Issue 56 · August 2023

Join the conversation on Twitter @Briefer_Yet and LinkedIn.

Did a colleague forward the The Brief to you? Sign up via the button below to get your very own copy.

Upcoming Events

Members of C&E team and our associates will be speaking at or attending a number of upcoming industry events:ALPSP Annual Conference and Awards 2023 – Nicko Goncharoff will be speaking on the topic of “OA China: Positioning publishers to succeed in the world’s largest market for articles” (Session 10) on September 15th. Nicko recently collaborated with C&E on our recent International STM Publishing in China: State of the Market Report 2023. We encourage you not to miss this session if you will be attending ALPSP this year (and to purchase this essential report). James Butcher will also be attending this event.

Silverchair Platform Strategies – Colleen Scollans will be speaking on the topic of “Digital to Data: Marketing strategies & technologies for the next era of publishing.” Look for her and other members of the C&E team at this conference.

STM Dinner & Conference 2023 – Michael Clarke will be moderating the panel titled “Achieving open, visible, and impactful research at scale. Where have we got to and what’s next?” Michael’s colleagues Colleen Scollans, David Lamb, and Nicko Goncharoff will also be attending the meeting.

Frankfurt Book Fair– Michael Clarke, Colleen Scollans, David Lamb, and Nicko Goncharoff will also be attending the meeting.

Contact us to set up a meeting in conjunction with any of these events.



This month, Frontiers published its 2022 Progress Report, which is packed full of facts and figures about article output, citations, publishing speed, and customer satisfaction. The picture for 2022 was largely rosy, but next year’s annual report may not include as many good news stories – Frontiers’ article output has slowed down in recent months.

The picture for the 2022 calendar year was very encouraging, however, as the 92-page annual report makes abundantly clear. The headline figure was that Frontiers published 125,104 articles in 2022, up from 85,582 the previous year (46% year-on-year growth). Submissions increased by 52% year-on-year to 240,037 in 2022.

Frontiers launched 49 new journals in 2022, bringing the total to 189 by year end (there are now 219 journals listed on the Frontiers website, 30 new journals having already launched in 2023). Between them, the journals were run by more than 243,000 academic editors – an average of ~1,300 editors per journal – 90,000 of whom were recruited in 2022.

Frontiers had 2,201 staff in December 2022 (that figure is now 2,500, according to their website), meaning that, on average, 57 articles were published per staff member in 2022. By contrast, MDPI published 303,000 articles and had 6,750 staff on the books in 2022 according to their annual report, giving an average of 49 articles published per staff member. Frontiers appears to be slightly more efficient than MDPI, although it seems likely, given the location of its offices, that its overhead per staff member is higher.

Frontiers puts the citation performance of its portfolio front and center of its progress report. “Frontiers is the 3rd most-cited among the 20 largest publishers, with an average of 5 citations per article” appears at various points in the report, which includes 19 pages detailing the performance of the journals in Clarivate’s Journal Citation Reports and Elsevier’s CiteScore.

The report omits a discussion of the underlying cause of this excellent citation performance, however. The implicit message is that the research Frontiers publishes is of better quality than other publishers. This may very well be the case but to the extent quality is correlated with an average number of citations per article for a given journal, the performance by Frontiers is assisted by the large number of review articles they publish. Frontiers has published just over 464,000 articles (all article types) since 2007: 20% are classified as “review”, “opinion”, “mini review” or “perspective”; and, in fact, only 64% of Frontiers’ output is classified as “original research” according to the metrics provided on their website. Review articles are cited more frequently than original research papers, which helps to boost the average citations per paper.

It’s a considerable achievement to publish so many review and opinion articles, especially since a significant number are commissioned. Commissioning an academic to write a review is challenging at the best of times but asking them to pay an article processing charge for the privilege of being commissioned isn’t an easy sell. With so many review articles being published across the scholarly ecosystem, the market is likely to become saturated. 

As of August 30, 2022, according to the Frontiers website, 112,758 articles have been published in the last 365 days, which is a decrease of 10.8% when compared to the 365 days of the 2022 calendar year. The graph below shows the number of articles (of all types) that were published per day in each quarter in 2022 and 2023.

In Q1 2023 Frontiers published 339 articles per day, on average. In Q2 that number fell to 242 articles per day and is currently running at around 222 articles per day for Q3. It’s not clear what has caused this drop in output. 92% of Frontiers authors rated their experience as “good” or “excellent,” according to the 2022 Progress Report. Furthermore, the average time from submission to final decision fell from 77 days in 2021 to 73 days in 2022. Acceptance rates appear to be steady at about 60%. Perhaps submission-to-publication times have increased in 2023, created by a backlog of papers undergoing peer review, or perhaps more stringent checks have been put in place off the back of Clarivate delisting more than 50 journals earlier this year. Alternatively, submissions may have leveled off after considerable growth in recent years or in the face of increased competition. 

MDPI’s growth rates have also been hit after two of their journals were delisted from the Journal Citation Reports; 5 out of 6 authors walked away when MDPI told them that International Journal of Environmental Research and Public Health had been delisted back in March. MDPI published 524 ‘articles’ per day in 2021, 672 per day in 2022 and is currently running at 676 per day (source: MDPI website); so, although output has increased in 2023, the rate of growth is much lower. Does this signal just a pause or the end of the unprecedented growth for both Frontiers and MDPI? If the later, other scholarly publishers across the ecosystem will be breathing a huge sigh of relief.

The Low Hanging AI Fruit


An intriguing aspect of the rapid rise (and uptake) of generative artificial intelligence (AI) and large language model (LLM) technologies lies in not knowing the eventual value they’ll bring to scholarly research and what sorts of products and services will be built around them. We’re in the very early days of coming to grips with these technologies, but as technologist, product development guru, and friend of The Brief Steven Sinofsky notes, “there are parallels to learn from and help guide us on how technology will evolve. Not the one path, but the sorts of paths that can follow.” Based on the history of technology development, Sinofsky’s first prediction, made in February 2023, has already come to fruition: 
…in the next 6–12 months every product (site/app) that has a free form text field will have an “AI-enhanced” text field. All text entered (spoken) will be embellished, corrected, refined, or “run through” an LLM. Every text box becomes a prompt box.
Six months later, it seems like every scholarly bibliometric database or management tool announced just that – the development (or in Clarivate’s case, the plans to develop) an AI-enhanced query system. Elsevier has launched an alpha test in ScopusDigital Science has launched a private beta test in Dimensions, and Papers has announced a waitlist for a beta test. Clarivate announced a partnership with AI21 Labs as “part of its generative AI strategy.” Meanwhile, Research Solutions (the company behind Reprints Desk) announced the purchase of ResoluteAI, which (according to the press release) will provide Research Solutions with “integrations of taxonomies, ontologies, and knowledge graph technology, together with the latest [LLMs to]make information discovery and retrieval highly efficient.” Each of these services offers (or will offer) researchers the ability to ask natural language questions and have the AI Assistant curate the top papers in the field into a written synopsis, or provide plain language summaries of specific papers.

As Sinofsky notes, this is the low hanging fruit of AI. His chosen comparison is that of spellcheck, which originally was a standalone product, then became a feature of other products, and now is just something that happens as one types, whether in a browser or a product window. In Joe Esposito’s classifications, these services are not a “business” nor a “product,” and may not, after long, even be considered a “feature.” What will be truly revolutionary are the products built from the ground up using LLMs that will subsume the existing tools.

As generative AI technology is built into bibliometric databases and other discovery services, it raises questions for publishers. Such databases are typically viewed by publishers as helpful discovery tools driving traffic to journal articles (along with providing useful bibliometric data, market intelligence, and valuable information concerning authors). AI interfaces have the potential to upset that view. First, they may lead to lessdiscovery of, and traffic to, individual papers. What percentage of readers really wants to dig into the details of an article compared with how many will be satisfied with a quick AI-written synopsis of its contents? Second, indexing publisher content to build a bibliometric or discovery database is one thing. Ingesting publisher content to train an LLM is another thing entirely. Publishers are increasingly viewing such use cases as a different category of licensing than traditional indexing services – and one that may ultimately have significantly more economic activity associated with it. There is also the not insignificant matter of open access publishing and the fact that publishers are increasingly issuing CC BY licenses, which explicitly allow commercial uses such as ingestion into LLMs.  

There is also a smaller market of publishers, librarians, research societies, and researchers using Scopus, Dimensions, and Web of Science for bibliometric study of the literature itself, where integration of LLM-driven search will have a significant impact. Most of these tools require a non-trivial level of expertise to interrogate and interpret, which often involves downloading reams of data for analyses in other software – even to answer relatively simple questions. Although the current AI implementations are geared toward research discovery, one hopes they will at some point extend to bibliometric analyses. Being able to ask a plain language question and have it answered in a similarly straightforward manner has the potential to dramatically expand not only the questions asked but also the number of people that seek to ask questions. 



After the deal to sell publishing house Simon & Schuster (S&S) to Penguin Random House was blocked by the US Department of Justice, the company has found a new suitor, and S&S will be sold to private equity firm KKR & Co. The sale has raised concerns among authors and book lovers as KKR has a history of “vulture capitalism” – it was one of the entities behind the purchase and subsequent collapse of Toys ‘R’ Us. KKR is also the owner of the OverDrive ebook and audiobook platform for libraries.
CACTUS Communications has merged with Kolabtree, a freelancing platform for scientists.



Sarah Tegen has moved to a new position as the Senior Vice President and Chief Publishing Officer at the American Chemical Society.
Randy Townsend is now a Consultant with Origin Editorial.
We sadly note that this summer has seen the untimely deaths of several well-loved and respected members of our community. Our thoughts and condolences go out to friends, family, and colleagues of Justine Covault (Massachusetts Medical Society/New England Journal of Medicine and also an accomplished professional musician), Adrienne Vaughan (Bloomsbury), and Mohammed Yahia (Springer Nature/American Chemical Society).

Briefly Noted


While Congress and the White House continue to wrangle over funding the requirements set out in the Nelson Memo, the US Department of Agriculture (USDA) is the latest agency to release their public access implementation plans. Like most plans already released, the USDA’s is largely an iteration of its existing Holdren Memo plan, but with the embargo reduced from 12 months to zero. Interesting wrinkles include mandating authors to use individual persistent identifiers (ORCID iDs are suggested) when depositing research papers in the agency’s publications repository, PubAg, and that the agency’s data repository, Ag Data Commons, is migrating to the commercially owned Figshare platform.

Some welcome news from cOAlition S this month was the launch of their Hybrid Open Access DashboardThe data offered here are interesting and should prove valuable in tracking a rapidly changing open access landscape.

In yet more examples of how AI is “ruining” everything, authors must now be increasingly aware of illegal AI-generated books being sold under their names. Author Jane Friedman was alerted to a series of books being listed on GoodReads and sold on Amazon where she was credited as the author. The AI-generated books were eventually taken down from both sites, but not without a struggle (and the ability of a well-known author to generate negative publicity). Meanwhile, it’s not just actors who are concerned with being replaced by AI dopplegangers. “Digital narration” seeks to replace talented audiobook narrators with AI.

Oxford University Press released its 2022–2023 Annual Report, noting an increase of 5.6% in annual turnover with an 11% increase in net profit.

It’s not what you know, it’s who you know. A study in Nature Communications shows the career benefits of coauthoring papers with well-known and well-cited researchers

Can book reviews survive in an open access world? The authors of this piece (from De Gruyter) suggest that Subscribe to Open (S2O) is a possible savior. 

Speaking of S2O, the American Society for Microbiology announced that their seven subscription journals will move to S2O starting in 2025.

In last month’s issue of The Brief, we noted that, in instances where journals are tied in the Journal Citation Reports category rankings, the tied titles were listed alphabetically. Clarivate has made a change to its listings, and now lists tied journals by their Journal Citation Indicator (JCI) metric as a secondary sorting characteristic.

The mystery of the “Vickers Curse” has finally been solved. An editorial about moth pheromones in Current Biology has been cited over 1,200 times, often in wildly inappropriate papers. The reason for this stems from a glitch in Google Scholar and the way it interpreted wrongly formed DOIs. The default prefix of a DOI for an Elsevier journal is 10.1016, which is usually followed by a slash and a DOI suffix, which indicates the specific journal and the specific article. But if there’s an accidental space in there, and just the prefix is recognized, for some reason Google Scholar would default to the Vickers editorial. Google Scholar has now fixed the bug (no pun intended).

Everyone’s job looks easy from the outside. But until you actually try to automate a job out of existence or replace the workflow or tooling of a job, you don’t really know what is the human part or where the “work” happens. – Steven Sinofsky, AI, ChatGPT, and Bing…Oh My