Gusev, Ilya · Nov 19, 2026
Twice-Told Critiques: Claim-Level Homogenization of Peer Review in the LLM Era
Markdown · JSON · Original source
Ilya Gusev argues that peer reviews of the same paper are becoming less distinct, and that reviews flagged by MELD, a detector of -generated text, overlap more than unflagged reviews. Peer review depends on readers identifying different problems; if their critiques increasingly repeat one another, a paper may receive less varied scrutiny. Gusev reports rising critique overlap across ICLR, NeurIPS, ICML and journal sources, and says a third ICLR review added fewer new problems in 2027 than in 2026. The MELD comparison is an association: the abstract presents evidence that flagged reviews overlap more, but does not establish that LLM use caused the pattern. Escape Meridian interpretation: the results point to a change in the variety of critiques, not a demonstrated change in review quality or its effects on publication decisions. A timing issue also affects interpretation: the abstract reports results for 2027, although the paper is listed as submitted on 19 November 2026. The abstract does not explain this chronology or give enough methodological detail to assess how the overlap was measured.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
cs.AI · Article
Source details
Original source date: Nov 19, 2026 (2026-11-19).
Date basis: Date basis not provided.
Added to Escape Meridian: 2026-11-20T06:32:03.725Z. First observed: 2026-11-20T05:07:25.337Z. Retrieved: 2026-11-20T06:30:58.829Z.
Referenced by
Recorded citations in public readings. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
The abstract reports results for 2027, although the page lists the paper as submitted on 19 November 2026; it does not explain this chronology. The supplied abstract does not describe the overlap-measurement method or establish that -generated text caused the reported differences. The supplied evidence is the abstract, not the paper's full text.
Open question
What explains the greater overlap between MELD-flagged reviews, and how does the paper distinguish the role of -generated text from other possible explanations?
Source-stated
Gusev reports that measured critique overlap at ICLR rose between 2026 and 2027.
At ICLR, measured overlap rose from 12.0% in 2026 to 21.3% in 2027
Source-stated
The abstract reports year-on-year rises in measured overlap at NeurIPS, ICML and the F1000 and PLOS journals.
with year-on-year rises at NeurIPS, ICML and the F1000 and PLOS journals.
Source-stated
Gusev reports that a third ICLR review added fewer new problems in 2027 than in 2026.
A third ICLR review of a paper added 5.46 new problems to the other two in 2027, against 6.12 in 2026.
Source-stated
The abstract says two MELD-flagged reviews of the same paper overlapped more than two unflagged reviews across ICLR and journal sources.
Within a paper, two reviews that MELD flags overlap more than two unflagged reviews
Escape Meridian interpretation
The abstract presents the MELD comparison as evidence of an association and says it suggests critique diversity is falling; it does not state that LLM use caused the pattern.
These results provide evidence that LLM-flagged reviews of one paper are less distinct from each other than unflagged reviews, and suggest that the diversity of critiques among reviews of a single paper is falling.
Source and contributor context
Gusev, Ilya · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Large language model
A language model with many learned numerical settings, used to process and generate language.
arXiv
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Editorial provenance
Version 1. Publication ID: d98ebd0e-a38d-499c-88f3-913319356532.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: passed by anthropic/claude-opus-5.5 on 2026-11-20T06:32:03.725Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: useful_perspective.
Observation: 1d38a57e1d065ce18f98994c46fe4cb8686fcc35c484a2f3a9f781a61725e978. Source text hash: 28ba9dd9a0805248b51c7b84d19b5cfdfe6bd53a24390c9ac5d4e3222c526878.
Revision and source-check history
Version 1 · 2026-11-20T06:32:03.725Z
Twice-Told Critiques: Claim-Level Homogenization of Peer Review in the LLM Era
Ilya Gusev argues that peer reviews of the same paper are becoming less distinct, and that reviews flagged by MELD, a detector of -generated text, overlap more than unflagged reviews. Peer review depends on readers identifying different problems; if their critiques increasingly repeat one another, a paper may receive less varied scrutiny. Gusev reports rising critique overlap across ICLR, NeurIPS, ICML and journal sources, and says a third ICLR review added fewer new problems in 2027 than in 2026. The MELD comparison is an association: the abstract presents evidence that flagged reviews overlap more, but does not establish that LLM use caused the pattern. Escape Meridian interpretation: the results point to a change in the variety of critiques, not a demonstrated change in review quality or its effects on publication decisions. A timing issue also affects interpretation: the abstract reports results for 2027, although the paper is listed as submitted on 19 November 2026. The abstract does not explain this chronology or give enough methodological detail to assess how the overlap was measured.
Original source
cs.AI · Article
Source details
Original source date: Nov 19, 2026 (2026-11-19).
Date basis: Date basis not provided.
Added to Escape Meridian: 2026-11-20T06:32:03.725Z. First observed: 2026-11-20T05:07:25.337Z. Retrieved: 2026-11-20T06:30:58.829Z.
Limits
The abstract reports results for 2027, although the page lists the paper as submitted on 19 November 2026; it does not explain this chronology. The supplied abstract does not describe the overlap-measurement method or establish that -generated text caused the reported differences. The supplied evidence is the abstract, not the paper's full text.
Open question
What explains the greater overlap between MELD-flagged reviews, and how does the paper distinguish the role of -generated text from other possible explanations?
Source-stated
Gusev reports that measured critique overlap at ICLR rose between 2026 and 2027.
At ICLR, measured overlap rose from 12.0% in 2026 to 21.3% in 2027
Source-stated
The abstract reports year-on-year rises in measured overlap at NeurIPS, ICML and the F1000 and PLOS journals.
with year-on-year rises at NeurIPS, ICML and the F1000 and PLOS journals.
Source-stated
Gusev reports that a third ICLR review added fewer new problems in 2027 than in 2026.
A third ICLR review of a paper added 5.46 new problems to the other two in 2027, against 6.12 in 2026.
Source-stated
The abstract says two MELD-flagged reviews of the same paper overlapped more than two unflagged reviews across ICLR and journal sources.
Within a paper, two reviews that MELD flags overlap more than two unflagged reviews
Escape Meridian interpretation
The abstract presents the MELD comparison as evidence of an association and says it suggests critique diversity is falling; it does not state that LLM use caused the pattern.
These results provide evidence that LLM-flagged reviews of one paper are less distinct from each other than unflagged reviews, and suggest that the diversity of critiques among reviews of a single paper is falling.
Source and contributor context
Gusev, Ilya · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Large language model
A language model with many learned numerical settings, used to process and generate language.
arXiv
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Editorial provenance
Version 1. Publication ID: d98ebd0e-a38d-499c-88f3-913319356532.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: passed by anthropic/claude-opus-5.5 on 2026-11-20T06:32:03.725Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: useful_perspective.
Observation: 1d38a57e1d065ce18f98994c46fe4cb8686fcc35c484a2f3a9f781a61725e978. Source text hash: 28ba9dd9a0805248b51c7b84d19b5cfdfe6bd53a24390c9ac5d4e3222c526878.
passed · 2026-11-20T06:32:03.725Z · anthropic/claude-opus-5.5
Publication: d98ebd0e-a38d-499c-88f3-913319356532
attribution — pass: The reading attributes the argument and results to Ilya Gusev, the listed author. Findings are framed as reports in the abstract, not as established facts.
Authors:Ilya Gusev
summary — pass: The reading leads with the author's thesis and states the source's own premise that peer review relies on different readers finding different problems. It explains MELD as an LLM-text detector and keeps the association-versus-causation limit. ICLR, NeurIPS and ICML are not glossed; a short gloss such as 'major AI conferences' would help, but they read clearly as review venues. The chronology note is hedged as a gap, not an error. The question follows from the evidence.
Peer review relies on several readers finding different problems.
date — pass: originalPublished is 2026-11-19, which matches the arXiv submission date and the v1 submission history. The feed date of 20 November is later and is correctly not used.
[Submitted on 19 Nov 2026]
support — pass: Each claim matches the abstract's verbatim excerpts: the ICLR overlap rise, the rises at other venues, the fewer new problems added by a third review, and the MELD-flagged overlap at ICLR and journal sources. The Escape Meridian interpretation is labeled and conservative, with no causal or quality claims added. Flagging the 2027 data against a November 2026 submission is a reasonable stated gap, not an invented contradiction.
These results provide evidence that LLM-flagged reviews of one paper are less distinct from each other than unflagged reviews
media — pass: The selected media is the verified arXiv abstract page for this paper, the direct original source. No better media was found.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
A language model with many learned numerical settings, used to process and generate language.
Definition referencearXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXivA language model with many learned numerical settings, used to process and generate language.
Definition referenceA language model with many learned numerical settings, used to process and generate language.
Definition referenceA language model with many learned numerical settings, used to process and generate language.
Definition referencearXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXivA language model with many learned numerical settings, used to process and generate language.
Definition referenceA language model with many learned numerical settings, used to process and generate language.
Definition reference