In June 2024, Leigh-Ann Butler, Stefanie Haustein and colleagues posted a paper describing an open dataset of article processing charges, the fees many journals ask authors to pay so that a paper is free for everyone to read [1]. They gathered the price lists of six large publishers (Elsevier, Frontiers, PLOS, MDPI, Springer Nature and Wiley) for every year from 2019 to 2023 [1]. The result covers 8,712 journals and 36,618 journal-year entries, and anyone can download it [1].
That kind of openness about prices is welcome. What the numbers show is less comfortable. Open access has done a lot for readers. My worry is what happens on the other side of the paywall, where the question quietly shifts from whether you can read a paper to whether you can afford to publish one.
My view is simple: paying per article is one way to fund open access, and it should not be the default. When a fee rather than the work starts to decide where a paper goes, we have moved the barrier, not removed it.
What the price lists show
Averaged across the journals and years in the dataset, the list fee was $1,977 in fully open (gold) journals and $3,137 in hybrid journals, which are subscription journals that let authors pay to make one paper open [1]. The authors point out that hybrid fees run higher even though those journals are already paid for through subscriptions [1]. At the top end, 37 Nature journals listed $11,690 in 2023 [1].
Prices also moved in one direction. Of the journals with a fee in both 2019 and 2023, 89% raised it, and 40% raised it by more than the 19% inflation over that period [1].
Imagine a young researcher in a country where a month’s salary is well below $2,000. Her results are solid, and the journal that fits them best is open access. The paper is ready, but the invoice is bigger than anything her lab can pay.
That scenario is not about the quality of her work. It is about a price tag that sits between a finished paper and its readers. For a well funded lab, $3,000 is a line in a grant budget. For many others, unless a waiver or an institutional deal covers it, it is a wall.
Sticker prices are not the whole story
I want to be careful here, because list prices are easy to misread. The dataset records what publishers advertise, not what authors actually pay. The authors say so plainly: it “does not indicate waivers or discounts granted to individual authors,” because that information lives only in invoices and in the records of universities and funders [1].
Some fees are waived outright. In the dataset, 123 journal-year entries listed a fee of $0, which the authors explain could be a temporary waiver for marketing or a journal whose costs are covered by someone else [1]. Some authors never pay out of pocket at all, because a university or funder covers the fee [1]. Some institutions go further and sign read-and-publish agreements with publishers, and the authors suggest their data could help work out the costs and value of those deals [1].
So the honest version of the problem is not that everyone pays $3,000. It is that whether you pay, and how much, depends on where you work and who funds you.
Who can afford it
That is where the evidence gets pointed. In a 2023 paper, Thomas Klebel and Tony Ross-Hellauer analyzed 1.5 million articles in open access journals and found that fee levels were strongly stratified by scientific field and by the country of the authors’ institutions [2]. After accounting for field and country, how well resourced an institution was had small to moderate effects on the level of those fees, and those effects were largest in countries with low gross domestic product (GDP) [2].
Put simply, an institution’s resources were linked to the fees attached to its authors’ open access work, and that link was strongest in countries with low GDP. Klebel and Ross-Hellauer describe growing evidence that these fees “impede researchers with fewer resources” from publishing open access, and they call for alternative publishing models [2].
Waivers help, but to me they look like a patch. They depend on a publisher’s policy and on an author knowing to ask. Institutional deals help too, but I suspect they mostly help authors at institutions with the money to negotiate them.
The case for fees, and its limits
The other side deserves a fair hearing. Fee-based open access made a huge amount of research free to read, including for patients, teachers, and scientists at institutions that could never afford every subscription. Publishing also has real costs: editing, peer review management, typesetting, hosting and long-term preservation do not happen for free.
The best-known alternative shows exactly how real those costs are. Diamond open access journals are free for both readers and authors [3]. A 2021 study commissioned by cOAlition S, with support from Science Europe, estimated there are about 29,000 of them, publishing roughly 356,000 articles a year compared with about 453,000 in fee-based open access journals [3]. But most are small, publishing fewer than 25 articles a year [3]. In the study’s survey of 1,619 journals, 60% said they depend on volunteers, only just over 40% reported breaking even, and 25% reported a loss [3].
In the same survey, 70% of journals declared annual costs under 10,000 dollars or euros [3]. That number stays with me. Many journals run for a year on less than a handful of top-end article fees. The costs are real, but they are not always as large as the prices suggest.
The report also found that since 2018 the share of articles in diamond journals has been shrinking as fee-based journals grow [3]. That is the trend I would most like to see reversed.
What I would like to see change
As a researcher, I think a fee should never be what decides where a paper goes. The fit of the work should decide that.
Free reading was worth fighting for, but charging authors per paper just moves the paywall to the other side. I would rather see funders and universities pay for the publishing infrastructure directly, so that the work, not the budget, decides where it appears.
In practice, that points to a few steps. Funders and universities can put steady money into diamond journals and the shared platforms they run on, instead of paying fees one article at a time. Publishers can make waivers automatic and visible, rather than something authors must discover and request. And when institutions sign publishing deals, the terms and prices should be public, so that researchers and datasets like this one can check whether the money is well spent. The 2024 dataset is a good start at that kind of transparency, and I hope the rest of the system follows it.
References
[1] L.-A. Butler, M. Hare, N. Schönfelder, E. Schares, J. P. Alperin, and S. Haustein, “An open dataset of article processing charges from six large scholarly publishers (2019-2023),” arXiv:2406.08356, Jun. 2024, doi: 10.48550/arXiv.2406.08356.
[2] T. Klebel and T. Ross-Hellauer, “The APC-barrier and its effect on stratification in open access publishing,” Quantitative Science Studies, vol. 4, no. 1, pp. 22-43, 2023, doi: 10.1162/qss_a_00245.
[3] J. Bosman, J. E. Frantsvåg, B. Kramer, P.-C. Langlais, and V. Proudman, “OA Diamond Journals Study. Part 1: Findings,” Zenodo, Mar. 2021. [Online]. Available: https://zenodo.org/records/4558704
Related Articles
