Does peer review quality collapse under submission overload?

Peer review, in this excerpt, is a feedback loop rather than a fixed filter. The authors argue that more submissions "overtaxes the pool of suitable peer reviewers," review accuracy then drops because journals must "solicit assistance from less qualified reviewers or ask current reviewers to do more," and lower accuracy leads more authors to "try their luck at venues that might otherwise be a stretch," which adds submissions again. They call this a "pernicious feedback loop" within a "peer-review meltdown." The claim is structural: it follows from the model and the Discussion's argument, and the excerpt reports no measured effect.

Two features drive the mechanism. Submission choices carry information: peer review "partially reveals authors' private sense of their work's quality through their decisions of where to send their manuscripts." Review labor is "unpaid and largely unrewarded," so journals lack the wages that would recruit reviewers when supply runs short. The Discussion adds slower pressures, including publication growth that the authors put at "about 5% annually since 1952," and Figure 1 lists amplifiers such as a proliferation of journals and noisier review that reduces what authors learn from rejection.

The excerpt treats machine review as unsettled. It notes that the propriety debate "remains unsettled," and that some reviewers use LLMs "even when journal or conference guidelines forbid doing so," placing that use beside the labor shortage without tying the two causally. The randomized experiment in Does banning LLM use in peer review change review outcomes? measures what bans and limits did to scores and decisions and how often reviewers broke them, evidence this excerpt does not offer. The loop bears more directly on Can automated review loops handle AI-generated research at scale?: that note argues journals and arXiv cannot scale to AI-generated volume, and this excerpt supplies the human-labor bottleneck behind that point, though it proposes no automated fix. The loop runs through falling review accuracy, the step Can inference scaling help reviewers catch errors humans miss? targets, but the excerpt evaluates no such tool.

What the excerpt does not establish is how strong the loop is. It gives the model's setup and the Discussion's argument, but not the results section or the appendix proofs the authors cite, so the strength of the cycle cannot be checked here. The model also holds manuscript quality exogenous, which the authors flag as a limitation: authors choose how much effort to invest in anticipation of scrutiny, and revise-and-resubmit may push them toward manuscripts that are "less than polished." The bibliometric studies behind the rise in declined invitations are cited rather than reproduced, with [10] noted as an exception. The implication is that the loop is a credible model-based account of why review capacity limits submission quality, and its practical weight stays open until the results or empirical work are in view.