Peer Review Is Breaking Down as Paper Volumes Climb 5.6% a Year
Scientific peer review is under severe strain as publication volumes grow exponentially and editors struggle to recruit qualified reviewers, raising questions about how AI-era research will be validated.
👁 112 reads
Scientific peer review is buckling under exponential growth in submissions, with papers indexed in Scopus and Web of Science increasing at 5.6 percent per year and editors now struggling to recruit even a single willing reviewer per manuscript.
- Papers indexed in Scopus and Web of Science are growing at 5.6% per year, and one editor recently had to contact 30 researchers to find a single reviewer
- Researchers globally devote a collective 15,000 years of work to peer review annually - work that would cost $1.5 billion if paid, for the US share alone
- AI research in particular is already shifting toward informal publication channels such as blogs, bypassing traditional review altogether
The strain shows up in concrete ways. Steven Mack, an editor at Human Immunology, recently had to email around 30 researchers to find one who would review a paper. Five years ago, he estimates five to ten emails would have produced three willing reviewers. Haseeb Irfanullah, an independent consultant and editorial board member at Wiley’s Learned Publishing, put it bluntly: “I struggle like anything to get peer reviewers.”
The scale of the unpaid labour involved is striking. According to figures cited in Ars Technica’s reporting, researchers around the world collectively devote 15,000 years of work to peer review every year - labour that, if compensated, would cost $1.5 billion for the US share alone. Because that work is donated and anonymous, the system has no mechanism to expand when submission volumes rise.
The consequences are visible in individual cases. Health economist Jason Semprini at Des Moines University had a paper on HPV vaccine mandates rejected after a single reviewer misread the study’s central argument - confusing a policy-effectiveness question with a question about vaccine efficacy itself. Normally two or three reviewers would catch such a misreading, but only one was assigned. Microbiologist Sebastian Lourido from the Whitehead Institute described the overall publishing experience as “extremely protracted and painful.”
The problem is most acute in AI research, which has already shifted toward communicating findings through informal channels such as blogs, sidestepping formal review entirely. Peer review is, in Semprini’s words, “the bedrock of science” - but at this moment, he added, “it’s not standing strong.”
Practitioner takeaway: For developers and product builders who rely on AI research, the retreat from peer review means that influential findings - benchmarks, capability claims, safety results - increasingly arrive via blog posts or preprints with no independent validation. Treating such sources as provisional, cross-checking claims against multiple publications, and staying alert to the difference between a peer-reviewed result and an informal announcement is more important now than it was even a few years ago.
Google · Your sourcesQuantslant higher in your Google - one clickAdd to preferred sources →
More on this from QUANTSLANT
magazine·Zuckerberg’s 6,500-Word Manifesto Calls for Open Superintelligent AI