violations are governed by the exchange rate `(D − w)/ϖ`, not a vanishing rate

Maintainer decision, formalized and landed by the fourth follow-up. An after-the-fact violation risk a continuation knowingly carries is priced: the honest bid includes −ϖ p, the option beats inquiry iff gord − w ≥ ϖ p (BRIAFollowup2.priced_risk_wins_iff, C.5's exchange rate), and the agent accepts such risk at a rate that does not go to zero. The theorem: the weighted average expected violation count per winning block — every class together, given the history at opening — is at most (D − w)/ϖ plus (ρ 𝒜_K + M(K))/(ϖ Σ w_k), vanishing when the allowance and the noise bound are o(K) (violation_rate_le_exchange, violation_rate_le_exchange_rescaled), from three things only: the landed overestimation bound, the noise hypothesis over all blocks (a selection fixed at opening), and inquiry on the menu; settlement convention (ii) absorbs the pricing error of the forecast events (eval_sub_score_ii against eval_sub_score_i). **The design's tolerance for violation risk is set by ϖ**: inside the band D − w < ϖ < (D − w)/p_min the tolerated rate lies strictly between p_min and 1 (tolerated_rate_band); drills sharpen the prices, they do not lower it. What was wrong before: the second follow-up's rate bounds took the winners' margin over the non-incident blocks and the third applied the noise hypothesis to that selection, which is decided by the block's own outcome — on a stream with constant incident probability the non-incident noise sum is linear, NoiseBounded fails there, and incidents_le_signed allows the constant rate (Witness.constantRisk, nonincident_noise_linear, incidents_constant_rate, signed_bound_allows_constant_rate). The named hypotheses of the violation claim are the noise hypothesis over all blocks and the conditional-expectation bound E[S_k | opening] ≤ D − ϖ π_k; competitiveness and the honest tracker leave that list. Rejected alternative: a vanishing incident rate as the target (the second and third follow-ups).