PSDA.8:6 - Bias-Annotation

Scope: consequence comparisons for bounded decision-support use. Lenses: Onto/Epist separates alternative, forecast, observation, and result; Arch exposes interaction and distribution; Prag fixes the receiving question; Gov preserves conditions and authority; Did makes the comparison replayable.

Ranking bias prefers closure over partiality. Baseline bias compares change with an empty world. Aggregation bias hides losers and shared burdens. Model precision can conceal missing applicability. Counter these by exposing consequence paths, testing the common basis, and stating the exact relation actually supported.