Item Q2: What disciplines the grade-to-quantity relation?

Everything epistemic in the deference line rests on a relation between what the principal grades and what the intervention is worth. Assumed uniformly it makes the market dispensable — the conclusion follows in three lines with the bound attained — so the target has to be a statistical relation, derived. But the relation mentions only the principal and the world, never the agent's credence, so no coherence or no-exploitability condition on the agent can establish it. No round is filed for that unresolved hypothesis shape.

What is missing: either a reformulation whose subject is something the agent's own dynamics can constrain — the candidate is discipline on the agent's estimate of the discrepancy once grades are themselves scored — or an argument that the residue is irreducibly a competence assumption and should be declared rather than derived.