claim 'urms2-0.51' has no recorded blind attack: the review is not standing until one runs
id: r_fb9c81
question: Does a blind attack find a structural or analytic failure in the URMS2 0.51 bandwidth claim, without access to the author's reasoning?
frontier: the claim asserts the URMS2 bandwidth extends past the half band to 0.51, resting on the stated assumption that the true logarithmic frequency separation is preserved rather than collapsed into a cutoff estimate; no blind attack had been recorded against it, so the review was not standing
proposed_attack: ask what modification of the world would preserve the appearance of the result while making its interpretation false, specifically whether perturbing the polynomial frequency spacing or the coefficient bounds can keep the mean-value integral looking sound while falsifying it, then probe the Montgomery-Vaughan off-diagonal bound numerically
dead_routes:
- reading the author's derivation before attacking; that makes the review white-box and a separate exercise, and this one is blind by construction
- arguing from the shape of the assumption alone; the blockpos-0.672529 kill shows an assumption can read as sound and still fail on an exact witness, so the attack must compute
- treating a bounded off-diagonal error as sufficient on its own; it must be compared against the main diagonal term, since domination is the property in question
required_oracles:
- the Montgomery-Vaughan mean-value theorem as stated in the primary literature
- exact evaluation of the polynomial spacing cost sum n |c_n|^2 by direct computation, not by quoting the asymptotic
- the main diagonal term U log x, recomputed in the same probe rather than carried in from the claim
kill_conditions:
- the off-diagonal error diverges, or is not strictly dominated by the main diagonal term, at any admissible alpha and delta
- the spacing cost exceeds its O(x log x) bound once the upper polynomial length W is allowed to exceed the integration block U
- the probe cannot reproduce the claim's own regime, in which case the attack has not reached the claim and reports that instead of a verdict
agents_may:
- search
- derive
- code
- attack
agents_may_not:
- declare novelty
- declare theorem status
- promote their own claim
- read the author's derivation before the probe has run