Peer review is collapsing under the weight of its own success

The machinery of scientific validation is grinding to a halt, not because researchers have lost faith in scrutiny but because they can no longer afford to provide it. Peer review, the unpaid labour that stands between a submitted manuscript and publication, was conceived in an era when journals were few, submissions modest, and academics enjoyed the luxury of unstructured time. That world has vanished, yet the system persists largely unchanged, sustained by a social contract that grows more threadbare each year.

The arithmetic is unforgiving. Global research output has expanded relentlessly over recent decades, driven by growing numbers of scientists, intensifying pressure to publish, and the proliferation of journals eager to capture market share. Each manuscript typically requires two or three reviewers, each of whom must donate hours to read, critique, and recommend revisions. The supply of qualified experts has not kept pace with demand, and those experts face competing claims on their attention that did not exist a generation ago.

Academic employment has shifted from a model that rewarded broad intellectual citizenship to one that counts publications, citations, and grant income with unforgiving precision. Reviewing a colleague's paper generates none of these metrics. It appears on no curriculum vitae, attracts no funding, and contributes nothing to the tallies that determine hiring, promotion, or tenure. The incentive structure has become actively hostile to the behaviour the system requires.

Institutions have compounded the problem by colonising the discretionary time that once made unpaid service feasible. Teaching loads have grown, administrative burdens have multiplied, and the expectation of continuous grant applications has turned research itself into a form of piecework. The hours that might once have been devoted to reading a manuscript with care are now claimed by compliance paperwork, student emails, and the relentless cycle of proposal writing. Senior researchers, who possess the expertise most journals seek, are precisely those whose time is most fragmented and whose opportunity cost is highest.

The result is a market failure in slow motion. Journals struggle to find reviewers willing to accept invitations, and those who do agree often take months to return reports. Editors send dozens of requests before securing a panel, and manuscripts languish in queues that stretch credibility. The delay is not merely inconvenient; it distorts the entire publication process, rewarding speed over thoroughness and encouraging authors to submit to less selective venues where acceptance comes quickly.

Quality control has not disappeared but rather migrated to new arenas that operate outside the traditional gatekeeping structure. Preprint servers, which publish manuscripts without peer review, have become the de facto first venue for many fields, particularly in the life sciences and physics. Researchers increasingly read and cite work that has never passed through formal scrutiny, relying instead on their own judgement and the informal commentary that accumulates in online forums and social media.

Post-publication review has emerged as a parallel system, one that rewards engagement rather than anonymity. Platforms that allow open commentary on published papers attract researchers willing to critique work publicly, often with more rigour than the original peer review provided. The incentive is reputational rather than altruistic: a well-argued critique can raise a researcher's profile in ways that anonymous reviewing never could. The shift is profound, moving validation from a private transaction between editor and reviewer to a public performance subject to its own social dynamics.

The traditional journal, meanwhile, faces a crisis of legitimacy. If the most important work circulates as preprints and the most penetrating critiques appear on blogs or post-publication platforms, what function does the journal serve? The answer increasingly lies in prestige rather than quality control. Publication in a high-status venue signals that a manuscript has cleared a bar, but that bar is no longer defined primarily by intellectual merit. It reflects instead the scarcity of acceptance, a filtering mechanism shaped as much by reviewer availability and editorial workload as by scientific rigour.

Some disciplines have adapted more successfully than others. Mathematics, with its smaller community and culture of meticulous proof, has maintained a reviewing culture that still functions, albeit under strain. Computer science has embraced conference publication, where reviewing is concentrated in intense bursts and tied to specific deadlines, creating a different set of pressures but avoiding the indefinite drift of journal review. The biomedical sciences, facing the largest volume of submissions and the most fragmented reviewer pool, have seen the most acute breakdown.

Proposals to reform the system circulate regularly but founder on the same obstacles. Paying reviewers would require journals to generate revenue or institutions to provide subsidies, neither of which seems likely at scale. Requiring researchers to review in proportion to their own submissions would create perverse incentives, encouraging perfunctory reports and discouraging thorough critique. Open review, in which reports and identities are published alongside the paper, has attracted interest but also resistance from researchers wary of retaliation or social discomfort.

The most significant experiments involve decoupling review from publication. Some platforms allow researchers to submit a manuscript once and have it reviewed by a central service, then take the reviews to any participating journal. The model reduces duplication and spreads the reviewing burden, but it requires coordination across publishers with competing interests and has yet to achieve widespread adoption. Other initiatives attempt to create review as a standalone product, with reports that can be attached to any version of a manuscript, whether preprint or published article.

The deeper question is whether the profession still values the kind of scrutiny that peer review was meant to provide. The process was never designed to catch fraud or verify data; it aimed instead to assess whether a manuscript made a coherent argument, addressed the relevant literature, and drew conclusions warranted by its evidence. That form of quality control requires time, attention, and a willingness to engage seriously with another researcher's work. It is precisely the kind of intellectual labour that metric-driven academia has devalued.

Elite researchers, who once saw reviewing as a professional duty and a way to stay abreast of their field, increasingly decline invitations. Their refusal is rational: the opportunity cost is high, the recognition nil, and the manuscripts they are asked to review often fall below the standard they consider worth their time. The result is a hollowing out, in which the most qualified reviewers withdraw and the burden falls to junior researchers, who may lack the expertise or confidence to provide meaningful critique, or to those with fewer competing demands, who may not represent the field's leading edge.

The diffusion of quality control into multiple venues creates a more chaotic landscape but not necessarily a worse one. Preprints allow rapid dissemination and enable researchers to claim priority without waiting for the publication queue. Post-publication commentary can be more substantive than traditional review because it addresses the work as published rather than as a draft, and because critics can build on one another's observations. The disadvantage is that this scrutiny is unevenly distributed: high-profile work attracts attention, while papers in niche areas may receive none.

The transition also shifts risk from journals to readers. Traditional peer review, for all its flaws, provided a minimal assurance that someone with expertise had examined a manuscript before publication. Readers could reasonably assume that egregious errors or unsupported claims had been filtered out, even if the review was imperfect. In a world where preprints circulate freely and post-publication critique is sporadic, that assurance evaporates. Readers must exercise their own judgement, a burden that may be manageable for specialists but problematic for those outside the field or for journalists and policymakers seeking reliable summaries.

The commercial dynamics of publishing add another layer of complexity. For-profit publishers depend on journals' prestige to justify subscription fees or article processing charges, and that prestige rests on the perception of rigorous gatekeeping. Yet those same publishers benefit from volume, which incentivises launching new journals and accepting more papers. The tension between selectivity and growth is managed through a tiered system, in which a few flagship journals maintain high rejection rates while a proliferating array of secondary titles absorbs the overflow. The result is a market segmented by prestige rather than by intellectual coherence, with reviewing standards that vary widely even within a single publisher's portfolio.

The pandemic offered a glimpse of what the system looks like under extreme stress. Submissions surged, particularly in biomedicine, as researchers raced to publish findings related to the virus. Preprint servers became the primary venue for disseminating results, and traditional journals struggled to keep pace. Some expedited their review processes, but speed came at a cost: papers were published with errors that would likely have been caught under normal scrutiny, and retractions followed. The episode demonstrated both the value of rapid dissemination and the risks of bypassing careful review.

What remains unclear is whether the profession will converge on a new equilibrium or continue to fracture along disciplinary lines. Some fields may sustain traditional peer review by reducing submission volumes, either through higher desk rejection rates or through cultural shifts that discourage publication for its own sake. Others may embrace preprints and post-publication review as the primary modes of validation, accepting the trade-offs in exchange for speed and transparency. Still others may develop hybrid models, in which initial dissemination occurs without review but formal validation follows for work that meets some threshold of importance.

The question is not whether peer review will survive but what will replace the functions it once performed. Gatekeeping, quality assurance, and the certification of knowledge are not optional; they are necessary for science to function as a cumulative enterprise. If traditional journals can no longer provide these services reliably, something else must. The alternatives emerging now are partial, experimental, and unevenly distributed. Whether they will cohere into a system capable of sustaining trust in scientific claims remains an open question, one that matters far beyond the academy.

This article was produced with AI assistance and reviewed against our editorial standards.