Graham Kendall studies academic publishing. He recently ran the numbers on what peer review costs the UK higher education sector. His calculations are worth quoting:

“In the UK, there are about 46,500 higher education academics. The total wage bill is about 11.5 billion. If 50% of those academics are active peer reviewers and they review 5 papers each year and each one takes two hours, the cost to the universities is 27.65 million. 

Two hours per paper is already a pretty conservative estimate. Most reviewers spend considerably more, and anyone who has worked through a paper with a comprehensive literature review knows that two hours barely covers a first read. So £27.65 million is probably the bare minimum of what universities are collectively subsidising when academic staff do peer review.

And that figure comes on top of what universities already pay publishers. Now we might want to argue that many of us do peer review in our ‘own time’, but it’s worth [pausing for a minute to hang into the logic of Kendall’s argument.

The academic publishing sector is a highly profitable business. One recent analysis (November, 2025) said 

“The dominant four are Elsevier, Springer Nature, Wiley and Taylor & Francis, which collectively generated over US$7.1 billion in revenue from journal publishing in 2024 alone, and over US$14 billion in profits between 2019 and 2024. Their profit margins have always been over 30% in the last five years, and for the largest publisher (Elsevier) always over 37%.”

Using North American as the ur case, the researchers went on to note that this was a significant proportion of the total amount spent on research. 

The cruel irony here is that this highly profitable academic publishing industry is subsidised at every stage by the sector it ‘serves’. Institutions pay subscription fees to access paywalled journals. They pay open access charges so that publicly funded research can be read without said paywall. And then we academics do the editorial and reviewing labour, generally for free, that makes the journals worth subscribing to in the first place. And institutions are locked in to this publishing economy with audit, reputational and human relations systems built around publications.

Now a new layer of unpaid expectations is being quietly added. Reviewers are increasingly being expected to assess the quality of a paper’s argument and evidence AND to check whether it has been written by an AI and whether its references are real. The last of these is not a trivial task. If you take three minutes per reference and an average paper has twenty references (and most have considerably more), you are adding at least an hour to the review, realistically much more. That is before you even read a word of the actual argument.

Editors of journals say it is already hard to find reviewers. Adding AI detection and reference verification to the reviewer’s job description will put off people who might otherwise agree to review. And it will wear out those who do. Neither outcome is good for the quality or speed of peer review, both of which are already under serious strain.

So the question all this raises is –  where does the publisher’s responsibility sit in reference checking? A pre-editorial check for  ghost references, run before a paper is sent to reviewers, seems like a minimum. Publishers have the resources, the technical capacity, and the commercial incentive to do this. They are not already stretched volunteers fitting a review around a teaching load. If a paper arrives with fabricated references, the publisher’s editorial office is better placed to identify that than a reviewer who agreed to give two or more hours to scholarly citizenship.

I am not saying peer review is not important. But I am saying that the scope of peer review seems to be expanding in ways that serve publishers more than it serves scholarship. The reviewer’s job is to assess whether a piece of work makes a genuine contribution to the field and has been conducted ethically and rigorously. Checking whether the references exist is arguably quality control which ought to sit with the organisation that profits from the work.

Publishers are bound to push back on reference checking. Detection tools are imperfect. Reference verification at scale might be labour-intensive. But these are operational problems for a very well-resourced industry to solve, not reasons to keep offloading the work onto academics who are already donating a lot to our very own multinational publishing succubus.

And can we expect our institutions to argue this case? Don’t answer, it’s just a rhetorical question.

Photo by Jonny Gios on Unsplash