Editor’s Note: Today’s post is by Christos Petrou, founder and Chief Analyst at Scholarly Intelligence. Christos is a former analyst of the Web of Science Group at Clarivate and the Open Access portfolio at Springer Nature. A geneticist by training, he previously worked in agriculture and as a consultant for Kearney, and he holds an MBA from INSEAD.

Until recently, MDPI and Frontiers were known for their meteoric rise. At one point, powered by the Guest Editor model, the two publishers combined for about 500,000 papers (annualized), which translated into nearly USD $1,000,000,000 annual revenue. Their growth was extraordinary, but so has been their contraction. MDPI has declined by 27% and Frontiers by 36% in comparison to their peak.

Despite their slowdown, MDPI and Frontiers have become an integral part of the modern publishing establishment. Their success reveals that their novel offering resonates with thousands of researchers. Their turbulent performance, however, shows that their publishing model is subject to risk, and its implementation should acknowledge and mitigate such risk.

IJERPH’s freefall

MDPI’s largest journal, the International Journal of Environmental Research and Public Health (IJERPH), was delisted from the Web of Science (WoS) in March 2023. What came next was both expected and staggering. The journal shrunk by 88% in comparison to its peak. It published about 1,800 monthly papers in the three-month period from December 2022 to February 2023, and it dropped to about 200 monthly papers in summer 2023.

IJERPH’s decline is a stark reminder that in the absence of an Impact Factor, other journal traits matter little. Fast publishing, a high acceptance rate, and a low APC are unattractive to authors if they are not accompanied by a good (or in some cases, any) Impact Factor and ranking.

Contagion, silos, and valuations

While the demise of IJERPH was not surprising, the implications for MDPI’s broader portfolio were harder to foresee. A month after IJERPH’s delisting, the rest of MDPI’s journals declined by 17%, from 28,500 papers in March to 23,500 in May. Since then, they have published fewer papers on every month, dropping to 22,500 papers in August.

Including IJERPH, MDPI declined by 27% in summer 2023 in comparison to its peak. Summer 2023 also marks the first year-on-year decline for MDPI (5%) in a long time. In addition, the decline has been similar across regions and paper types (Guest Edited papers and regular papers).

When asked, MDPI’s CEO, Stefan Tochev, suggested that ‘while there was a drop in publication numbers in April… during the following months MDPI submissions and publications have been steady’. Stefan also mentioned that MDPI has ‘implemented methods to improve the relevance of publications’. This should reduce the risk that another journal is delisted for being out of scope, as in the case of IJERPH. Given the currently accessible information, discerning the impact of author flight versus the impact of the new editorial policy on MDPI’s slowdown is a challenge. However, given the interchangeability of MDPI’s journals, it seems improbable that the new policy has significantly contributed to the slowdown.

The contagion across the MDPI portfolio reflects a source of risk that is unique to publishers with relatively uniform portfolios. When all journals are similarly branded and share the same editorial and operational policies, a journal’s reputation reflects the reputation of the entire portfolio and vice versa. Trouble in one journal may be viewed negatively by authors of other journals. If IJERPH got delisted by WoS, is there any guarantee that other journals will not have the same fate? Does the perceived reduction in IJERPH’s reputation translate to a similar reputational drop for MDPI’s other journals?

It is unlikely that the same extent of contagion would be observed in other large publishers, such as Elsevier and Springer Nature. Their multiple, uniquely branded portfolios and journals can act as silos in case of reputational challenges in one or more journals. A problem in a Cell-branded journal would not likely have much impact on the perception of a journal from The Lancet, nor would an issue with a BioMed Central journal hurt the reputation of a Nature Portfolio title. Such silos pose other challenges for traditional publishers, as for example, it becomes more difficult to implement policies and programs across their portfolios, but at least they limit the damage from reputational issues.

This raises interesting questions about the financial valuation of uniform, fully-OA publishers. The APC OA business model is publication volume-driven, so a drop in papers published means an immediate drop in revenue. Further, uniform portfolio publishers may be at higher risk for reputational damage done by one journal than diverse, traditional publishers. If your profit is less reliable and is subject to the whims of indexing services, then there may be less value presented than the better protected profits of diverse portfolios. Elsevier and Springer Nature can predict their income in the coming 3-4 years with high accuracy. MDPI could get it wrong by year’s end.

There is also a discussion to be had about the capacity of indexing services to wipe out more than $1,000,000,000 of an organization’s market cap (a figure arrived at using the price that Wiley paid for Hindawi as a yardstick for MDPI). Such consequential decisions require transparent and fit-for-purpose decision mechanisms from indexing services and their metrics.

Reputation comes knocking

When I first wrote about MDPI in summer 2020, their reputation was not stellar, but it was improving. It would take several paragraphs to list all the reputational challenges that MDPI has faced since then. They have encountered institutional challenges, such as the inclusion of several of their titles in the Warning List by the Chinese Academy of Science. And they have faced a steady stream of negative press in the form of typically poorly researched blog posts and scholarly articles.

However, nothing seemed to stick. Up until the moment of IJERPH’s delisting, MDPI was successful despite the attacks to its reputation. Its annual growth was about 35% in the months before the delisting, and the publisher was on track to exceed 350,000 papers in 2023, bringing them close to the size of the second largest publisher, Springer Nature.

But now it looks as if questions about MDPI’s reputation have begun to catch up with it. First, the delisting of IJERPH likely is a consequence of MDPI’s reputation. It is possible that WoS would not have taken such aggressive action if MDPI was viewed in better light. Then, the wide contagion across the MDPI portfolio is also likely a consequence of the publisher’s reputation. It is possible that authors would have treated the delisting of IJERPH as a one-off event, if MDPI was viewed with less skepticism across the scholarly world.

MDPI does not deserve the entirety of its poor reputation, but it has not been proactive in its defense. When MDPI reacts to negative developments, it appears unconvincing. For example, their response to a paper by Oviedo-Garcíathat discussed self-citation rates of MDPI’s journals included a chart that showed MDPI standing out for its intra-publisher self-citation rate, in essence supporting the author’s argument (the article has since been retracted and revised). Meanwhile, MDPI’s self-citation rates are high but unspectacular, and certainly not indicators of predatory behavior.

MDPI’s poor reputation is partly the result of inattentiveness. And if papermills are ever discovered to have operated at MDPI en masse, the root cause will likely be the same, namely inattentiveness.

What does this mean for MDPI’s future performance? There are two extreme scenarios that can combine in various ways. The bad scenario is a vicious circle of declining reputation and administrative action against MDPI that leads to declining paper output. Having previously benefitted from a virtuous cycle of improving reputation, MDPI will witness the other side of the coin.

The good scenario is a brand refresh that combines with the short memories of forgiving academics and the ranking of several MDPI journals in next year’s Impact Factor release. This could place MDPI back on the growth track.

In any case, MDPI’s performance in the short-term will be beyond its control. No matter how hard it works to improve its reputation, it will not be able to stop another delisting from a major index or a regional anti-MDPI policy. Meanwhile, it appears to be temporarily stabilizing just above 250k annual papers.

Frontiers, same but different

As with MDPI, Frontiers is a fully-OA publisher that is based in Switzerland, has a uniform portfolio, relies on the Guest Editor publishing model, and grew phenomenally up until 2022. There are some differences though. For example, Frontiers publishes papers fast, but slower than MDPI, it has a higher reliance on the Chinese market than MDPI, it charges higher APCs than MDPI, and its journals are slightly better ranked than those of MDPI.

Another difference between MDPI and Frontiers has to do with their reputation. While MDPI is not in good control of the narrative, Frontiers has a tighter grip on it. For example, they highlight, on seemingly every possible opportunity, that they are the 3^rd most-cited among large publishers (which is true in some ways, but not in every way), and they run a series of reputation-building initiatives such as the Frontiers Forum, Policy Labs, and Frontiers for Young Minds.

Instead of getting caught in unflattering online exchanges, Frontiers’ founder, Kamila Markram, has received a series of glamorous awards. And in Frontiers’ flagship event in 2023, the Frontiers Forum Live, the line-up of speakers included former secretary-general of the United Nations Ban Ki-moon, historian and author Yuval Noah Harari, primatologist Jane Goodall, and former vice president of the United States Al Gore. Such an event might not be to everyone’s taste, but it certainly makes for good PR.

In the face of clearcut misconduct, WoS would not hesitate to penalize the journals of any publisher no matter their reputation. But it is possible that for less egregious misconduct, such as a journal publishing papers out of scope, a publisher with the reputation of Frontiers would be given the benefit of the doubt, when a publisher with the reputation of MDPI was not.

The plot thickens

Given the similarities between the two publishers, one would expect Frontiers to benefit from the slowdown of MDPI. Instead, Frontiers has experienced a protracted slowdown of its own, starting in October 2022. While the trigger of the contraction of MDPI (and Hindawi) is easy to pinpoint, the thread related to Frontiers’ contraction is harder to untangle.

In September 2022, news broke out that Hindawi planned to retract about 500 papers in 16 journals. Hindawi’s announcement was followed by an initial drop of Frontiers’ paper output. Hindawi’s first round of retractions was followed by more retractions and the delisting of 19 of its journals from WoS in March 2023, when MDPI’s IJERPHalso got delisted. This was followed by another drop of Frontiers’ paper output (6,600 papers in April 2023 was its lowest output in 26 months). In the three-month period from May to July 2023, Frontiers’ paper output was 36% lower than its peak in Q3 2022.

Hindawi and Frontiers had something in common before the retraction news surfaced: the wholesale application of Guest Edited collections and a high exposure to papers from China, where perverse incentives have resulted in papermill activity. Much of Frontiers recent growth had come from that region; at the time of the announcement, about half of Frontiers’ papers had authors based in China. The retractions at Hindawi might have made Frontiers uneasy. Frontiers’ ensuing slowdown was the steepest in China, as the country’s paper contribution dropped by half.

The timeline of events suggests that Frontiers self-moderated its paper output by applying additional checks and stricter editorial criteria. When its paper output first declined by 17% from Q3 to Q4 2022, the output of MDPI grew by 19%, suggesting that Hindawi’s retractions did not deter authors from the fully-OA route, and the contraction at Frontiers was the publisher’s choice. While the contraction at MDPI has been geographically agnostic, the contraction at Frontiers has been highest in China, where it had an unhealthy exposure, again indicating a choice rather than market dynamics. Finally, the contraction of Frontiers has been steeper than that of MDPI, despite Frontiers not facing a direct reputational challenge as MDPI, pointing once more at a self-moderation choice.

When asked, Tom Ciavarella, Head of Public Affairs and Advocacy, North America for Frontiers, shared a statement that largely aligns with the conclusions of the analysis. ‘The publishing market has been cooling since 2021 and there will be many reasons for that, not least the economic overhang of the pandemic. In that context, all research publishers are seeing slower article submissions. At the same time, the threat of fraudulent science is growing, exploiting goodwill and eroding trust, so we have responded with even tighter thresholds for acceptance and we are rejecting a higher proportion of papers than ever.’

Controlling the narrative

Why would Frontiers moderate its paper output? The publisher is likely focusing on its brand, sacrificing short-term profitability to stay in control of the narrative. It has taken the time to pursue the right editorial action to protect its reputation and eventually return to growth.

We might already be seeing the fruit of this intervention. In a recent post, Frontiers announced the retraction of 38 papers linked to the practice of authorship-for-sale. Such schemes rely on authorship changes after submission. In response, Frontiers announced policy changes, whereby ‘requests for authorship changes will only be granted under exceptional circumstances’.

It remains to be seen whether these retractions conclude a period of introspection or whether they are the pebble that starts a landslide as in the case of Hindawi. Either way, this is in line with Frontiers attempting to stay in control of the narrative. As their statement concluded, ‘like any business, scientific publishers live and die by their reputation, and any publisher that prioritizes profit over their customers is likely to fail’.

Meanwhile, authorship-for-sale papers have been flagged at MDPI since September 2021, but MDPI does not appear to have taken action in a timely manner to the frustration of the researchers that flagged the papers in the first place.

The unholy grail of publishing

The publishing model of MDPI and Frontiers has resonated with thousands of researchers. The two publishers offer a high likelihood of acceptance and speedy publication in well-ranked journals, typically coupled with peer recognition (and pressure) when publishing in guest-edited collections.

MDPI and Frontiers are not the ideal destination for every paper of every author, but at one point, they combined for about 500,000 papers (annualized), while growing at a rapid pace. This translates into nearly $1,000,000,000 annual revenue, powered by papers in Guest Edited collections.

Although the Guest Editor model is associated with high reputational risk and unpredictable performance, publishers will not let an opportunity of this size drop. Instead, they will keep tweaking the model until they find the right formula. By the time that one or more publishers achieve sustainable success, a few will have tapped out.

There are several contenders facing a diverse array of challenges. MDPI will have to prove that it is adaptable and not a one-trick pony. Frontiers will have to prove that it did not inadvertently expose itself to unmanageable risk before tweaking its editorial policies. And newcomers to this publishing model, such as Springer Nature’s Discover series and Taylor & Francis’ Elevate series, will have to learn from the lessons at Hindawi, MDPI, and Frontiers before pursuing the model at scale.