E-Paper
Fraudulent scientific papers are booming
A subset of journal editors may be partly responsible
SCIENTIFIC JOURNALS exist to do one thing: provide accurate, peer-reviewed reports of new research to an interested audience. But according to a paper published in PNAS on August 4th, that lofty goal is badly compromised. Scientific fraud, its authors conclude, happens on a massive scale and is growing quickly. In fact, though the number of scientific articles doubles every 15 years or so, the number thought to be fraudulent is doubling every 1.5 years (see chart).
That state of affairs is enormously harmful to science, argues Luís Nunes Amaral, a physicist at Northwestern University in Chicago and the study’s senior author. Without better guardrails, researchers will not be able to trust the scientific literature and might end up wasting time and grant money trying to repeat fraudulent experiments. If nothing is done, “The scientific enterprise in its current form would be destroyed,” he says.
It has long been clear that publication fraud rarely comes from lone fraudsters. Instead, companies known as paper mills prepare fake scientific papers full of made-up experiments and bogus data, often with the help of artificial-intelligence (AI) models, and sell authorship to academics looking to boost their publication numbers. But the analysis conducted by Dr Amaral and his colleagues suggests that some journal editors may be knowingly waving these papers through. Their article suggests that a subset of journal editors are responsible for the majority of questionable papers their publications produce.
To arrive at their conclusion, the authors looked at papers published by PLOS ONE , an enormous and generally well-regarded journal that identifies which of their 18,329 editors is responsible for each paper. (Most editors are academics who agree to oversee peer review alongside their research.) Since 2006 the journal has published 276,956 articles, 702 of which have been retracted and 2,241 of which have received comments on PubPeer, a site that allows other academics and online sleuths to raise concerns.
When the team crunched the data, they found 45 editors who facilitated the acceptance of retracted or flagged articles much more frequently than would be expected by chance. Although they were responsible for the peer-review process of only 1.3% of PLOS ONE submissions, they were responsible for 30.2% of retracted articles.
The data suggested yet more worrying patterns. For one thing, more than half of these editors were themselves authors of papers later retracted by PLOS ONE . What’s more, when they submitted their own papers to the journal, they regularly suggested each other as editors. Although papers can be retracted for many causes, including honest mistakes, Dr Amaral believes these patterns indicate a network of editors co-operating to bypass the journal’s usual standards.
Dr Amaral does not name the editors in his article, butNature, a science magazine, subsequently made use of his analysis to track down five of the relevant editors. PLOS ONE says that all five were investigated and dismissed between 2020 and 2022. Those who responded toNature’s enquiries denied wrongdoing.
Compelling as Dr Amaral’s analysis is, it does not conclusively prove dishonest behaviour. All the same, the findings add to a growing body of evidence suggesting some editors play an active role in the publication of substandard research. An investigation in 2024 by RetractionWatch, an organisation that monitors retracted papers, and Science, another magazine, found that paper mills have bribed editors in the past. Editors might also use their powers to further their own academic careers. Sleuths on PubPeer have flagged papers in several journals which seem to be co-written by either the editor overseeing the peer review or one of their close collaborators—a clear conflict of interest.
Detecting networks of editors the way Dr Amaral’s team has “is completely new”, says Alberto Ruano Raviña of the University of Santiago de Compostela in Spain, who researches scientific fraud and was not involved with the study. He is particularly worried about fake papers remaining part of the scientific record in medical fields, where their spurious findings might be used to conduct reviews that inform clinical guidelines. A recent paper in the BMJ , a medical journal, found that 8-16% of the conclusions in systematic reviews that included later-retracted evidence ended up being wrong. “This is a real problem,” says Dr Ruano Raviña.
Yet the incentives for fraud continue to outweigh the consequences. Measures including a researcher’s number of publications and citations have become powerful proxies for academic achievement, and are seen as necessary for building a career. “We have become focused on numbers,” says Dr Amaral. This is sometimes made explicit: staff at Indian medical collegesare required to publisha certain number of papersin order to progress. Some journals, for their part, make more money the more articles they accept.
Breaking either trend will take time. In the meantime, publishers are rolling out new screening tools for suspicious content, including some which spot “tortured phrases”—nonsensical plagiarism-evading paraphrases generated by AI models such as “colossal information” instead of “big data”—or citations in the wrong places.
There is also increasing pressure on publishers to root out bad papers. Databases of reputable journals, such as Scopus or Web of Science, can “de-list” journals, ruining their reputations. It’s up to the publishers to bring about a relisting, which means tidying up the journal. “If we see untrustworthy content that you’re not retracting, you’re not getting back in,” says Nandita Quaderi, editor-in-chief of Web of Science. But whether publishers and the many editors who work hard to keep bad science out of their journals can keep up with the paper mills remains to be seen.
