Per-pair pedagogical readings of diff_with_*.json outputs. Audience: researchers learning to interpret diff artifacts. Each annotation walks one corpus pair field by field, distinguishing substantive movement from non-comparable dimensions and showing what a writeup looks like.
Per-pair pedagogical readings of peer_with_*.json outputs. Audience: researchers learning to interpret peer comparison artifacts. Each annotation walks one corpus peer pair field by field, distinguishing individual-pair findings from corpus-level claims and naming what the comparison cannot conclude.
Three illustrative rating files (good, mediocre, insufficient) showing what substantive vs uninformative Phase 2 submissions look like. Audience: prospective raters evaluating their own work before committing. The same document is rated three different ways; the contrast IS the teaching.