Fit the target.
Do not score beauty.

Observed ↔ target is a local objective

13.04 answers only one question: “How well does this observed dominance structure fit this explicit target profile under this explicit evaluation profile?” It does not answer whether the whole composition is globally good.

Composite-first scoring

DominanceEvaluationProfile

No implicit equal weights. A zero weight is an explicit choice to keep a condition diagnostic-only for this composite.

Six target conditions

Canonical MetricResult family

Weighted mean of applicable components

INVALID > N_A > KEEP

Only same-objective comparison

Correction request is not mutation

Failed applicable target conditions may emit non-mutating correction requests. The owner route is explicit; 13.04 does not decide or execute foreign geometry changes.

Target + evaluation policy are identity-bearing

No beautyScore by another name

Topic13 local chain

CANON → LAB → DATA → IDEAS

Visual laboratory hook

13.04 regression suite

Tests cover six condition semantics, explicit target N_A handling, weighted composite, MetricResult metadata, same-target comparison, zero-vs-N_A firewalls, routing/correction synthesis and consolidated Topic13 ownership boundaries.

DATA FOR LAYOUT ENGINE

{
  "module": "Compo 13.04",
  "name": "scoring_correction_and_tests",
  "version": "1.0-canonical",
  "canonical_main_title": "Scoring, correction и tests",
  "topic": "13_ДОМИНАНТА_И_ГЛАВНЫЙ_ФОКУС",
  "scope": "OBSERVED_VS_TARGET_COMPONENT_EVALUATION_COMPOSITE_LOCAL_TARGET_PROFILE_FIT_CORRECTION_REQUESTS_COMPARISON_AND_TOPIC13_TESTS",
  "main_topic_contract": {
    "topic_title": "Доминанта и главный фокус",
    "topic_goal": "Define dominance as a measurable leader of attention through focal candidates, dominance margin and competing/co-dominant cases without assuming a universal mandatory single focal center.",
    "expected_output": "DominanceDescriptor, focal rank, dominance margin, target profiles",
    "planned_files": [
      "13.00 Dominant role / focal candidate model",
      "13.01 Attention factors: size, contrast, isolation, position",
      "13.02 Dominance margin и competing dominants",
      "13.03 Target dominance profiles",
      "13.04 Scoring, correction и tests"
    ]
  },
  "ownership": {
    "owner": "Compo13.04",
    "owns": [
      "DominanceEvaluationProfile",
      "DominanceTargetConditionResult",
      "DominanceTargetCondition MetricResult components",
      "dominance.system.localTargetProfileFit composite MetricResult",
      "DominanceTargetEvaluation",
      "same-target/same-profile local evaluation comparison",
      "DominanceCorrectionRoutingPolicy",
      "DominanceCorrectionRequest synthesis",
      "Topic13 consolidated scoring/correction regression contract"
    ],
    "does_not_own": [
      "focal candidate eligibility — Compo13.00",
      "attention factors/raw measurements — Compo13.01",
      "attention aggregation/ranking/margin/observed class — Compo13.02",
      "target dominance profile declaration/binding — Compo13.03",
      "foreign geometry/layout mutation",
      "global hierarchy quality",
      "global rule priority",
      "universal beauty score"
    ]
  },
  "condition_keys": [
    "classification",
    "leaderIdentity",
    "leaderCount",
    "coDominantCount",
    "competingCount",
    "margin"
  ],
  "evaluation_profile_schema": {
    "evaluationProfileRef": "stable identity",
    "profileSourceRef": "explicit source",
    "profileSourceOwnerRef": "owner",
    "componentWeights": "exact six condition keys -> finite >=0; at least one >0",
    "aggregation": "WEIGHTED_MEAN",
    "targetDeclaredNAHandling": "DROP_EXPLICIT_TARGET_NA_ONLY",
    "missingObservedHandling": "N_A",
    "comparisonEpsilon": "finite 0..1, used only for same-profile composite comparison",
    "provenance": "canonical provenance class",
    "status": "OK|INVALID",
    "mutationAllowed": false,
    "default": "NONE — no implicit equal weights and no universal dominance target weighting"
  },
  "condition_evaluation": {
    "classification": "1 iff observed classification is in targetProfile.allowedObservedClassifications; else 0",
    "leaderIdentity": {
      "ANY_SUBJECT": "N_A by explicit target declaration; dropped from composite denominator",
      "ANY_OF": "1 iff at least one target subject is in observed leader subjects",
      "ALL_OF": "1 iff every target subject is in observed leader subjects; extra observed leaders allowed",
      "EXACT_SET": "1 iff observed leader subject set exactly equals target subject set"
    },
    "leaderCount": "1 iff observed leader count is within inclusive target leaderCountRange; else 0",
    "coDominantCount": "1 iff observed coDominantCandidateRefs count is within inclusive target range; else 0",
    "competingCount": "1 iff observed competingCandidateRefs count is within inclusive target range; else 0",
    "margin": {
      "IGNORE": "N_A by explicit target declaration; dropped from denominator",
      "REQUIRED_RANGE": "1 iff observed absoluteMargin is within inclusive target range; else 0"
    },
    "firewall": "all component fits are target-local binary fit descriptors; mismatch is a valid 0 under KEEP, not REJECT"
  },
  "metric_result_contract": {
    "minimumFields": [
      "key",
      "owner",
      "value",
      "applicability",
      "confidence",
      "status",
      "provenance",
      "composite",
      "componentOf",
      "dependsOn",
      "derivedFrom",
      "scoringGroup",
      "independentVoteAllowed",
      "evidence",
      "repairHints"
    ],
    "compositeKey": "dominance.system.localTargetProfileFit",
    "componentPattern": "dominance.target.componentFit.<conditionKey>",
    "compositeAggregation": "weighted mean of positive-weight applicable condition components",
    "targetDeclaredNA": "leaderIdentity for ANY_SUBJECT and margin for IGNORE may drop denominator only because target non-applicability is explicit",
    "observedNA": "required observed evidence N_A => composite N_A",
    "confidence": "minimum upstream rank-entry confidence across contributing evidence; never dominance strength",
    "provenance": "DERIVED"
  },
  "independent_numeric_vote": {
    "key": "dominance.system.localTargetProfileFit",
    "owner": "Compo13.04",
    "composite": true,
    "componentOf": null,
    "scoringGroup": "dominance-local",
    "independentVoteAllowed": true,
    "meaning": "EXPLICIT_TOPIC13_TARGET_PROFILE_FIT_ONLY"
  },
  "component_metric": {
    "keyPattern": "dominance.target.componentFit.<conditionKey>",
    "owner": "Compo13.04",
    "composite": false,
    "componentOf": "dominance.system.localTargetProfileFit",
    "scoringGroup": "dominance-local",
    "independentVoteAllowed": false
  },
  "evaluation_outcome_policy": {
    "outcomes": "INVALID|N_A|KEEP",
    "precedence": "INVALID > N_A > KEEP",
    "INVALID": [
      "malformed/incompatible lineage",
      "invalid evaluation profile",
      "invalid target profile/binding",
      "contradictory candidate-to-subject mapping"
    ],
    "N_A": [
      "target binding N_A",
      "required observed ranking/margin/class evidence unavailable",
      "missing confidence metadata for a positive-weight applicable component",
      "no positive-weight applicable component after explicit target-declared N_A drops"
    ],
    "KEEP": "all required comparable observed evidence is available; target mismatch remains a valid soft fit 0..1",
    "firewall": "Topic13 target mismatch is not a hard REJECT in 13.04 v1"
  },
  "composite_formula": {
    "formula": "localTargetProfileFit = sum(weight_i * fit_i) / sum(weight_i) over positive-weight APPLICABLE components",
    "componentFitDomain": "0|1 for applicable target conditions",
    "compositeDomain": "[0,1]",
    "zero": "0 is a valid complete target mismatch under KEEP; zero is not N_A or INVALID",
    "one": "1 means all positive-weight applicable target conditions fit; it is not global composition quality",
    "firewall": "no hidden falloff, no universal weighting, no conversion of N_A/INVALID into zero"
  },
  "target_condition_result_schema": {
    "conditionResultRef": "stable identity",
    "evaluationProfileRef": "explicit evaluation profile",
    "targetProfileRef": "13.03 profile",
    "targetProfileBindingRef": "13.03 binding",
    "conditionKey": "closed six-condition taxonomy",
    "observedValue": "structured observed value",
    "targetValue": "structured target contract",
    "satisfied": "true|false|null",
    "fit": "1|0|null",
    "applicability": "APPLICABLE|N_A",
    "status": "OK|N_A|INVALID",
    "confidence": "upstream evidence confidence or null",
    "reasonCode": "explicit",
    "derivedFrom": "canonical refs",
    "mutationAllowed": false
  },
  "dominance_target_evaluation_schema": {
    "dominanceEvaluationRef": "stable identity",
    "targetProfileRef": "profile",
    "targetProfileBindingRef": "binding",
    "evaluationProfileRef": "profile",
    "focalRankingRef": "observed ranking",
    "dominanceMarginRef": "observed margin",
    "competitionDescriptorRef": "observed competition",
    "conditionResultRefs": "six deterministic condition results",
    "componentMetricKeys": "six canonical component keys",
    "compositeMetricKey": "dominance.system.localTargetProfileFit",
    "outcome": "INVALID|N_A|KEEP",
    "value": "0..1 only under KEEP; null otherwise",
    "allEvaluatedConditionsSatisfied": "boolean under KEEP",
    "status": "OK|N_A|INVALID",
    "mutationAllowed": false
  },
  "comparison_policy": {
    "requires": [
      "both evaluations outcome KEEP",
      "same targetProfileRef",
      "same evaluationProfileRef",
      "same scoringGroup dominance-local"
    ],
    "relations": [
      "LESS",
      "EQUAL_WITHIN_EPS",
      "GREATER",
      "INCOMPARABLE"
    ],
    "epsilon": "evaluationProfile.comparisonEpsilon",
    "meaning": "EXPLICIT_TOPIC13_TARGET_PROFILE_FIT_ONLY",
    "firewalls": [
      "cross-target-profile comparison is forbidden",
      "cross-evaluation-profile comparison is forbidden",
      "higher localTargetProfileFit is not automatic final-layout winner",
      "comparison is not beauty/global hierarchy quality"
    ]
  },
  "correction_routing_policy_schema": {
    "correctionRoutingPolicyRef": "stable identity",
    "policySourceRef": "explicit source",
    "policySourceOwnerRef": "owner",
    "targetOwnerRefByCondition": "exact six condition keys -> explicit non-empty owner refs",
    "provenance": "canonical provenance class",
    "status": "OK|INVALID",
    "mutationAllowed": false,
    "firewall": "13.04 routes non-mutating correction requests; it never edits foreign-owned state"
  },
  "correction_request_schema": {
    "requestRef": "stable identity",
    "dominanceEvaluationRef": "source evaluation",
    "conditionKey": "failed applicable condition",
    "targetOwnerRef": "explicit routing owner",
    "reasonCode": "one of target mismatch reason codes",
    "observedValue": "copied observed evidence",
    "targetValue": "copied target contract",
    "evidenceRefs": "canonical refs",
    "suggestedAction": "REVIEW_UPSTREAM_DOMINANCE_EVIDENCE_OR_POLICY",
    "status": "OPEN|INVALID",
    "mutationAllowed": false
  },
  "correction_reason_codes": {
    "classification": "TARGET_CLASSIFICATION_MISMATCH",
    "leaderIdentity": "TARGET_LEADER_IDENTITY_MISMATCH",
    "leaderCount": "TARGET_LEADER_COUNT_OUT_OF_RANGE",
    "coDominantCount": "TARGET_CO_DOMINANT_COUNT_OUT_OF_RANGE",
    "competingCount": "TARGET_COMPETING_COUNT_OUT_OF_RANGE",
    "margin": "TARGET_MARGIN_OUT_OF_RANGE"
  },
  "critical_firewalls": [
    "observed dominance structure != desired dominance structure",
    "target profile != score",
    "target binding != target satisfaction",
    "target condition fit is local to one explicit targetProfileRef",
    "classification mismatch != REJECT",
    "leader mismatch != REJECT",
    "count mismatch != REJECT",
    "margin mismatch != REJECT",
    "target mismatch may produce numeric 0 under KEEP",
    "numeric 0 under KEEP != N_A",
    "N_A != 0",
    "INVALID != 0",
    "target-declared N_A may drop denominator only when non-applicability is explicit",
    "missing observed evidence cannot silently drop denominator",
    "component fit != independent numeric vote",
    "component metrics cannot vote independently",
    "only one independent Topic13 numeric vote",
    "dominance.system.localTargetProfileFit != global composition quality",
    "dominance.system.localTargetProfileFit != beautyScore",
    "attentionIndex != Topic13 final vote",
    "dominance margin != Topic13 final vote",
    "focal rank != Topic13 final vote",
    "candidate count != Topic13 vote",
    "leader count != Topic13 vote",
    "factor weights != global rule priority",
    "evaluation component weights != global rule priority",
    "no implicit equal evaluation weights",
    "no universal dominance target weights",
    "cross-target-profile ranking is forbidden",
    "cross-evaluation-profile ranking is forbidden",
    "higher localTargetProfileFit != automatic final-layout winner",
    "correction request != foreign mutation",
    "correction request != automatic repair",
    "module execution order != rule priority",
    "local != global",
    "descriptor != score unless explicitly owned MetricResult",
    "no universal beautyScore",
    "IDEA != CANON",
    "EXPERIMENT != CANON"
  ],
  "IDEAS_REFS": [
    "IDEA-01 full human test after all 72 topics; compare composite target-profile fit with human judgment",
    "IDEA-02 semantics needed before placement; semantic target meaning remains explicit/deferred",
    "IDEA-03 explicit rule priorities; evaluation weights never become hidden global priority",
    "IDEA-04 collect structured data for future composition model",
    "IDEA-05 preserve reasons for human corrections",
    "IDEA-06 structured machine-readable layer for website",
    "IDEA-07 graphics-for-post-text research remains downstream/separate"
  ],
  "DATA_CAPTURE": [
    "sceneRef",
    "targetProfileRef",
    "targetProfileBindingRef",
    "evaluationProfileRef",
    "componentWeights",
    "conditionResultRefs[]",
    "conditionFitsByKey",
    "conditionApplicabilityByKey",
    "conditionConfidenceByKey",
    "componentMetricKeys[]",
    "compositeMetricKey",
    "localTargetProfileFit",
    "evaluationOutcome",
    "allEvaluatedConditionsSatisfied",
    "correctionRoutingPolicyRef",
    "correctionRequestRefs[]",
    "comparisonRelation",
    "humanReasonCodes[]"
  ],
  "DEFERRED_SEMANTIC_INPUTS": [
    "semantic meaning of target-profile importance",
    "global hierarchy priority",
    "brand/editorial tradeoffs among local composition modules",
    "future learned condition weights",
    "future global final-layout selection policy"
  ],
  "HUMAN_FEEDBACK_HOOKS": [
    "TARGET_PROFILE_FIT_TOO_HIGH",
    "TARGET_PROFILE_FIT_TOO_LOW",
    "CLASSIFICATION_FIT_WRONG",
    "LEADER_IDENTITY_FIT_WRONG",
    "LEADER_COUNT_FIT_WRONG",
    "CO_DOMINANT_COUNT_FIT_WRONG",
    "COMPETING_COUNT_FIT_WRONG",
    "MARGIN_FIT_WRONG",
    "EVALUATION_WEIGHTS_WRONG",
    "ZERO_SCORE_SHOULD_NOT_MEAN_INVALID",
    "N_A_SHOULD_NOT_BE_ZERO",
    "CORRECTION_ROUTING_WRONG",
    "CORRECTION_REASON_WRONG",
    "HIGHER_LOCAL_FIT_NOT_BETTER_GLOBALLY"
  ],
  "CONFLICT_PRIORITY_ASSUMPTIONS": [
    "module execution order is not priority",
    "evaluation component weights are local scoring parameters, not global rule priority",
    "targetProfileRef defines the local objective and prevents cross-target ranking",
    "same-profile composite comparison does not choose the final layout globally",
    "correction routing names owners but grants no mutation authority",
    "future reviewed global conflict-priority owner remains authoritative"
  ],
  "LAB_HOOKS": [
    "human A/B: same target profile with different component-weight profiles",
    "human A/B: same composite score but different failed target conditions",
    "test whether binary condition fits predict perceived target satisfaction",
    "test zero-score complete mismatch versus N_A missing-evidence perception",
    "compare correction reason usefulness to human designer actions",
    "collect cases where higher localTargetProfileFit is not preferred globally"
  ],
  "topic13_local_chain": [
    "13.00 focal candidate root model",
    "13.01 explicit attention factors",
    "13.02 explicit aggregation + ranking + margin + observed competition",
    "13.03 explicit target dominance profiles",
    "13.04 observed-vs-target local fit + correction requests"
  ],
  "observation_record": {
    "moduleRef": "Compo13.04",
    "metricResults": [
      "dominance.system.localTargetProfileFit",
      "dominance.target.componentFit.<conditionKey>"
    ]
  },
  "stop_condition": "STOP after 13.04. Do not create closing audit in this file."
}