Life After Paper Mills

You won't like this.

One of the nice things about living startup life for a few years is that it becomes natural to think with your ‘business brain’.

A kind of rapacious and extractive circuit starts up in your neuralses. The instant reaction you have to a new situation changes. Rather than trying to understand the problem more deeply (regular thoughts for me) you start to think ‘how would I make money out of this?’

This is often analogous to ‘how would I exploit this?’

Sometimes these thoughts are unpleasant, and happily I don’t have to pay any attention to them, and I’m certainly not compelled to follow them up.

As such, they’re very useful thoughts. Predicting future bastardry is a wonderful comfort to have. It prepares you for both specific and general situations.

Today, one of those thoughts has led me here.

Here’s how I think some bright spark, at some publisher, will start their own paper mill. Doing this will massively disrupt the paper mill market, which is quickly growing into a bigger and bigger business.

Did He Just Say That?

Yep, I have some explaining to do.

I just read this great, great preprint: https://arxiv.org/pdf/2503.21267

This is an absolutely heroic treatment of a very sickly Taylor and Francis journal.
And "the case of the journal Bioengineered" — in this genre, any time someone uses 'the case' in a title, you're going to have a bad day.

And what a bad day it is.

In 2021, Bioengineered got absolutely frantic with publishing new papers, at a rate about ~6x their regular.

God knows what that meant for the submission rate, it was probably higher still as the flood of papers going out implies a proportional flood going in.

They didn't notice anything untoward with that, because the same thing happened again the next year.

Why? This is what paper mill targeting looks like.

A paper mill is a bit like a parasitic infection — they're looking for a soft target, and once one is found, they burrow right inside. Why go through the hide when you can go straight through the eye?

So, they hit this hapless journal hard for a few years until the scheme was clearly too audacious to hide.

Anyway, the story breaks, and Bioengineered farted around haplessly for a while but didn't clean up the mess. Most of it they just... left sitting there.

So, these authors did it for them. (COI: I know all five of them, and they’re selfless heroes with great skill, dashing good looks, and they all carry a rapier at all times like a kind of science Zorro).

Washing out the journal contents to just preclinical studies, 226 out of 878 had visible image problems (~25%). Only 35 identifiable retractions were present out of that lot, so likely *only 15%* of the necessary retractions.

Ouch!

The obvious question is: how does this compare to the Wiley / Hindawi debacle? Because this is a colossal publishing fuckup, and we should probably try to understand its fuckupitude on the same kind of scale.

Answer: Hindawi was more papers total (maybe 10x), but it was better dealt with (although was still itself incompletely handled). The Hindawi cleanup was faster, and more transparent. Unresolved: how well either publishing group learned from their mistakes. There is no equivalent data on new submissions from either Hindawi or Taylor & Francis from the period after they realised just how badly they got owned.

Now The Bad Thoughts Start

I increasingly suspect that the entire publishing ecosystem — authors, journals, publishers, and universities — *cannot* deal with this problem in the way that these authors have.

Not will not deal with it, cannot. I think we've deluded ourselves that publishers are capable of handling this issue but lack some kind of … I don’t know, ‘gumption’ or something. That they are capable but lack the will.

Rather, maybe, it's a skill and a money issue. If publishers cared enough, they'd be sinking several millions of dollars into this problem, right now. Instead there’s some fairly ropey efforts to prevent things going bad in future. Nothing is invested in mitigation, and because of that it means we have no granular data on what went wrong previously.

In business, preventing a problem from happening again at a functional organization means you need a frank and detailed post mortem. Good organizations run these really, really well. They excel in combining the maximum self-criticism with the maximum freedom from judgment. Different formal evaluative mechanisms in medicine can also be amazing at this — maximum insight, minimum judgment.

But this clearly hasn’t happened here, because if they had the data, they’d issue the damn retractions. And it isn’t a time issue. They’ve had years.

Now, the publication ecosystem is trying, a bit, to deal with the absolute bin fire of global publishing at present. I think there are individual people in that space trying very hard!

But collectively: the urgency is off, the budget is off, and — most tellingly — the people who are really good at this are writing preprints about it instead of working the problem from the inside.

I see a failure in strategy. The whole 'parallel economy' of people trying to deal with forensic metascientific issues behind a corporate veil:

(1) generally aren't good enough at it, and

(2) usually don’t have the resources they need to become good at it, and

(3) DEFINNNNNITELY aren't fast enough at it.

Paper mills (and individual authors) are nimble and clever. Publishers aren't. They’re big stupid lumbering things with oligopolistic tendencies and quarterly statements and SEC filings and meetings about meetings.

It's the same startup / establishment vibe we see elsewhere in commercial life. David and Goliath was never a fair fight — anyone who’s used a sling can tell you Goliath never had a prayer.

Goliath’s only chance was for the Philistines to simply go round to David’s house mob handed and burn it to the ground.

And now we’re at exactly that happening, the March of the Philistines.

You know what happens next? You're not going to like this. I promise you. This is the beginning of the end. Ready?

Prediction: a publisher is going to start some kind of 'paper mill-like' business

Yes, seriously.

What are the economics of publishing paper mill papers? In OA, the transaction occurs is when a paper is accepted. However, under a paper mill commercial model, money has already changed hands upstream, as authors pay a ‘managing author’ or organizer at a mill for access to authorship on a paper.

So if a publisher wants to ‘vertically integrate’, they can bring it in-house.

Under the banner of 'helping authors' and 'democratizing access', of course.

Not ‘selling access to papers’. More… ‘creating opportunities’.

Here’s how I’d do it if I was a publisher:

• you start a training program, give it a cool name ‘Publisher Bob’s Fellowship Program’ or ‘Special Clever Boy Scheme’

• you gate access to people who are already minimally accredited (bachelor’s? interest in the field? MD? could depend on the outlet)

• they you make these people pay, and it doesn’t have to be cheap

• then you administer a kind of ‘training program’ where you drag these hapless individuals through some kind of process to teach them to do research; two viable options there:

  • (1) teach them ratchet quality meta-analysis or systematic reviews (there are already too many of those, so what’s a few more)

  • (2) take input papers from the global scientific community where the authors of those papers 'need help with finishing the manuscript', where the author providing the data or half-finished manuscript gets to publish without fees or APCs on an accelerated timeline, and the journal takes money from 'contributing authors' for training/coordinating.

• maybe invent a new journal that ONLY produces these papers, or get a deal with one of the existing ones that makes these papers ‘pre-vetted’

• manage the editorial structure so these paper get a pretty soft touch on their way to publication; lots of ways to do that, compliant editors, gating the nature of the peer review process, etc.

• count the money

Why would I do it this way?

Because this business model already exists, already works, and has shown itself to be at least minimally defensible (specifically, to accusations of it being a pay-to-play horseshit scheme designed to generate revenue for whoever owns it and gum up global publishing even more with rafts of nonsense).

I don’t think this scheme has an accepted name. I call it the low calorie paper mill or Mill-er Lite if I’m being cute.

(It would be possible for it to work if the training and the output were really good, but they aren’t. They’re shit. But: they do publish papers, and you can’t accuse them of fabricating crap by definition. They ‘train’ people who pay money to write real-but-bad papers.)

What I’ve just outlined would result in a massive transfer of wealth from paper mills to publishers. Of course, publishers already make money from mills, because we know paper mill papers generally pay APCs (if they didn’t, they would be revenue negative and publishers would do something about them).

But this way they would get the paper mill’s business AND roll their APC into the upfront charge to the authors.

If there was a Formal Publishing Scheme for this, why would you take a chance on a dodgy Telegram channel trying to sell you an ersatz publication when for more or less the same price, you could get essentially the same product from a legit marketplace?

The whole thing would quickly devolve into totally dishonest nonsense almost instantly when scaled up, of course. But: there would be no place for paper mills, as dishonest authors could get their shitty manuscripts or datasets finished BY SOMEONE ELSE, FOR FREE and researchers who want to buy papers could simply do it through a scheme with better face validity and no likelihood of getting punked by someone writing an impertinent preprint.

Let’s look at some rough costs.

Paper mill model: fakes paper (wage cost maybe $100), sells access to five authorship slots (maybe $3000 revenue total), manages publication and editing and whatnot (maybe another $100), pays APC (very variable between target journal, and also there are presumably a lot of country-based waivers handed out!)

Mill-er Lite Model: authors all pay for training and paper access (probably $1.5K-$3K per training/article combo per author, if my memory of this business model is serviceable), five authors per paper. Paper mill goes hungry and has to go back to scamming pensioners.

The numbers are all wrong here, but the scale is right. And if all you care about is maximizing profits, doesn’t that beat the hell out of having to investigate and remove paper mill output?? Where you have to spend Real Money to find and remove the problem, which is also decreasing your revenue??

What to do?

There is a one hundred percent chance that the implementation of any broad scheme like this will devolve into a mockery of the scientific process when it is expanded and formalized beyond a very narrow, heavily scrutinized, well-overseen pilot project.

There is already almost zero research integrity oversight, and this would require a lot more of it to function effectively! But as we’ve established, there is no money in oversight. Therefore there is no ‘fiscally responsible’ way to supervise it. A scheme like this starts pre-deranged and quickly graduates to utterly deranged.

So, what to do?

Two things, both simple.

But first: support me, I’m a goddamn delight.

First thing: remember this post. If it doesn’t happen in the next two, maybe three years, write yourself a calendar reminder to call me Chicken Little Dr. Metascience Pissbaby and we’ll have a laugh about my aggressive flight of fancy and how wrong I was.

Second thing: if a publisher tries this, you pick up a stick and you HIT THEM.

Refuse to review for the first publisher who tries this ever again.

Don’t submit to their journals.

Flip them if you can.

Punk them by signing up for their silly program and writing about how bad it is for the newspapers.

And try to make everyone else do the above as well.

And if you see a millimeter of this nefarious plan being floated in the public domain, don’t hesitate. Saddle up.

Let’s HOPE I’m wrong. That would be not very dangerous and pretty funny, and this can go down as an unpleasant thought experiment of Business Brain.

But we’ll see.

Stay vigilant.