{"documentVersion":"3.6.0","scoringVersion":"3.2.0","unmeasuredNeutral":50,"scoreRange":{"min":0,"max":100},"inputReachability":{"riskTolerance":{"item":"c_risk_tolerance","web":"never","api":"always"},"reversibility":{"item":"c_reversibility","web":"always","api":"always"},"constraintLoad":{"item":"c_constraints","web":"always","api":"optional"},"safetyNet":{"item":"c_safety_net","web":"never","api":"optional"},"exitPlan":{"item":"c_exit_plan","web":"conditional","api":"optional"},"deadlinePressure":{"item":"c_deadline","web":"always","api":"always"},"agency":{"item":"c_agency","web":"always","api":"always"},"avoidance":{"item":"c_avoidance","web":"always","api":"always"},"socialPressure":{"item":"c_social_pressure","web":"always","api":"always"},"decisionConfidence":{"item":"c_decision_confidence","web":"never","api":"optional"},"valuesClarity":{"item":"c_values_clarity","web":"never","api":"optional"}},"inputTransforms":{"constraintLoad":{"from":"c_constraints","form":"count of selected constraints, excluding the sentinel \"none\", divided by 6, times 100, then clamped to 0..100","denominator":6,"excludes":["none"],"unmeasuredWhen":"the answer is absent or is not an array"},"deadlinePressure":{"from":"c_deadline","form":"ordinal answer mapped to a fixed point; any other value is unmeasured","map":{"now":90,"weeks":60,"months":30,"later":10}},"regret":{"from":"c_regret","form":"ordinal answer mapped to a fixed point; any other value is unmeasured","map":{"waited_too_long":100,"both_equal":50,"unclear":50,"committed_too_soon":0}}},"contradictionRules":{"riskToleranceVsGrowth":{"severity":18,"requires":["c_risk_tolerance","c_priorities"]},"socialPressureVsAutonomy":{"severity":16,"requires":["c_social_pressure","c_priorities"]},"avoidanceVsRegret":{"severity":16,"requires":["c_avoidance","c_regret"]},"irreversibleVsRiskTolerance":{"severity":14,"requires":["c_reversibility","c_risk_tolerance"]},"agencyVsImpact":{"severity":11,"requires":["c_agency","c_priorities"]},"pickedVsImmediateLeaning":{"severity":12,"requires":["c_pick_today","c_gut_feeling"]},"leaningVsEmotionalState":{"severity":10,"requires":["c_gut_feeling","c_emotional_state"]},"valuesVsIdentity":{"severity":12,"requires":["c_values_clarity","c_identity_alignment"]},"confidenceVsPastOutcome":{"severity":10,"requires":["c_decision_confidence","c_past_similar"]}},"contradictionNoteLimit":5,"formulas":{"riskScore":{"form":"weighted mean over measured dimensions only, then clamped","weights":{"riskTolerance":0.3,"reversibility":0.25,"constraintLoad":0.12,"safetyNet":0.1,"exitPlan":0.08,"deadlinePressure":0.15},"inverted":["riskTolerance","reversibility","safetyNet","exitPlan"]},"agencyFriction":{"form":"weighted mean over measured dimensions only, then clamped","weights":{"agency":0.4,"avoidance":0.22,"socialPressure":0.1,"decisionConfidence":0.16,"constraintLoad":0.12},"inverted":["agency","decisionConfidence"]},"conflictScore":{"form":"contradiction severity plus conditional terms, then clamped","terms":{"contradictionSeverity":0.58,"unsureBonus":16,"weakConsensusBonus":14,"preferenceAgainstEvidenceBonus":12,"socialPressureThreshold":55,"socialPressureAboveThreshold":0.12,"valuesClarityInverted":0.12,"weakConsensusFraction":0.3333333333333333},"predicates":{"contradictionSeverity":"always: the summed severity of the fired contradiction rules, times the published multiplier","unsureBonus":"added when the current-preference answer is \"unsure\" or no option was picked","weakConsensusBonus":"added when at least two option signals were supplied and consensusStrength is at or below weakConsensusFraction","preferenceAgainstEvidenceBonus":"added when a preference and an evidence leader were both supplied and name different options","socialPressureAboveThreshold":"added as max(0, socialPressure - socialPressureThreshold) times the published rate; contributes nothing at or below the threshold","valuesClarityInverted":"added as (100 - valuesClarity) times the published rate; contributes nothing when valuesClarity was not measured","form":"the sum of the terms above, then clamped to the score range"}},"clarityScore":{"form":"base minus weighted penalties plus weighted supports, then clamped","terms":{"base":100,"conflict":-0.44,"risk":-0.2,"agencyFriction":-0.18,"valuesClarity":0.09,"decisionConfidence":0.09}}},"notes":["A dimension the person was never asked is excluded from its weighted mean rather than counted as a neutral answer. Only the fallback when nothing at all was measured is the neutral value.","Scores are withheld entirely from the public map unless every defined score input was supplied, so a partial answer set cannot be read as a precise number.","These weights are not empirically validated coefficients. They encode a stated ordering of concerns and are published so a result can be checked, not because a study established them.","Contract 1.11.0 made every weighted input suppliable over the public API, so a caller that supplies all of them reaches the whole published model — including the two positive clarity terms, which no surface could reach before. They remain optional: an existing request returns a byte-identical result, and each response now reports in dataQuality.modelReach how much of the declared weight it actually reached.","A declared weight is not a promise that the input behind it is collected. See inputReachability: safetyNet, decisionConfidence and valuesClarity are supplied by no shipped surface, and exitPlan is only asked when a follow-up rule fires. So the risk score in the guided web scan is a weighted mean over three of six dimensions carrying 0.52 of the declared weight, rising to four of six and 0.60 only in the scans where the exit-plan follow-up is reached; and the clarity score never receives its two positive terms.","Six of the nine contradiction rules require an input the guided web scan never asks for, so they cannot fire there. Measured by severity, a web visitor’s conflict score is drawn from 43 of 119 declared points; the remaining rules are reachable only where the missing inputs are supplied. This was verified by sweeping the engine across the live question space, not by reading the conditions.","Document 3.6.0 records that the API can now supply safetyNet, exitPlan, decisionConfidence and valuesClarity, which 3.5.0 correctly reported as reachable from no surface. Document 3.5.0 adds inputTransforms and the conflict-term predicates, without which riskScore, agencyFriction and conflictScore could not be recomputed from this document at all: the weights were published over quantities whose derivation was not. Verified by recomputing a live API result from the document alone. Document 3.4.0 corrects 3.3.0, which stated the 0.60 best case as though it were the shipped case while its own inputReachability table already marked exitPlan conditional. 3.3.0 remains served verbatim at its own address. Document 3.3.0 revised 3.2.0 by adding inputReachability, the contradiction rule table, and changes no number. Scoring engine 3.2.0 is unchanged and its golden vectors are byte-identical; 3.2.0 remains served verbatim at its own address."]}