Round 1 — multi-review synthesis

Five majors: the P5-3 existence claim fired on any of 16 personas with no familywise control; the human comparator was not protocol-matched (“δ-matched” and N-fold language retired); human microstructure was unproven; the broad realism-vs-effect thesis was already occupied by concurrent work; and the draft was three papers in one. Disposition: the contribution narrowed to a fixed-panel strategic-interaction example with auditability as a credibility layer, not the sole novelty.