One academic paper’s journey through the mill

In the summer of 2020, Anna Abalkina came across an advertisement online: a website was selling authorship of a chemistry paper that it said was to be published in a reputable journal in February 2021.

The ad didn’t come as a surprise to Abalkina, a sociologist at the Free University of Berlin, because she had been tracking such advertisements for a while. She found this instance on 123mi.ru, a website that claims to have brokered tens of thousands of deals for authorships in journals indexed by leading databases, such as Web of Science and Scopus.

123mi.ru is known as a paper mill, which, for a fee, lets academics add their names to a research paper that is already accepted for publication in a peer-​reviewed journal. Such papers are a way for unscrupulous professors to boost their publication record with little effort.

Paper mills are a scourge for publishers of academic journals, which must also contend with manipulation, falsification and fabrication of data, plagiarism, and honest errors. Most publishers want to stop the paper mills, but the mills always seem to be one step ahead.

Since it surfaced in 2018, 123mi.ru claims to have brokered more than 20,000 authorship slots in over 4,000 scholarly papers. It sells a slot for up to $5,000 each, with first authorships usually the most expensive.

The site is one of many paper mills that churn out subpar studies. Abalkina’s understanding of 123mi.ru’s process is that it usually instructs ghostwriters to pen a paper only if there is interest in the topic.

Completely fake papers are often written by ghostwriters in languages other than English—often Russian—and then translated. In some cases, paper mills advertise plagiarized English versions of actual studies that were published in non-English-​language journals. And the mills occasionally market authorship slots on real papers written by one of the listed coauthors.

The 123mi.ru ad for this particular chemistry paper, about the binding of the drug sulfanilamide with zinc oxide, seemed to be an example of the last. The ad contained a note that one authorship slot was reserved for the researcher who authored the paper; it also held lots of clues for Abalkina’s sleuthing, including how many authors the final paper would have, its month and year of publication, and the abstract the final paper would carry.

Moreover, the paper was unique because 123mi.ru doesn’t usually advertise papers that are originally written in English, Abalkina says. This paper was also easier to track than most because the ad contained a whole abstract. “This advert is a special one,” she says. “There are not so many offers like that.”

Ads such as these boast that the final paper will be indexed in prominent databases like Clarivate’s Web of Science and Elsevier’s Scopus—a prerequisite for career advancement in some countries. Earlier this year, Web of Science delisted 19 journals run by the open-access publisher Hindawi for several reasons, including paper mill activity.

Abalkina entered the ad for the drug-binding paper to her tracking spreadsheet. She easily found the paper in 2021, when she was hunting for published studies corresponding to the ads she had collected.

In December 2021, Abalkina published a preprint—an article published before peer review—chronicling the hundreds of papers she was able to match with ads on 123mi.ru. As part of her study, she analyzed more than 1,000 offers advertised on 123mi.ru. The latest version of her study is due to be published in Learned Publishing.

Different publishers have different attitudes to her findings, Abalkina says. Some regularly refer to her spreadsheet, carry out investigations, and retract papers. Others are slower to take action, and when they do, they don’t mention paper mills in retraction notices.

Abalkina found the drug-binding paper published in the Journal of Sulfur Chemistry, which is run by Taylor & Francis, a British academic publisher that is part of the publishing and conference company Informa. The publishing giant Elsevier also hosts the paper on its ScienceDirect platform as part of its pilot program for third-party publishers.

C&EN emailed one of the study’s coauthors, Pengfei Luo of the University of Science and Technology Beijing, but didn’t receive a response. An email to corresponding author Tianying Yang of Northeastern University in China bounced back. C&EN couldn’t find contact details for the last author, A. Sarkar of the Indian Institute of Science. 123mi.ru didn’t reply to a request for comment.

A Taylor & Francis spokesperson says in an email that the paper is being investigated by a publication ethics and integrity team. “The team are following our established investigation processes and any action taken will be informed by Committee on Publication Ethics (COPE) guidelines. We are not able to comment any further on this case while the investigation is ongoing,” the spokesperson says.

An Elsevier spokesperson says in an email, “If Taylor and Francis did decide to investigate and subsequently retract the article it would automatically be removed from Science Direct.”

Taylor & Francis and Elsevier are part of a group of publishers that are testing or using a tool launched earlier this year to help them detect papers produced by paper mills. The tool was developed by the International Association of Scientific, Technical, and Medical Publishers (STM), a trade group with more than 100 members. The American Chemical Society, which publishes C&EN, is a member.

Hylke Koers, chief information officer for the STM’s technical operations arm in the Netherlands, says the organization is in the process of launching a pilot that will help find simultaneous submissions of papers to a number of journals. “We understand that paper mill material is often submitted to multiple journals, multiple publishers at the same time,” he says.

Bernhard Sabel, a neuropsychologist at the University of Magdeburg who built his own fake-paper detector, says paper mills need to be prevented from tainting scholarly literature. “It’s in the interest of society to have clean science rather than polluted science,” he says.

Publishers that want to stop mill-​generated papers can look for red flags, including compromised peer review, a low likeliness of collaboration between authors, and odd turns of phrase.

Paper mill studies are often plagued with bogus or rigged peer review. Rigged review occurs when authors submitting papers suggest fictitious peer reviewers with fake email addresses. These reviewers are difficult to verify as they usually have personal email addresses; universities in some countries don’t have institutional ones. Those posing as peer reviewers typically offer glowing feedback.

Abalkina says one solution to this problem is to stop allowing academics to recommend reviewers for their work. Previous studies on the topic conclude that author-suggested reviewers are more lenient than those chosen randomly by journals.

When there’s evidence that a paper likely originated from a paper mill, journals and publishers should flag it quickly—without waiting to hear from authors, says Jennifer Byrne, an oncologist at the University of Sydney who has worked extensively on the problem of errors in published genetic sequences. If such papers are indeed found to have been associated with paper mills, she adds, they should be retracted.

“Responses have to be hard, tough, and fast. At the moment we see none of those things,” Byrne says.

Byrne contributed to part of another tool, the Problematic Paper Screener, that identifies errors in genetic sequences. This online tool, created by concerned researchers, also searches for so-called tortured phrases, which are strange turns of phrase for established terms. Such phrases are probably introduced by paraphrasing software to circumvent plagiarism checkers.

In 2021, when the tool was launched, thousands of papers contained these tortured phrases. Now, research integrity sleuths say, paper mills are getting smarter, and fewer studies containing such turns of phrase are being published. Similarly, Byrne notes, some once-common mistakes in genetic sequences stopped appearing after she drew attention to them.

Paper mills are also churning out better papers, according to Byrne. Some “don’t have mistakes in them,” she says.

These improvements at paper mills mean publishers and journals may need to use other indicators of poor quality or fabricated research. One potential red flag can be collaboration between researchers in different parts of the world, but Byrne cautions that such unlikely alliances have also been shown to be a defining feature of highly innovative research.

Recent years have seen enhancements in translation and paraphrasing programs. As artificial intelligence gets better, “the capacity of paper mills to generate more plausible and error-free manuscripts will increase,” Byrne says.

She notes the literature probably already contains many papers masquerading as real science that actually originated in paper mills. As time goes on, “that distinction between the invented and the real will converge,” Byrne says. “I’m very worried about that.”

Dalmeet Singh Chawla is a freelance science journalist based in London.