‘A waste of time for all of us’: caught in the crossfire of the peer-review wars

In April 2023, Michael Börsch received what he describes as the laziest peer review in his 30-year career as a biophysical chemist. Reviewer 1 had given detailed feedback on his manuscript about an enzyme that drives the production of the energy-carrying molecule ATP, but Reviewer 2’s comments were, in Börsch’s view, “stupid”.

Reviewer 2 had pasted sections of Börsch’s abstract before listing a series of generic demands. One was “conclusion part is very long, make it brief”, a second asked him to “improve the English grammar” and a third mentioned measurements that weren’t part of the paper. The reviewer also recommended that Börsch and his co-authors include “some latest literature”, pointing to two unrelated studies about Alzheimer’s disease¹,².

“Really, this was a waste of time for all of us,” recalls Börsch, who is at the Friedrich Schiller University Jena in Germany. After revising the manuscript according to Reviewer 1’s remarks and rejecting Reviewer 2’s requests in a thoughtful, if slightly exasperated letter, his paper was published in the International Journal of Molecular Sciences³, which is part of the portfolio of MDPI, a publisher in Basel, Switzerland. Their work was done and dusted — or so Börsch thought.

In December 2023, marketing researcher Maria Ángeles Oviedo-García at the University of Seville in Spain posted on the post-publication peer-review platform PubPeer that language highly similar to that used by Börsch’s Reviewer 2 was also used in 85 reviewer reports for other papers. Oviedo-García included reviewer reports from Börsch’s study and dozens of others in a 2024 paper⁴ published in the journal Scientometrics about review mills — a term she coined to describe individual or groups of scientists who duplicate reviewer reports, probably with the goal of increasing the number of times their or their collaborators’ papers are cited.

Oviedo-García’s findings prompted MDPI to add a yellow warning notice at the top of dozens of affected papers, including Börsch’s study. The note expresses concern that the paper hadn’t been vetted adequately by one of the reviewers and that it was being put through additional peer review. Suddenly, in Börsch’s view, the paper he and his team had worked on for 15 months was invalidated for the wider academic community and was seemingly tainted by allegations of wrongdoing. And yet, we did not make any mistakes, Börsch says. “We did solid science. We did the review process,” he adds, visibly angry as he recounts the story on a Zoom call. How, he wondered, could this have happened?

Slapdash reviewer reports and ones that include unnecessary citation requests have long plagued academic publishing, but over the past few years the production of some has become much more sophisticated. Publishers, including Springer Nature and Taylor & Francis, and the journal PLoS ONE have told Nature that they have observed a worrying increase in reviewer reports with boilerplate language being reused for several manuscripts, and that these reports often also contain irrelevant citation requests. (Nature’s journalism is editorially independent from its publisher, Springer Nature.)

For instance, 0.05% of reviews received by Institute of Physics Publishing (IOPP) journals in 2025 were 100% identical and 0.3% had 80% overlap. One analysis of nearly 150,000 reports in MDPI journals, several PeerJ journals, The BMJ and a selection of computer-science venues found that between 0.5% and 1.3% were highly similar to other reviews, according to work⁵ presented at the 2025 International Congress on Peer Review and Scientific Publication.

Review mills not only threaten to sully the academic record with irrelevant references and compromise the quality of peer review, but are a bane to authors such as Börsch, who are finding themselves caught in the crossfire between review mills, the sleuths that hunt them and the journals scrambling to undo the damage. Understanding review mills requires a deep look at the systemic forces that drive fake reviews and the pressures faced by authors and journals to publish studies as quickly as possible.

Unless solutions are found, review-mill activity will only increase, warns psychologist Dorothy Bishop, an emeritus professor at the University of Oxford, UK, who has been investigating review mills with Oviedo-García. “I think it could be quite a common phenomenon,” she says, “and we’re probably only just [seeing a] tiny bit of it.”

Reviewers under pressure

One reason for the rise in reports by review mills is that publishers are increasingly making reviews public and hence accessible to sleuths like Oviedo-García. That’s why much of her evidence for review mills comes from MDPI journals — and PLoS ONE and PeerJ. “There should be more publishing of peer reviews, precisely because it is one way to be aware of this phenomenon,” Bishop says. The growing use of artificial-intelligence tools probably contributes, too, as they make it easy to produce generic review content.

“MDPI is much better at publishing the peer reviews than most other publishers. So there’s a sense in which it would be rather unfair to treat this as something to bang MDPI over the head with,” adds Bishop. An MDPI spokesperson says that improving transparency was a key reason for the publisher’s 2018 decision to give authors across all its journals the option to publish peer-review reports (reviewers can choose whether their name is public). The practice “is a key factor in ensuring openness”, says the spokesperson. “It boosts transparency and can serve as a means of increasing accountability.”

But specialists and sleuths also blame the well-documented publish-or-perish pressure faced by academics that favours the quantity of publications and citations over the quality of the research. Nature reached out to more than 30 writers of similar or duplicated reviews identified by Oviedo-García. Only one agreed to a substantive interview. Abd Dagestani, a sustainability researcher at Central South University in Changsha, China, says that much of the pressure comes from journals. He conveyed the sense that it was the feeling of being unvalued as a reviewer that led him to duplicate sections of his reviewer reports for several studies sent out by MDPI for review. When, during his PhD, journals were asking him to review papers, he recalls feeling honoured, hoping his contribution could somehow help to improve the work. Some journals — such as many published by MDPI — offered discounts on paper submissions in return for reviewing.

Sometimes, he’d spend a week reviewing papers, only to find that some editors would accept papers regardless of whether the authors had addressed his comments. “An editor did this once or twice, and I was like, ‘okay, so why am I spending time to review the paper?’,” he says. He felt that this was especially unfair, given that scientists are asked to review for free while publishers make money through authors’ work. Meanwhile, the review requests kept mounting; sometimes tallying ten a week from multiple publishers, including MDPI. At one point, he decided to provide more-generic comments on the studies rather than scrutinizing them in depth, he says, and ended up using similar phrasing across several reviews.

In a 2025 post on social-media platform X, Oviedo-García wrote that she had identified two identical reviewer reports written by Dagestani, and two others that were nearly identical, submitted by an anonymous reviewer. Dagestani’s reviews — for 2022 papers in the MDPI journals Sustainability and Processes — used some of the same statements, including “the abstract should briefly describe the research background” and “research gaps should be well mentioned in the introduction”.

Dagestani also suggested citing his own sustainability papers, but he says he only did this so that authors could learn from his studies, thus saving him time in evaluating their work in detail. “It was just a friendly suggestion,” he says. Dagestani was shocked when — “unfortunately”, he says — some authors included all requested references in their paper (the authors of the study in Processes included all six of his suggestions). Perhaps the authors thought they’d have an easier time getting their paper accepted if they pleased the reviewer, Dagestani speculates. Ultimately, however, Dagestani began declining many of the requests; he now reviews only for journals that he feels value his work.

When Nature journalists contacted Dagestani to fact-check details of his account, he ceased to meaningfully cooperate but added, “I don’t duplicate anything”.

When contacted for comment about Dagestani’s account, an MDPI spokesperson said that the publisher values reviewers’ time and expertise and “does not expect or encourage the use of duplicated or generic review reports”. Although the spokesperson recognized that researchers face growing demands on their time because of high submission volumes and review requests across publishers, editors are nevertheless expected to consider reviewer comments carefully and ensure that substantive concerns are addressed. By offering discounts to reviewers on article-processing fees, as well as reviewer awards and formal recognition of reviewing activity, MDPI helps to support reviewers and recognize their contributions. The publisher also works to identify reviewer ‘cartels’ and to flag misconduct. “Oviedo-García’s work serves as an important starting point for discussion among both academic communities and publishers, highlighting the need for a deeper understanding of how aspects of the publishing process can be manipulated,” the spokesperson said.

Publishing patterns

Citations, citations, citations

According to Bishop, most people involved in review mills are seeking to boost their citation numbers. These guide university decisions around hiring, promotions and tenure, and influence university rankings and funding decisions. The more that citation numbers are used as a metric of academic success, “the more people will try to game the citations”, she says.

Evidence for this comes from the observation that some people who provide duplicate reports are also habitual self-citers. Fabian Wittmers, a marine biologist and science sleuth at Oregon State University in Corvallis, mentions the engineer Rabiul Awual as one example of this phenomenon. Wittmers says that Awual’s high citation count — his work has been cited nearly 43,000 times according to Google Scholar — is partly due to a long-standing self-citation habit. In a now-retracted study about caesium removal from waste water⁶, for instance, Awual and his co-authors cited their own research 123 times — Wittmers considers 10–15 self-citations per paper to be the acceptable limit. The retraction notice, issued in October 2025, notes that the citations to Awual’s research, most of which were added at the revision stage, were of “either borderline relevance or no relevance whatsoever to the article”, and calls this practice “a violation of publishing ethics”. The notice also says that the authors disagree with the grounds for the retraction.

When contacted about their reasons for disagreeing with the retraction, co-author Nazmul Hasan asked Nature to refrain from sending them e-mails. Other co-authors, including Awual, did not meaningfully reply to requests for comment, instead responding with hostility and profanity.

Many more citations could stem from Awual’s review practices, too. Public reviews that Awual has written for papers in MDPI journals reveal that he often uses the same statements across reports and includes citation requests to his own research. Wittmers says that he has discovered dozens of studies by other researchers in Elsevier journals that cite large numbers of Awual’s studies. The reviews for those studies aren’t public but Wittmers suspects that these were reviewed by Awual or his collaborators.

One 2022 study⁷ in the journal Food Chemistry about the structural properties of dough in steamed bread, for instance, cites four of Awual’s papers about wastewater treatment. “This is not a one-off situation. There’s multiple papers in Food Chemistry where papers by Awual were cited that make no sense,” says Wittmers.

Wittmers suspects that for Awual and his collaborators, these kinds of review practice might be necessary to get their studies cited, because he deems many of the papers to be of poor quality. For instance, in a 2025 PubPeer post, Wittmers reported evidence of image duplication in six of Awual’s studies. “No one wants to cite these paper-mill products,” Wittmers says. “So you need to request that they get cited. And that’s why you review a lot.”

When Awual was invited to comment on Wittmers’ allegations of self-citation, review milling and image duplication, he asked Nature to stop sending “foolish” e-mails.

One author of a study in an Elsevier journal — who requested anonymity to avoid possible conflicts with reviewers or publishers — says that they felt coerced into citing Awual’s studies because of the anonymous review that they had received. After generic suggestions that the title be rephrased and that the study’s novelty be explained in the introduction, the review contained numerous reference requests. The reviewer concluded with a request to see the revised manuscript, implying — the author thought — that they might recommend rejecting the paper if the requests weren’t fulfilled. “I felt pressured by the reviewer,” the author says.

They say that they’ve asked the journal to review the irregular references and that they’ll alert journal editors if they encounter similar reviewer reports in the future. In general, the author says, they’ve seen a worrying increase in incoherent reviewer reports that seem to be referring to entirely different manuscripts. “It seems that many journals do not perform even minimal checks on the coherence and relevance of peer reviews, leaving scientists vulnerable to inadequate reviewing practices,” they note.

When asked for comment about Elsevier’s handling of reviewer reports, a spokesperson said that the publisher’s policies address citation manipulation by reviewers and that the publisher has training in place to help editors to recognize such practices. “Reviewers should not suggest citations to the reviewer’s (or their associates’) work unless the suggested citations are for genuine scientific reasons,” note the policies. “Editors may remove citations that, in their view, are inappropriate from reviewer comments.”

Why Awual was invited to review for journals outside his scientific expertise is unclear. In general, Wittmers suspects that review mills often make an editor’s job easier, because they struggle to find reviewers among an increasingly overworked academic workforce. Review mills might respond quickly to review requests, helping journals to reduce the time spent on article reviews.

Nature journalists asked Springer Nature for comment on its processes for dealing with coercive peer-review practices and promoting academic integrity. A spokesperson said that the publisher has been working to identify and prevent review-mill activity by examining peer-review content and investigating networks of compromised individuals.

“For cases where we can document repeated attempts to manipulate citation practices, individuals are suppressed from acting as reviewers in future and may be subject to further sanctions, such as being blocked from publishing in our journals,” the spokesperson says. Authors can report coercive citation requests to a journal’s editor-in-chief or to the publisher, and editors should inform their relevant publishing contact, they added.

Mapping citations

To retract or not to retract

Review-mill activity can go unnoticed for years and be tricky to spot, Bishop says. Moreover, many people engaged in such activity use identifiers from the PubMed biomedical-science database, rather than full references to make their citation requests, Bishop adds. This means that overworked editors might not recognize improper requests. People involved in review mills also escape detection by requesting that authors cite their collaborators’ papers rather than their own. This might have been the case with Börsch’s Reviewer 2: chemist Qaiser Mahmood at the Guangdong Laboratory of Chemistry and Fine Chemical Engineering in Shantou, China, who Oviedo-García found to have duplicated reviews. Nature journalists determined that the papers¹,² Mahmood asked Börsch and his colleagues to cite were co-authored by some of Mahmood’s collaborators. Altogether, Oviedo-García has identified 85 similar reviewer reports that Mahmood, his collaborators or anonymous researchers have written for various journals, requesting up to 54 studies to be cited. Mahmood did not respond to requests for comment about his and his collaborators’ review practices.

Suggesting that manuscript authors cite the reviewer’s collaborators’ papers rather than the reviewer’s own could be part of a quid-pro-quo dynamic in which those associates reciprocate in kind when they are reviewers, Bishop says. Such networks can grow into large review cartels. For example, Bishop and Oviedo-García identified a group of at least nine Italian oncologists who used similar review templates to get their work cited in 170 articles⁸. “In this mill, we’re pretty sure an editor is also involved, which helps them allocate papers to friends who will do this,” Bishop says.

Although journals are getting better at detecting review mills, it’s often thanks to science sleuths that journals — and authors themselves — are made aware of this activity. Börsch only learnt of Mahmood’s behaviour through Oviedo-García’s PubPeer post and her Scientometrics paper⁴, which lists 263 reviewer reports with evidence of review-mill activity. Börsch feels that by implicating his study in her paper about misconduct, he and his co-authors were being unfairly criticized even though he — and many others whose papers are included in Oviedo-García’s article — had rejected the reviewers’ citation requests.

Börsch asked Oviedo-García to remove any mentions of his study from her paper, but received no response. Oviedo-García, whose paper doesn’t mention Börsch’s study directly, says that she associated only reviewers with fraud or misconduct. “It is crystal clear stated in the article that the authors are not to blame,” she says.

Börsch asked the editor-in-chief of Scientometrics to retract Oviedo-García’s paper, but his request was declined. MDPI’s publishers refused to remove the yellow label plastered above the online version of his paper. Börsch has also contacted Mahmood, but did not hear back. He has e-mailed his funding agency to see whether there was anything they could do — all to no avail. “For my PhD student, it’s a disaster to have his first paper branded like this,” Börsch says.

When MDPI sent Börsch’s study to another group of reviewers — which it did without the authors’ permission — it seems to have sent out the original manuscript rather than the updated version, in which Börsch had responded to the helpful first reviewer, he says. When one of the new reviewers had an entire series of further requests, Börsch decided that enough was enough. In his view, the post-publication peer review unfairly placed the responsibility on him to correct the scientific record when it was the publisher’s responsibility to ensure that studies were vetted properly. He sees the extra peer review as a “smokescreen” to cover up MDPI’s failure to ensure adequate review. “At one point, I decided not to respond to any of their e-mails anymore,” he says.

To his fury, in June, a representative of the International Journal of Molecular Sciences asked Börsch and his colleagues for an update. If the journal didn’t hear from the authors soon, it “may have to withdraw the manuscript due to loss of contact”, they wrote.

When asked about Oviedo-García’s findings, an MDPI spokesperson says that since the Scientometrics paper⁴ the publisher has introduced measures to prevent similar issues from arising in the future. “We have strengthened oversight of the peer-review process to support robust editorial decision-making, and we are vigilant about monitoring all submissions for potential forms of reviewer misconduct.”

And when asked for comment on Börsch’s views, the MDPI spokesperson acknowledged the broader challenges of assigning blame when correcting the scientific record. “No system is entirely immune to error or systematic forms of misconduct,” they say. Moreover, post-publication actions such as corrections or retractions are crucial for maintaining the integrity of the scientific record, they add. “The purpose of post-publication notices is to inform readers in a clear and transparent manner, rather than to serve as a punitive measure.”

Finally, when asked about the suggestion that MDPI might retract Börsch’s paper, the spokesperson said that “retraction would only be considered in cases where the findings of an article are found to be unreliable, and this has not been the case in any of the investigations undertaken to date”.

Caught in the crossfire

“It’s very, very unfortunate for authors who are actually innocent in this,” Bishop acknowledges. “They are really just used as fodder” and “do the dirty work for the reviewers”, who then get citations. But journals face tricky decisions when dealing with such cases. Is a paper’s quality compromised even if the authors did not cite the requested papers, as in Börsch’s case? Renee Hoch, the head of publication ethics at the publisher PLOS in San Francisco, California, says that the fact that a paper didn’t receive proper review can harm a journal’s trust in a particular article. In cases in which PLoS ONE has identified peer-review integrity issues that affect several studies — which can include review-mill activity — the journal will consider retracting affected papers or taking other action, such as publishing an expression of concern, depending on the nature and severity of the concerns. “We don’t use retraction as a form of punishment. It’s just to correct the published literature, but [authors] perceive it sometimes as either a punishment or as an allegation,” Hoch says.

PLoS ONE also removes people involved in review mills from its active reviewer databases and removes any involved editors from editorial boards. But fully stamping out review mills can be a game of whack-a-mole. Awual and one of his colleagues, for instance, seem to have been banned from reviewing for one Elsevier journal, but Wittmers has seen Awual’s studies being cited in other Elsevier journals, suggesting that he or his collaborators continue to review for some of the publisher’s outlets, Wittmers says.

Awual listed the Dhaka Institute for Materials Science (DIMS) in Bangladesh and the Japan Atomic Energy Agency as his most-recent affiliations. In March, Elsevier retracted a 2024 environmental-chemistry paper⁹ after Wittmers questioned on PubPeer why the paper contained citations to ten of Awual’s studies. According to the retraction notice, “the authors said that these were added in response to the reviewer comments. However, the authors do not appear to have checked or questioned the relevancy of these references to their work.” Elsevier has also retracted some of Awual’s own studies for reasons such as extra affiliations being added for the authors without the editor’s approval.

When asked about actions taken with regard to Awual’s or his collaborators’ review activity, an Elsevier spokesperson said, “this case is currently under investigation, and we cannot disclose further information until the investigation is complete”.

When Nature approached a principal scientist at the Materials Science Research Laboratory at the University of Dhaka, which, according to some of Awual’s papers¹⁰, is the parent institution of DIMS, they responded that “there is no existence of any institute named DIMS associated with the University”. The website dims-bd.org — which one of Awual’s e-mail addresses is registered to — is “under construction”. Nature contacted the Japanese Atomic Energy Agency in Funaishikawa about Awual; a spokesperson says that he is not currently affiliated with the agency and that they’ve requested that publishers remove the incorrect affiliation.

Some journals are working on ways to prevent review-mill activity to begin with. IOPP, for instance, has released a duplicate-review checker to help editors to identify reviewer reports that contain language similar to that of reviews for other IOPP journals. Kim Eggleton, IOPP’s head of peer review and research integrity in Bradford, UK, says that because review mills operate across journals, the providers of peer-review submission systems could incorporate such checks into their platforms so that publishers can detect reviewers who have given duplicate reviews at other journals.

Wholly preventing the issue, however, would require funders, institutions and publishers to address the pressures on scientists that lead them to duplicate reviews and to try to game the citation system.

In the meantime, Börsch says he no longer submits papers to MDPI journals. Although he has given up on trying to remove the yellow notice from his ATP enzyme study, he hopes that it will serve as a warning to other scientists and prevent them from falling into a similar trap. However, for him, “it’s not a nice reminder”, he says. “It just recalls the troubles we have, and that they are not solved.”

Katarina Zimmer is a science journalist based in Berlin.

References

1. Khan, S. et al. Molecules 27, 7368 (2022).
2. Ujan, R. et al. J. Mol. Struct. 1246, 131136 (2021).
3. Pérez, I., Heitkamp, T. & Börsch, M. Int. J. Mol. Sci. 24, 8442 (2023).
4. Oviedo-García, M. Á. Scientometrics 129, 5805–5813 (2024).
5. Labbé, C., Hubert, G., Swart, W. & Cabanac, G. In 10th Int. Congr. Peer Rev. Sci. Publ. (American Medical Association, 2025).
6. Hasan, N., Shenashen, M. A., Hasan, M., Znad, H. & Awual, R. Chemosphere 270, 128668 (2021); retraction 387, 144640 (2025).
7. Guo, J. et al. Food Chem. 368, 130853 (2022).
8. Oviedo-García, M. Á., Aquarius, R. & Bishop, D. V. M. Account. Res. 33, 2640012 (2026).
9. Wang, Y. et al. Environ. Res. 251, 118734 (2024); retraction 294, 123816 (2026).
10. Alotaibi, A. N., Alsafrani, A. E., Albishri, A., Awual, M. R. & Rahman, M. M. Chem. Eng. J. 522, 168056 (2025).

Credits

• Author: Katarina Zimmer
• Illustration: Andreion De Castro
• Infographics: James Bayliss
• Data analysis: Vera Nienaber
• Art direction: Wes Fernandes
• Subeditor: Francisca Schultz
• Editor: Jack Leeming

Springer Nature