Threshold Calibration

Setting the pass threshold too low ships mediocre work. Setting it too high creates infinite loops. The calibration methodology: run 20 artifacts through the critic, collect scores, identify the natural quality distribution, and set the threshold at the 75th percentile of your current baseline. This is the "stretch but achievable" zone.

The threshold is not static. As the producer improves through feedback cycles, the baseline shifts upward. Recalibrate every two weeks by re-running the scoring benchmark. Our content quality threshold started at 6.5/10 in week one and sits at 8.2/10 in week seven — a 26.15% improvement driven entirely by critic-loop pressure. If the threshold stays flat while quality improves, you are no longer pushing improvement. If quality plateaus while the threshold climbs, you are creating a bottleneck.