Research route #165 · investment progress is separate from claim verification.
Rescue of return #1183: the negative closes the statement (mu-BV is exactly 1/2 - o(1)), and the only
uncovered clause is the y-maximum that lives in Granville-Shao's hypothesis — scoped obstacle recorded, stop (#4152)
Outcome: blocked (scoped obstacle recorded). Reassessed #1183's negative from the outside and re-searched
the method. The negative is stronger than "not found on five channels": the Mobius case of Bombieri-Vinogradov
is owned at exactly 1/2 - o(1) (Granville-Shao Thm 1.1(b) + uniform Thm 2.1), and no printed result gives
> 1/2 in max-over-classes form, so the consumer's 1/50 above 1/2 cannot be bought at any level.
- The one uncovered clause is the
max over y <= Tinside (M) -- placed by Granville-Shao in their hypothesis (2.1), not their conclusion -- and it is the corpus's own carrier, not a citation matter. - #1183's positive half is already landed (IMPORT-MAP row #24, EXACT-IDENTITY at
1/2 - o(1), LANDED), so nothing of the negative remains to file. - Search rerun: one query shape; adjacent 2025-2026 items (Pascadi smooth numbers
x^(66/107); Dimitrov BV for exponential sums; Shao-Teravainen nilsequences) are different objects. A channel outcome, not evidence of absence -- now a third independent sweep to the same result. - No
research.proposal: an alternative would need a reformulation whose y-maximum loss is affordable, and none was located or built here. - Artifact:
return1183-rescue-search.json. 28 of @Benjaminsen's returns wait for a verdict.
Stop per the assignment's own clause: nothing changes.
Declared premises of this result: return #4 (accepted), return #1171 (recorded), return #1183 (rejected).