Opinion: Is Open Access Worth the Cost?

ABOVE: © istock.com, jindu savaliya

On January 11, 2022, the journal Nature Neuroscience announced its article processing charge for open-access papers in 2022. Despite being unchanged from the previous year when open-access (OA) became available for the first time in all Nature Springer family journals, a tweet about the substantial fees of €9,500/US $11,390/£8,290 immediately triggered more than 400 replies, 2,300 retweets, and of course, memes, most of which conveyed a single sentiment: “How dare you charge so much!” This was not the first time scientists protested such steep OA publishing fees, and it likely won’t be the last.

PREVIEW THIS ISSUE

There is a massive range of journal article processing charges (APCs), typically ranging from around $1,000 to more than $10,000. Just five years ago, an APC of approximately $5,000 charged by Cell was considered outrageously high by some and raised questions about how these fees were justified. (See “Opinion: Understanding and Coping With Rising Publication Costs,” The Scientist, September 2017.) Much has happened in the OA marketplace since then. With the European Commission–backed Plan S, which mandates OA publishing for research funded by participating public agencies, going into effect last year, major scholarly publishers introduced OA options to existing journals, created a mirror journal that was completely OA, or both. Under a Transformative Agreement, a subscription-based journal must gradually increase its share of OA papers and meet specific criteria annually. These hybrid journals, though, are usually expensive to publish in.

The million-dollar question is: Are these extremely high APCs reasonable? Because many types of running costs, such as staff wages, scale with the volume of the journal, it is not difficult to grasp that the average cost per article depends on the number of annual publications. A journal that publishes fewer articles needs to charge a higher APC to maintain the same profit margin.

Let’s look at the journal eLife, the only journal published by eLife Sciences Publications, a nonprofit organization that still relies on funding agencies to a significant extent to keep revenues and expenditures balanced and thus may be treated as a fair reference model for analyzing publication costs that are not primarily fueled by APCs. eLife was a no-cost-to-publish OA journal when it was launched in 2012, but five years later, it imposed an APC of $2,500, which was further increased to $3000 in April 2021 (though authors may request a waiver). During 2015–2020, the number of articles published in eLife and the associated expenses steadily increased. By dividing expenditures by the annual number of articles, we can estimate how the cost per article scales with journal volume. Although the amount a publisher spends on development and marketing varies each year, a linear correlation is evident on a five-year scale. (See graph on next page.) This analysis shows that OA journals positioned and running on a level similar to eLife’s can at least be self-sustaining with an APC at somewhere between $2,700 and $4,700 if they publish around 2,000 articles per year. Large OA publishers that made fee breakdowns publicly available, such as Frontiers (which published approximately 85,000 articles across 139 journals in 2021) and MDPI (which published about 240,000 articles in 386 journals that same year), typically charge an APC within this range or lower. (Full disclosure: I am a guest associate editor at Frontiers in Chemistry, which charges and APC of $2,950.)

What researchers are paying for is a certification service—the credibility lent to research when it is published in a peer-reviewed journal.

When we turn our attention back to high-end titles such as the Nature Springer research journals, which each typically publish fewer than 200 articles per year, the linear model estimates a per-article cost range of $7,400 to $9,400. Albeit calculated based on a nonprofit operation model, these astonishing numbers are comparable to estimates made more than nine years ago. Indeed, the controversial APC rates unveiled at the beginning of this year now seem reasonable. A high rejection rate in these journals further increases their running costs. Keep in mind that these journals are not only for-profit, but all the editors are full-time employees rather than academics who work with journals for little or no compensation.

But there is still an unaddressed question. As customers—the scientists who author papers to flesh out the corpus of the OA literature—what are we buying from the publishers?

To come up with an answer, I asked myself why I still seek to publish findings in journals at all, even though file sharing is absurdly easy in this day and age. I can post my manuscripts on a WordPress blog or Reddit with a few clicks or choose a more formal path by leaving them on arXiv or other preprint servers without ever submitting a manuscript to a journal. Either way, there is no publishing cost or paywall at all.

What researchers are paying for is a certification service—the credibility lent to research when it is published in a peer-reviewed journal. This was not financially clear in the pay-to-read era, but such services are backed by a collection of key components, including the journal’s branding/reputation, the editorial and review services, and broader multimedia promotion. In particular, studies in life or physical sciences are usually not easily reproducible by peers without considerable human resources and money. Therefore, rather than seeking post-publication judgment (which, of course, is emerging but still far from mainstream), most researchers want to get a stamp of approval from a recognized journal to certify that their manuscript meets the novelty and quality requirements associated with the brand name. Likewise, readers tend to choose references carrying a trustworthy reputation so they do not have to verify every conclusion on their own benchtop.

SCALABLE COSTS Annual publishing volume, expenditure, and average costs to publish an article in the nonprofit OA journal eLife from 2015–2020. Inflation adjustment to the 2022 level was calculated based on the UK CPI data.

Things are slightly different in computer science, mathematics, and theoretical physics, where openly sharing preprints, source code, and data is a common practice. In these fields, audiences can usually verify and build upon the results published by others via standardized toolsets: programs should run in the same way on any computer, and derivations should follow the same norms in a chosen theoretical framework. Consequently, there is less need for a third-party authority, such as a pricey high-profile journal, to certify the quality of a study. Computer scientists’ discontent with commercial journals eventually led to a boycott of the journal Nature Machine Intelligence, which was founded in 2018.

The subscription model was inherited from the paper media era when communicating knowledge was expensive, and readers chose to pay for what they could read. In this internet age, the ongoing evolution to OA publishing is more compatible with the needs of both authors and funding agencies, at least in experimental sciences. Under this new model, grant money is now used to certify and publish resulting studies in a journal, whether it be society-run or commercial, meaning that funders and employers, not readers, are responsible for the cost of evaluating research.

Before the transformation fully concludes, journals will still court allegations of “double-dipping,” that is, receiving money from both the libraries and authors when OA and paywalled articles coexist in an issue. Several publishers have responded by either reducing the subscription fee or providing discounted or fully waived APCs to institutional subscribers. However, establishing a fee model that satisfies all parties remains a challenge. There is more need now than ever for high-quality certification and validation services provided by publishers. Funders and institutions must recognize the urgency of shifting resources from subscription to publication, and play an active role in negotiation with publishers on fees and benefits.

Most importantly, resources should only be allocated to publishers that offer exemplary service and value in the market. And please don’t forget that scientists’ voices make a difference in decision-making processes, even at large, well-established publishers and funders. By urging grant providers and institutions to allocate sufficient budgets toward publication and providing feedback on each publisher’s service quality, we will be able to navigate this pay-to-publish landscape as a cohesive research ecosystem. 

Jingshan S. Du is a Washington Research Foundation Postdoctoral Fellow at Pacific Northwest National Laboratory. The views he expresses here do not necessarily reflect those of his employer or of the US government. He is an Early Career Editorial Advisory Board member of ACS Biomaterials Science & Engineering and Guest Associate Editor of OA journal Frontiers in Chemistry.