations:
The identified Select application binds the candidate set, finite comparison basis, required coverage, token union, selection-use arguments, policies and returned SelectionSlot. If the account also asserts dated selection U.Work, first establish its A.13 core: the performing System, local agential-role kind and membership criterion, and an obtaining assignment of its declared direct species for the relevant scope and time. A.15.1 independently admits that performance; F.6 applies afterward only for assignment-bound attribution. The Work identity and extent are not automatically those of one Select application.
When the account asserts them or the receiving use consumes them, A.2.4 governs evidence use with its own claim scope and relevance window, A.10 governs reliance and provenance, and G.11 governs source or assertion-edition currentness.
A durable selected-set episteme, when needed, is governed by C.2.1, and any current entity-identity inception claim by A.15.PROD. No universal work-result, comparison-result, or selection-result relation is presumed.
To replay the selection, recover:
- the candidate set and required binary comparisons; for every basis member, the CPM application, pair, realized
GuardDecision, and its own output binding or explicit absence; and the trace from every consumed token to its producing member; - one
U.ClaimScope, selected A.2.6 context slices, the optional A.19 predicate basis (explicitly absent when none governs), effective reference scheme and plane, and evaluation point or interval shared as required by the selection conditions; CNSpecRef.edition,CGSpecRef.edition, andTaskSignatureRef.editionwhen TaskSignature is used;- the effective MinimalEvidence policy, either the explicit override or
CGSpecSlot.MinimalEvidence; - the Selector’s realized
GuardDecisionand, fordegradeorabstain, the current failure-behavior policy; - the candidate-set value and derived union bound to the Selector’s
ComparisonResultSlotargument; - the effective criteria and selector-default refs; and
- the selected-set result; any current obtaining F.9 Bridge and its separate C.2.1 bounded-use claim; any optional
CLor observed-loss note consumed; and any ReferencePlane relation and policy. Recover A.10 reliance or B.3 assurance only when the selection use consumes it.
These neighboring objects support replay. The finite basis is a binding of the selection application, and none of them is selector-declaration content or a generic result container.
- the candidate set and required binary comparisons; for every basis member, the CPM application, pair, realized
A.19.SelectorMechanism:4.2 - Boundary and layering rules
-
Selection conditions are explicit values, not a new object kind. The application binds
CriteriaSlotplus effective selector-policy refs, defaults, anddegradefailure behavior. Acceptance and admission predicates remain separate.SelectionSlotcontains only the resulting candidate set; eligibility, conditions, scope, evidence, and replay metadata stay outside it. -
Selection consumes a traceable finite basis of upstream CHR products; it does not invent them. Any indicator-derived scoring or comparison input retains its A.19 basis and the UINDM positions selected under that declaration. A changed basis requires rechecking affected upstream position resolutions and dependent results, preserving reuse where their grounds still hold; the selector cannot carry old naked indices into the new declaration.
The selection application binds binary CPM applications separately and supplies
ComparisonResultSlotonly as the union of their justified outputs. The kernel MUST NOT perform normalization (UNM), indicatorization (UINDM), scoring (USCM), folding (ULSAM), comparison (CPM), batch-result fabrication, or missing-pair completion insideSelect. If a scalar “overall score” is desired, it must be declared upstream as an admissible scoring or comparator choice, not invented inside selection. -
Threshold discipline (acceptance is not selection). Acceptance and admission thresholds are not selection criteria: they remain in their governing declarations and are applied only through
SelectEligibility. Selection-level tie-breakers,PortfolioMode, and selected-set constraints may exist, but they MUST be explicit in current criteria or policy refs and bound by the selection application, never hidden as unnamed constants. -
Report‑only summaries inside suite closure. Any scalar summaries, illumination metrics, or auxiliary “why not chosen” telemetry are report‑only unless explicitly promoted by policy, and MUST NOT be used as hidden dominance rules (
A.19.CHR:4.3.3). Publication and telemetry are separately governed continuations outside suite closure.G.10governs shipping Part-G outputs as aSoTA-Pack;E.24.PUBgoverns a claim about episteme availability, form or carrier. -
Declare specializations under these restrictions. A proposed specialization of
SelectorMechanismmust retain these local restrictions:- SlotKind invariance for inherited operations,
- no new mandatory inputs to inherited
Select, - added capabilities appear as declared new operations or additional results under the local
⊑⁺notation.
For the refinement, conservative-extension or equivalence claim, apply its own
A.6.1 §4.8test, preserving the applicable operation, application and binding meanings. CC‑A19SelectorMechanism‑10 retains the predicate, endpoint facts and A.6.RCD gap branch. -
Planned use and actual binding. On a P2W path, an A.15.2 baseline selects TaskSignature, CG-Spec and evidence-policy editions. A.15.3 supplies planned filling only for an independently declared receiving position. Select arguments and returns are established under this declaration; the planned values remain available for a separately governed comparison with what occurred.
A.19.SelectorMechanism:5 - Archetypal Grounding — informative
A.19.SelectorMechanism:5.1 - Selection under partial comparisons
When comparisons are partial or set-valued, selection must not pretend there is a single best candidate by default. SelectorMechanism makes selection explicit, policy-bound, and replayable: it returns a set, including a singleton when the criteria determine one candidate.
A.19.SelectorMechanism:5.2 - Deployment options
Scenario. A platform team must pick a set of deployment options for a subsystem under multiple criteria: latency, cost, and regulatory risk. Comparisons are multi-criteria and do not induce a total order.
-
CandidateSetSlot = {OptionA, OptionB, OptionC}. -
CriteriaSlotrequires Pareto selection over the three unordered pairs{A,B},{A,C}, and{B,C}, returns all non-dominated admissible candidates, and preserves the full selected set unless an explicit current criterion requires a singleton. -
The finite upstream comparison-application basis covers all three required pairs:
Compare(OptionA, OptionB, ...)hasGuardDecision = pass: OptionB is strictly better than OptionA on latency, while OptionA is strictly better than OptionB on cost under the declared comparator. ItsComparisonResultSlotrecords those strict opposite wins, so neither dominates the other;Compare(OptionA, OptionC, ...)hasGuardDecision = degradebecause OptionC lacks the required risk attestation. In this constructed case, the admitted comparator and partial-evidence policy define the empty token set as the lawful output for this restricted use; its output binding contributes no relation token about OptionC; andCompare(OptionB, OptionC, ...)has the samedegradebasis and likewise contributes no relation token about OptionC.
The Selector’s
ComparisonResultSlotargument is the union of those justified member outputs, so its two tokens both trace to the{A,B}CPM application. No equality, worse-than, orabstaintoken is fabricated for OptionC. -
MinimalEvidenceSlot?is absent, so evidence is evaluated againstCGSpecSlot.MinimalEvidence. -
The selection application binds the three CPM applications and their pair, eligibility, and output bindings; the required-pair coverage and token trace; the deployment-option claim scope and selected regulatory
U.ContextSlicemembers; the same predicate basis or explicitnone; the reference plane and evaluation interval; and adegradepolicy that permits exclusion of OptionC.
Outcome.
- Under that bound
degradepolicy,SelectEligibilityreturnsdegrade, excludes OptionC without coercing unknown evidence, andSelectionSlotreturns{OptionA, OptionB}. - If either required comparison involving OptionC instead had
GuardDecision = abstain, that basis member would have no output binding,SelectEligibilitywould returnabstain, and no selected-set value would be created. Neither guard value is a member ofComparisonResultSlotorSelectionSlot. - The selection or eligibility application, finite CPM application basis, policies and returned value or explicit absence preserve the branch taken. An A.10 provenance account is added for the bounded reliance at issue; any dated Work claim requires the independent grounds in §4.1.
A.19.SelectorMechanism:5.3 - Analysis-method families
Scenario. A methods group selects a declared set of analysis-method families for a task. Candidates are refs to those families. The group wants diversity in the selected set, but does not want diversity metrics to silently become dominance criteria.
-
CandidateSetSlot={Family1, Family2, Family3, Family4} -
The selection conditions declare which binary method-family comparisons are required. A finite basis identifies every relied-on CPM application, its pair, eligibility value, and own output binding; the Selector’s
ComparisonResultSlotargument is their justified-token union. -
TaskSignatureSlotis present and is the single policy-default slot or ref:PortfolioModeand dominance regime,- budgeting and telemetry hooks (when used).
-
CriteriaSlotdeclares that diversity signals are telemetry unless explicitly promoted by policy.
Outcome.
SelectionSlotreturns a selected set of those families; any archive‑style behavior is a specialization and policy choice, not a hidden kernel default.- The
Selectapplication retains itsTaskSignatureRef.edition, effective criteria andSelectionSlotbindings for later explanation. Add provenance and dated Work only under their independent grounds in §4.1.
A.19.SelectorMechanism:6 - Bias-Annotation — informative
This pattern biases selection authoring toward explicitness and admissibility.
- Governance bias. Bias toward explicit criteria and policy-reference records rather than implicit constants. Risk: perceived overhead. Mitigation: use existing prose or code when it makes the criteria recoverable, and centralize defaults via
TaskSignatureSlotwhen used. - Architecture bias. Bias toward set‑return semantics and against forced total orders. Risk: consumers may expect a single winner. Mitigation: make the choice rule explicit; a declared total order alone does not choose a candidate.
- Epistemic bias. Bias toward fail‑closed evidence handling and against unknown coercion. Risk: more
degradeorabstainearly. Mitigation: improve evidence pins and policy clarity; do not relax the kernel. - Practice bias. Bias against embedding telemetry and publication into selection. Risk: teams want one step to select and report. Mitigation: keep those relations under their governing patterns; retain replay through selection bindings and the separately applicable Work, evidence and currentness relations in §4.1.
A.19.SelectorMechanism:7 - Conformance Checklist
| ID | Requirement |
|---|---|
| CC-A19SelectorMechanism-0 | Mechanism declaration completeness: one U.Mechanism episteme, its selection-operation-family EntityOfConcernRef, its effective reference scheme, the operation-local argument/result declarations, OperationAlgebra, LawSet, AdmissibilityConditions, and Applicability are recoverable under A.6.1. |
| CC‑A19SelectorMechanism‑1 | Single governing pattern: the SelectorMechanism U.Mechanism declaration is governed by A.19.SelectorMechanism:4.1; other descriptions cite this section rather than restating the kernel law. |
| CC‑A19SelectorMechanism‑2 | Set‑return default: a conformant Select MUST be set‑returning by default; it MUST NOT silently collapse partial orders or incomparabilities to a single winner. |
| CC‑A19SelectorMechanism‑3 | No hidden thresholds or constants: a conformant SelectorMechanism publication MUST NOT smuggle thresholds, weights, dominance rules, tie‑breakers, or default PortfolioMode fields. Selection‑level commitments MUST be explicit in CriteriaSlot and policy defaults when used (e.g., via TaskSignatureSlot). Acceptance thresholds remain governed by AcceptanceClauses, TaskSignature, or GateProfile records and MUST be applied only via SelectEligibility. |
| CC‑A19SelectorMechanism‑4 | No hidden scalarization: if ComparisonResultSlot is set‑valued or partial, a conformant publication MUST consume it as such; scalar summaries are report‑only unless explicitly promoted by policy outside suite closure. |
| CC-A19SelectorMechanism-5 | Evidence gating: SelectEligibility returns pass, degrade, or abstain; missing or unknown evidence never yields pass. Candidate exclusion or restricted use is explicit in the current criteria or policy bound by the selection application. |
| CC‑A19SelectorMechanism‑6 | SlotKind discipline: SlotKind tokens used in the SelectorMechanism intension MUST come from the CHR SlotKind lexicon (A.19.CHR:4.2.1). New SlotKinds require lexicon extension first. |
| CC-A19SelectorMechanism-7 | Bridge and reference-plane discipline: a semantic crossing cites an F.9 Bridge only between two F.17 SchemeSenseCell values when its profile applies and its predicate obtains; its C.2.1 bounded-use claim is separate and CL is optional. A ReferencePlane crossing cites its relation and policy separately. A scheme, cell, or plane difference alone establishes neither relation. A false or unresolved required Bridge predicate permits no pass; use only the degrade or abstain route already declared. A.10 enters only for current reliance, and B.3—including any locally defined R_eff calculation—only for a named assurance claim under its declared domain model and calculation. All remain outside selector-declaration content. |
| CC-A19SelectorMechanism-8 | Replay basis completeness: recover the Select or SelectEligibility application. Identify bound or returned values and required inputs that were missing: the candidate set, required binary comparisons, every upstream CPM application with pair, eligibility and own output binding or absence, token-to-producer trace, criteria and policy, U.ClaimScope, selected A.2.6 context slices, any governing A.19 predicate (or its explicit absence), reference scheme and plane, evaluation window, derived token union, guard judgment, and any returned SelectionSlot binding. An abstain judgment has no selection result. Independently recover dated Work through the A.13 core and A.15.1 admission in §4.1, evidence use under A.2.4, reliance and provenance under A.10, currentness under G.11 and a result episteme under C.2.1 only when asserted by the account or consumed by the receiving use. A Select application does not itself require separately admitted upstream comparison Work. The output values carry none of this metadata. |
| CC-A19SelectorMechanism-9 | Planned-filling separation: on a P2W path, A.15.2 carries intended editions/policies; A.15.3 typed filling applies only to independently declared positions. Select applications carry effective argument/result bindings; dated selection Work and any relied-on A.10 provenance remain separately established. |
| CC‑A19SelectorMechanism‑10 | Extension discipline: a proposed SelectorMechanism specialization MUST preserve inherited SlotKind designators and their meanings and add no mandatory input to inherited Select. For a claimed refinement, conservative extension or equivalence, apply the corresponding A.6.1 §4.8 preservation test, including application/binding predicates, identity and extent. The comparison predicate and endpoint facts must be established; use A.6.RCD’s missing-governor/substrate result when they are absent. The symbols ⊑ and ⊑⁺ alone establish none of these comparisons. |
| CC-A19SelectorMechanism-11 | Guard and gate separation: SelectorMechanism publishes neither GateDecision nor DecisionLog; SelectEligibility returns pass, degrade, or abstain separately from the selected set. |
| CC-A19SelectorMechanism-12 | Selection-condition completeness: CriteriaSlot, effective selector policies and defaults, and any degrade failure behavior are explicit and bound by the application; acceptance and admission predicates remain separate. |
| CC-A19SelectorMechanism-13 | Selection-scope completeness: every admitted Select application binds candidate universe, finite binary CPM application basis, required comparison coverage, token-to-producer trace, U.ClaimScope, selected A.2.6 context slices, any A.19 predicate governing the comparisons or selection criteria, effective reference scheme and plane, and explicit evaluation point or interval. No generic context input, optional structure, batch result, or label supplies them. |
| CC-A19SelectorMechanism-14 | Output separation: SelectionSlot contains only the by-value selected candidate set. On abstain no output is fabricated; a reduced set under degrade requires explicit current policy. |
| CC-A19SelectorMechanism-15 | Comparison continuity: selection cites the finite basis of binary CPM applications and may not change its membership, required coverage, member pairs, eligibility or output bindings, predicate basis, scope, selected slices, plane, or window. A justified change is a new application and may require recomparison. |
| CC-A19SelectorMechanism-16 | No generic result relation: the A.6.1 operation application binds SelectionSlot; C.2.1 governs a durable selected-set episteme when needed; subject patterns govern other result relations. |
| CC-A19SelectorMechanism-17 | Finite comparison-basis coverage: selection conditions derive the required binary comparisons; each is discharged by a CPM application, and every consumed relation token traces to its producing member output. A missing pair, untraceable token, or required member that abstains forces selector abstention rather than a fabricated batch result. |
A.19.SelectorMechanism:8 - Common Anti-Patterns and How to Avoid Them — informative
| Anti-pattern | What it looks like | Remedy |
|---|---|---|
| GateDecision leakage | Select emits GateDecision or writes a decision log | Keep gate decisions in their governing patterns. SelectEligibility remains the mechanism predicate; the application retains its realized eligibility value and replay basis. |
| Forced single winner | Select always returns exactly one candidate even under incomparability | Return a declared selected set. If a singleton is required, state the choice rule in CriteriaSlot and derive that result from the admitted comparison basis; any ordering used must be declared and admissible. |
| Hidden tie-breakers | “If incomparable, pick lower cost” without declaring that as policy | Move tie-breakers into explicit criteria or into declared comparator policies; never embed inside the kernel |
| Scalarization by convenience | Replace set-valued comparison with a scalar “summary score” treated as decisive | Keep summaries report-only unless declared as admissible comparator outputs |
| Unknown coerced to pass | Missing evidence treated as acceptable | Use tri-state SelectEligibility; unknown maps to degrade or abstain |
| Selection does comparison | Selection stage recomputes scoring or comparison internally | Keep comparisons upstream; SelectorMechanism binds CPM applications and consumes only their justified-token union in ComparisonResultSlot |
| One binary comparison treated as a batch | One Compare(left,right) application is said to cover three or more candidates, or a token union loses its producing applications | Bind a finite basis of binary CPM applications, derive required pair coverage from the selection conditions, and trace every consumed token to a member output |
| Selected set as replay record | Candidate universe, comparison ref, criteria, scope, evidence, or currentness are placed inside SelectionSlot | Keep SelectionSlot to selected candidates; bind use arguments on the application and use evidence, provenance, and currentness relations |
| Boundary drift | Selection reuses a token union after comparison-basis membership, coverage, member pair or eligibility, predicate, scope, selected slices, plane, or window changed | Treat it as another selection application and perform the required binary comparisons again when their governed basis changed |
| Publish inside selection | Selection emits a publication or telemetry relation as part of mechanism semantics | Keep publication and telemetry under their governing patterns; application bindings and applicable relations retain replay |
A.19.SelectorMechanism:9 - Consequences
Benefits
- Preserves incomparable candidates when the criteria require them.
- Makes thresholds, weights and scalarization choices inspectable; correct selection still depends on applying the admitted criteria to the justified inputs.
- Supports replay through application bindings and the relations that the receiving use consumes.
- Supports evolvability: new method families and selection styles can be wired without changing the kernel signature.
Costs and trade-offs
- Selected-set results can require downstream handling when a single decision is needed.
- Evidence discipline increases early
degradeorabstainuntil criteria and evidence policies are explicit. - Teams must make the selection criteria recoverable; existing descriptions may already suffice.
A.19.SelectorMechanism:10 - Rationale
Selection is where many systems accidentally convert admissible but nuanced information into an unjustified scalar decision. Making selection a separate, explicit mechanism boundary achieves two things that matter for engineering management:
- Technical integrity: the declared guard tests the admissibility and evidence conditions needed by this selection.
- Organizational clarity: it makes defaults and thresholds discussable, reviewable, and maintainable as explicit policy references.
The set‑returning default is not a preference for large retained sets; it is a correctness safeguard when the order is not total. Single-winner outcomes remain possible when the criteria select one candidate from the admitted comparison outcomes.
A.19.SelectorMechanism:11 - SoTA-Echoing
The research lines below supply application families and design comparisons. Their distinct output meanings and conditions remain with the cited methods; they do not establish the selector laws by themselves.
Selector-family SoTA packages are cited through their current Part G pack or claim sheet when one governs the use. They connect through CriteriaSlot and TaskSignatureSlot references while kernel semantics remain unchanged.
A.19.SelectorMechanism:11.1 - SoTA alignment map (normative)
| SoTA practice pointer, post‑2015+ | Primary source examples, post‑2015+ | Where it connects to SelectorMechanism | Adoption status |
|---|---|---|---|
| Retain several candidates under explicit criteria | Pugh, Soros and Stanley (2016), Quality Diversity: A New Frontier for Evolutionary Computation, §4.2 | QD supplies retained-set application families with different niche and retention rules. Declare the choice rule in each use; an archive rule is not the general Pareto-selection rule. | Adapt |
| Use declared archive-insertion criteria | Vassiliades, Chatzilygeroudis and Mouret, Using Centroidal Voronoi Tessellations to Scale Up the Multi-dimensional Archive of Phenotypic Elites Algorithm, author manuscript revised 2017 | CVT-MAP-Elites assigns a candidate to its nearest region and replaces a less-fit occupant. An FPF adaptation states the region and quality criteria; auxiliary telemetry supplies no undeclared dominance rule. | Adapt |
| Organize candidates as environment–solution pairs | Wang et al. (2019), Paired Open-Ended Trailblazer (POET) | POET pairs generated environments with optimized agents and transfers solutions between environments. Candidate and criteria structuring is an FPF adaptation; the source does not establish this kernel’s laws or FPF Method membership. | Adapt |
| Admit a declared rejection option | Geifman and El-Yaniv (2017), Selective Classification for Deep Neural Networks | The possibility of rejection informs FPF eligibility design. A selective-prediction rejection rule is not identical to this three-valued guard; apply each rule under its own conditions. | Adapt |
| Make architecture commitments traceable | ISO/IEC/IEEE 42010:2022 architecture description discipline | FPF assigns this declaration to one governing pattern; that assignment is an FPF design choice. | Adapt |
A.19.SelectorMechanism:12 - Relations
-
Builds on
A.6.1and its conformance checklist for mechanism identity, declaration content, applicability, and specialisation-chain discipline.A.19.CHRfor suite membership, suite protocol closure, SlotKind lexicon, and threshold and default discipline.G.0forCG‑Specadmissibility and evidence declarations.A.19for theCharacteristicSpacePredicatebasis when one governs selection.A.19.CPMfor every binaryCompareapplication in the finite basis, its pair, realized eligibility value, and own set-valued output binding.A.2.6forU.ClaimScopeidentity andU.ContextSlicemembership.A.19.CNforCN-Spec, whose declared comparison and candidate-use conditions supply an explicit input.C.22forTaskSignatureas a policy-reference artifact when used.A.6.1 §4.2for operation-local argument/result meanings and their SlotIndex projection; the local extension rule preserves inherited meanings.A.15.2for the edition/policy baseline on P2W paths;A.15.3plusA.19.CHR:4.7.2for typed filling of independently declared positions.C.27.TAfor the explicit selection-evaluation point or interval.A.2.4,A.10, andG.11for evidence-use scope, provenance, and currentness, separately from selection scope and output.
-
Used by
A.19.CHRas theselectstage in CHR pipelines.G.5as the primary conformance and specialization context for selector-based method dispatch andPortfolioModepolicies.E.18when this declaration is used by transformation-flow nodes. A.15.3 governs planned refs only for independently declared receiving positions; operation bindings and any dated selection Work retain their rules.
-
Coordinates with
CPMand other admissible comparison stages as producers of the result bindings whose justified-token union fills the Selector’sComparisonResultSlotargument.ULSAMand other admissible aggregation stages that must remain explicit rather than hidden inside selection.E.20for governing-pattern discipline;F.18when a reusable name or alias needs to be chosen. An F.9 semantic Bridge requires its own predicate, not merely a source term.