# Amendments to the second-pass protocol Amendment 1 (9 Nov 2026, 10:40 AM ET, before any grade was compared or reported): the authors had a reply window (29 Oct letter, 4 Nov reminder, replies asked by 6 Nov) and did not reply, so they get no second one. So the protocol line "it never changes the published verdict without a new right-of-reply letter" is replaced: a RECLASSIFY candidate that survives adjudication and the red team goes into the published check with a dated note that it was added on or after 9 Nov 2026, after the authors' reply window (29 Oct letter, 4 Nov reminder, replies asked by 6 Nov) had closed with no reply. No new letter and no new window. Private until final. Reader B, 9 Nov 2026, 10:44 AM ET: Gemini 3.8 Flash (Google), as in the first pass. Amendment 2 (9 Nov 2026, 10:51 AM ET, before any reader B grade was compared with reader A): reporting only, grades unchanged. Two readers found a gap: a choice the paper states and tests for sensitivity, but that no cited work uses, meets neither STANDARD nor the 'not disclosed' half of DEFENSIBLE. So every result is also reported with a plain disclosed yes/no (from reader A's notes, checked at adjudication), and DEFENSIBLE results are split in the write-up into 'disclosed' and 'not disclosed'. The authors' own earlier papers count as works the paper cites when the paper cites them (as the protocol's wording already says); results say how many STANDARD grades rest only on the authors' own work.