What this paper finds

• Discipline-led fan-out can multiply ambiguity when specialists rank the same population using different definitions of performance, scope, and evidence.
• The system stabilizes one observation, generates competing causal hypotheses, and assigns specialists to test predicted signatures and falsifiers instead of returning competing final answers.
• Reconciliation preserves dissent and separates observed severity, causal support, actionability, and uncertainty; agreement is treated as metadata, not proof.