RESPONSES TO AI REVIEW | VERSION 1.0 | 9 NOVEMBER 2026

H1 (minor): acquisition reference lacks exact title.
Response: accepted. Reference [6] now reads 'From Observable States to Acquirable Predictive Models'. Final hostile re-review verified resolution.

H2 (minor):2013 abstract-only reading and distinct2014 full-text evidence must not be conflated.
Response: accepted. Reference [3] explicitly states abstract/record-only scope and the PDF format error. Reference [7] separately names Ahmad, Huang and Yu (2014), with DOI, primary open text and inspected sections. Section 1 attributes the fuller one-step stopping comparison to[7]. Final hostile re-review verified resolution.

P1 (minor): 'five unlabelled stopping-tree shapes' is imprecise because after 0 and after 1 are different ordered patterns.
Response: accepted. Section 4 now says 'five observation-ordered stopping patterns before assigning terminal forecasts'. Counts, formula and proof are unchanged. Final hostile re-review confirmed this wording.

H3 (optional): abstract's reduction of policies could be read as uniqueness at endpoints.
Response: no further change required. The theorem explicitly provides one optimal policy and permits redundant endpoint ties. The reduction is in optimal value. Both proof and hostile reviewers judged this interpretation sound; no uniqueness claim is made.

Prior-work scope objection: standard Bayesian/POMDP tools already encompass this fixture, and known nonmyopic sensing can explain its qualitative behavior.
Response: accepted as a constraint on the contribution. Title, abstract, Section 1 and release recommendation use technical-addendum scope. Generic methods are classically attributed. Read-then-transition timing is not claimed to lie outside partially observable control. Observation-before-hazard and compiled equality are attributed to existing B-Theory work.

No supported major objection remains within this declared scope. This is the result of internal AI review, not a guarantee that an external reviewer would accept novelty or significance. Unavailable full texts and unverified historical priority are retained as explicit limits rather than marked resolved.