C.11:5.2 - Episteme grounding

Tell. A model-selection comparison takes three already-articulated explanations and asks whether one more observation or one more comparison pass is rational before preferring one explanation over the others.

Show. C.11 governs the decision doctrine over the current explanation set: one BeliefState, one OutcomeModel, one explicit PreferenceOrder or EvaluativeMeasure, and, when the case needs it, either one InterventionModel or one CounterfactualModel plus one SubjunctiveDependenceRelation, rather than one thinner evidential comparison. When another model comparison pass is on the table, ValueOfComputation belongs here as part of the current choice doctrine rather than as one later planning afterthought.

Show. If one more comparison pass cannot realistically change which explanation survives, the decision record should not end with “more analysis may help.” It should end with one ChoiceResult that prefers the current explanation now. If one more pass could still reverse the ordering and is cheap enough to justify, the decision record should say exactly which pass is worth doing and what ambiguity it is expected to resolve.

Show. A lawful closing line here is therefore something like: “ChoiceResult = choose model 2 now because the surviving uncertainty no longer changes the ordering under the current evidence” or “ChoiceResult = run one additional comparison pass on models 1 and 2 because the current outcome model still cannot distinguish their failure costs.” Anything vaguer leaves the decision question unfinished.

Show. This pattern does not yet govern open-ended hypothesis generation and does not yet govern operational rollout. Those questions stay outside this pattern even when the decision later feeds them.