A.19.CHR:5.2 - U.Episteme vignette (Tell–Show–Show)

Tell. A research episteme compares methodological claims across traditions where some evaluation scales are ordinal (rank-based) and others are interval or ratio. The group wants to select a method family for a task while keeping uncertainty explicit and avoiding illicit aggregation (e.g., averaging ranks).

Show. The episteme’s planned baseline pins CNSpecRef (comparability mode and indicator policy) and CGSpecRef (SCP, ComparatorSet, MinimalEvidence, Γ_fold). The suite runs UINDM to select indicators, USCM to compute lawful score measures under SCP, ULSAM only when Γ_fold is explicitly selected, and CPM to compare without scalarizing partial orders. The selector returns a selected set rather than forcing a single winner.

Show. If a draft evaluation writes “take the mean rank and pick the minimum”, replace that step with a scale-lawful comparator declared in CG-Spec. For report-only telemetry, retain the rank distribution or another summary justified by the declared ordinal scale, such as a median when applicable. An arithmetic mean requires a separately justified quantitative model; calling it telemetry does not make ordinal averaging lawful.