TDA manuscript: AI-review questions and framing

Interesting experiment- how well does AI critique? useful ones: what am i preserving, what gets discarded, why choose this over the simpler calculation?

source provenance: the files are stored under Claude Review. Their referee-report style and recommendations are AI review artifacts, not evidence of journal peer review or an editorial decision.

useful questions extracted from the later report: do the rank-deficit, normalization and entropy threads form a useful end-to-end procedure? Can a practitioner instantiate an error budget rather than only read abstract constants? Does the paper show when the bound changes an actual decision? Are claims about earlier methods tied to the precise operation being analyzed?

tradeoff to make explicit: persistent entropy can ignore uniform lifetime scale while discarding birth-position/rank-window information. A fixed-window deficit answers a more specific requirement but inherits the chosen ruler. Explain what each quantity measures rather than treating one as a universal replacement for the other.

review task: turn each critique into a question with a manuscript location, proposed edit/experiment and resolution status. The AI report's claims that it reproduced proofs or numerics are not independent verification performed by this notebook scan and should not be used as an endorsement.

version control: compare V1/V2/V3 critique files against the matching manuscript; a criticism resolved in a later draft should not remain labeled as currently open without checking.