The Hindawi Files. Part 1: The Timeline

And the biggest screwup in the entire history of academic publishing

Introduction

For those of you who don’t suffer the indignity of working in science all day, let’s run the ballistics.

Scientists - that is, people who do science for money - work at universities, research hospitals, private research organizations, private companies with a research focus (especially pharma companies), and other similar institutions. A key part of their job is publishing academic research.

Here’s what that looks like: you write a manuscript on something nerdy, and send it to an academic journal. This is an outlet run by a core team of people who handle the process of turning a manuscript into a published paper. First that manuscript is received by an editor, and if they like it, they send it out to external parties - who volunteer their time - for peer review.

If these ‘peers’ like the manuscript, usually they request some edits or modifications, sometimes some extra experiments, hopefully to improve it. Peer review isn’t perfect, but it’s the best imperfect system we’ve got.

Once the peers like it, the editor accepts it. Then the journal copyedits, typesets, and prepares the manuscript for publication, exactly the same way a book publisher or a newspaper would. Then, it’s published, which these days really means ‘released on the internet’.

A published manuscript is treated as the primary evidence that a scientist is doing their job. As a consequence, publishing more manuscripts in better journals - and they are rigidly ranked according to reputation and academic hierarchy - is considered better.

This process started in the 16th century, but organized, external peer review is a substantially more modern invention — 50, 60 years old. And during that time, we’ve increasingly noticed two things

(1) that the fact that the whole system relies on an implicit idea that ‘more publications are better’, and that idea increasingly leads to scientists cutting corners, self-promoting, and sometimes faking results altogether. And…

(2) because the entire system is based on trust — if you send a journal a manuscript, they often review it as if it was real, and never ask for evidence whether or not the research took place as described — the system is fragile.

The problem with the modern version of this system is publishers have the entire business completely sewn up. More than half of all science in the world is published in a journal owned by one of just five companies: Elsevier, John Wiley and Sons, Taylor & Francis, Springer Nature, and SAGE.

But that undersells the degree of control. More importantly, they control almost all the highly ranked English language journals in the world, and these control the reputation game at the center of publishing. If you have a new medical paper, one of the fanciest places you can publish it is the British journal The Lancet. That is owned by the publisher Elsevier. So, while you can get promoted - or maybe just keep your job - by publishing in The Lancet, access to your work is sold by the publisher back to the universities and other research institutions that scientists work at.

How much money does this make? Well, big publishers have operating profit margins are typically from 20 to 40%. That’s about the same level as Apple, which has an OPM of about 30%. They’re very successful as businesses because they control reward and culture.

(Starting a competing publication or service from scratch is astonishingly difficult, because to build a reputation you need good publications, and to get scientists to send you good publications… you already need to have that reputation. And you’re competing against academic publishing companies who control titles that have existed for many, many years - sometimes centuries. Starting a new journal to ‘compete with the old guard’ is like starting your own New York Times. They were there first. It’s not going to work out for you.)

However, over the four parts of this series, you’re also going to see: while these are very profitable businesses, they are also capable of breathtaking incompetence.

We won’t get into the broader case for why they are buffoons. The quick version is: the Incumbent’s Curse. If you control a domain of business completely, it stifles innovation, and makes your company slow and stupid. You have no pressure to change anything. Mistakes are tolerated because they have no consequences. Eventually the whole thing becomes decadent and stupid.

And we’re going to explore the depths of that stupidity.

How Publishing Could Go To Hell

Large publishers are not invulnerable. No business is.

So, I have often wondered how an academic publisher might melt down. Given that they’re highly stable businesses in a ridiculously protected market, and never forget that academic publishing is just one of their business units, it’s unlikely.

But they certainly could mess it up existentially. How?

It’s good to think about tail risks, which a term people sometimes steal from finance - it means ‘the chance of an uncommon sized loss from an uncommon event’. It might also be worth contributing to such a meltdown, if it starts to happen.

I think our hypothetical meltdown starts with some kind of disaster for a quality signal, a catastrophic loss of trust. For an academic publisher, this could be a number of simultaneous things, but let’s say the publisher is revealed to produce a significant amount of faked, unreliable or untrustworthy research documents.

Given that happened, and it would have to happen big, I think 3 additional outcomes come into play, maybe simultaneously, in combination, or in an unpredictable order.

• First, many of the academic journals managed by the same publication group are ‘delisted’ by databasing services. There are other companies that grade and rate academic outlets, they’re like the Moodys or the Standard and Poor’s of the publication system. After a catastrophic loss of trust, these companies would deny the journals owned by those publishers a rating. In nerd terms, that means a listing in formal academic databases: Scopus, WoS etc. While the publisher still owns the intellectual property of the journal, the brand, the copyright etc., publication in one of the titles they own does not ‘count’ any more as academic output.

• Second, these journals are also delisted by universities and governments, who both monitor academic publications because the publishers end up with a lot of their money. Different universities and governments around the world frequently ‘stop recognizing’ journals or publishers for being too low quality, at which point they stop adding to national or local metrics of academic output. In other words, if you publish in Nature, that counts. But, if you publish in Delisted Journal X, it’s no different from blogging or writing your research on a wall in permanent marker.

• Third, journals in the publisher’s portfolio are then removed from government or university open access mandates (i.e. the people who pay the publisher for ‘services’ no longer consider the cost of publication worth paying) and also from library subscriptions (so, universities who buy digital services back from publishers no longer see the cost as justified).

These are some of the assumed markers of quality that underlie the true fidelity of a journal. Whether or not someone will pay for it, count it, or database it is the actual core of academic legitimacy. If this quality slide has begun, and everything I just described starts to happen, what’s likely after that is exactly what you’d expect: no-one wants to work for them. And that includes the editorial staff who manage the papers, and the unpaid researchers who do the actual academic work of reviewing the papers. Also, no-one wants to send them their research.

This decline in participation is self-perpetuating - the more people leave, the more unreliable the journals seem. Revenue spikes downwards at this point, targets are missed, and the whole thing starts to circle the drain.

This is a lot like an ecosystem collapse. In an ecosystem, multiple interdependencies overlap, and their mutual influence maintains a complex system. Pull one of the threads too hard - for example, take all the wolves out of Yellowstone - and it all goes to hell.

Bear in mind, this is a thought experiment. I am not saying it’s likely or even remotely likely. 

But for the first time, I see how it might be possible.

Because there was: an incident.

And that brings us to…

John Wiley & Sons.

John Wiley and Sons (NYSE: WLY) is a publishing house established more than 200 years ago in New Jersey. They’re a massive publicly-traded company with an annual revenue of about 2B. Wiley is the fourth of the Big Five publishers (again, with Elsevier/RELX, Taylor & Francis, Springer Nature, and SAGE).

Wiley is traditionally distinguished from the other four by their stronger commitment to a closed-access publishing model (when you publish for free in a journal that someone has to pay to read). By the 2016 figures, about 4% of their journals were open access (where you publish for money in a journal that is free to read).

4% of all your portfolio being open access is very low. All the other big publishers have more than that.

That was, at least, until Wiley bought Hindawi (an open-access publishing company, with a portfolio of ~200 journals, all gold OA model). Gold OA journals publish free-to-read scientific papers, but authors pay a publication cost. Buying Hindawi outright cost Wiley $298M.

When Wiley made this acquisition, they made a gigantic strategic error. What you’re reading here is the first of a series about that strategic error.

Why they made this error is easy to explain, and I’m going to say it twice.

Everyone ignores both the frequency and severity of research integrity problems.

Again.

*Everyone* ignores both the frequency and severity of research integrity problems.

By ‘everyone’ I mean academics, universities, governments, public and private corporations AND the world at large. Even though there are severe issues at present in the accuracy of published science globally, a majority of people in all of these parties have a kind of laissez-faire attitude to the problem. They figure we’ll more or less muddle through the problem that science is a bit shit now, and the world will continue to turn.

And there’s a small, weird world full of people like me, who really care about research integrity, and we think everyone else is ignoring a giant wound in the side of global progress that is turning septic.

Only very occasionally is there a great big fat screwup in the publication process that makes it out of the research integrity world, and into the Oh Shit Consequences Are Real world.

This is the story of one of those errors.

Part 1: The Timeline

So, Wiley bought Hindawi. Let’s follow the story of this acquisition over the last few years.

The following details are all from quarterly earnings presentations unless otherwise stated, dated from their release, not the quarter they are reporting for.

***************************************

April 14th, 2020. SODP. From the Hindawi Director of Marketing.

“MA: One of the areas of our business that we spent time refining and improving over 2019 was our Special Issues program… The specific focus of Special Issues tends to produce high quality, timely articles and generate a greater level of interest, discussion and awareness around a dedicated topic. 

We worked throughout 2019 to hone the processes by which we identify, invite, approve and promote Special Issues, and the result was a more rigorous procedure and a higher quality output… In the first quarter of this year we have seen a wealth of new Special Issues being launched across many subject areas by researchers from all over the world. The quality of these Special Issues (and the researchers running them) is attested to by the fact that we have seen a 40% increase in submissions to Special Issues so far this year.”

January 5, 2021. Wiley press release. Acquisition of Hindawi by Wiley announced. $298M. “For the fiscal year ending December 31, 2020, Hindawi is projected to generate approximately $40 million in revenue with year over year growth of 50%.”

September 2nd, 2021. “M&A: M&A focused on adding scale or capabilities in key strategic areas of research and career-connected education. Hindawi (open research) integration on track.”

December 7th, 2021. “Article output +8% driven by Hindawi.”

March 9th, 2022. No mention of Hindawi in the investor presentation.

June 15th, 2022. “Drive Hindawi integration and revenue synergies -> Integration complete; 36% growth in Hindawi publishing output; strong double-digit growth in revenue and exceptional profitability”

August, 2022. A PROBLEM ARISES. ChemistryWorld reporting (04/23). “Jay Flynn, executive vice president at Wiley, was first alerted to the problem of paper mills at Hindawi in August [JH: presumably of 2022]. Further investigation uncovered a large-scale problem with special issues where a topic expert invites contributions and manages the peer review process… In addition to introducing additional AI-based screening, Flynn deployed 200 people from his editorial staff to conduct ‘a manual review of every single paper that we thought may have been compromised’.”

September 7th, 2022. No mention of Hindawi in the investor presentation.

September 28th, 2022. Retraction Watch reporting. Hindawi / Wiley plans ~500 retractions.

Mid-October, 2022.  Hindawi special issues pause submissions.

~20th October, 2022. ChemistryWorld reporting (04/23). “Wiley convened a meeting of publishers at the Frankfurt Book Fair in October [JH: FBF was 18th-22nd Oct 2022], inviting Clarivate, and disclosed to them problems they themselves had.”

December 7th, 2022. No mention of Hindawi in the investor presentation.

Mid-January, 2023. Hindawi special issues resume submissions.

March 3rd, 2023. “Projected impact to our Fiscal 2023 outlook is up to $30M in revenue and up to $25M in Adjusted EBITDA”.

March 9th, 2023. Press release. “Research was down … primarily due to a pause in the Hindawi special issues publishing program. The program was suspended temporarily due to the presence in certain special issues of compromised articles. As a result, Hindawi revenue declined $9 million vs. prior year, offsetting growth in other open access publishing programs.”

March 24th, 2023. UKSG.org reporting. 15 Hindawi, 4 Wiley-Hindawi, and 2 Wiley journals delisted by Clarivate.

April 4th, 2023. SKitch article. “In September 2022, Wiley identified and immediately alerted the industry to paper mill activity we found operating at scale. Specifically, we found fraudulent outside editors that had subverted our processes and workflows, leading to a proliferation of bad content. This scheme hit Hindawi’s Special Issues program hard… [We] issued an initial 511 retractions, and introduced additional AI-based screening tools. We are currently in the process of retracting an additional estimated 1,200 compromised papers.”

May 2nd, 2023. Hindawi blog. “Hindawi has today closed four journals, which have been heavily compromised by paper mills”

June 15th, 2023. Let’s reproduce the whole slide here.

September 7th, 2023.

▪ “Hindawi to recover growth trajectory in Fiscal 2025 and exceed Fiscal 2023 revenue in Fiscal 2026

▪ Wiley’s journal brand asset continues to grow; 112 Hindawi journals recently received first Impact Factor and 35 saw higher scores (out of a total of 200 journals); demand improving. Strong scores for Wiley journals in China”

October 10th, 2023. Wiley press release. CEO resigns. Investor day (October 12th) postponed.

October 23rd, 2023. Malaysia includes Hindawi journals in the list of publishers whose OA fees they won’t pay.

December 6th, 2023. RetractionWatch. Hindawi brand is cancelled. “Wiley will cease using the beleaguered Hindawi brand name, the publisher announced on an earnings call Wednesday morning. Wiley plans to integrate Hindawi’s approximately 200 journals into the rest of its portfolio by the middle of next year.“

March, 2024. Hindawi is reduced to a footnote on an earnings slide, just something that messes up the EBITDA margins.

***************************************

As a series of headlines, this feels like a regrettable but unfortunate business incident. One of things that ‘just happens’.

But this incident now represents the largest body of mass retractions in human history, and the single biggest failure of academic oversight in human history.

In the first year (specifically, 16th Nov 2022 to 29th Nov 2023), Hindawi mass retracted 7996 individual papers. At the time, this was more than the retraction count of all the science in all the world in 2022, and the figure was solely responsible for the gigantic retraction count in 2023.

There continued to be additional retractions, specifically:

2022: 8830

2023: 9936

2024: 1391

(Accurate at time of writing. Note: you are required to search year-by-year as the whole total is more than their website search allows.)

These figures are double-counted (as paper titles are changed to begin with ‘retracted’ and so are the retraction notices). The Wall Street Journal counted the revised total recently as 11,300. Some of those may not be from this incident, but it’s a reasonable estimate. Especially because it keeps going up.

So:

Hindawi was bought without the knowledge that they were publishing a lot of clearly fake scientific papers. After the acquisition, the problem got worse. A lot worse. And Wiley ran their new asset for an additional 20 months without noticing that the problem was getting worse.

You get a sense of how asleep people at Hindawi were in the first interview segment above: “We worked throughout 2019 to hone the processes by which we identify, invite, approve and promote Special Issues, and the result was a more rigorous procedure and a higher quality output”.

If that was true, it would just be a smug thing to say. But describing the greatest failure of scientific governance in history, it is… an astonishing admission of incompetence would be the nicest way I could phrase it. That’s really Captain of the Titanic stuff. It’s like announcing how great Enron was doing in July, 2000.

We’re going to follow through a few of the narrative threads that come out of this incident in future episodes, and there are three more parts to come.

Next time…

Part 2: Hindawi.