Bespoke Preprint Servers Fade

Like shiftless gamers overstaying their extended childhoods, bespoke, specialized preprint servers are being told by their parents to get out and make their own way in the world.

• It’s unclear there is any work for them any more, given the proliferation of second-rate OA outlets and “gaslight” journals. Why preprint when you can just print?

Last week, a job posting for a new CEO for arXiv made the rounds, with a proffered salary of approximately $300,000 for a $6M entity processing 200 papers each weekday.

It was also an announcement that arXiv would be establishing itself as an independent non-profit, for the first time trying to make a go of it outside of the protection of Cornell or the US government. The migration is scheduled to occur in July 2026.

• Paul Ginsparg may also be stepping aside, as one of the requirements for the CEO is, “Academic eminence, ideally in one of the major fields represented by arXiv, including Mathematics, Physics, and Computer Science; a demonstrated commitment to and experience advocating for open science is a plus.”
  • Speaking of recent birds leaving the nest, SPARC may be finding it harder to be independent, as rumors are that their expenses are well ahead of revenues for Year 1, creating a multi-million dollar headache . . .

Forced to fend for itself, arXiv’s reliance on funders is likely to continue, even as its ability to raise funds outside of the house isn’t certain.

In 2023, arXiv received $10M from the Simons Foundation and the NSF. In 2025, it received $7M from NASA and Schmidt Sciences. Cornell knows how to raise money, but will arXiv’s new CEO? Where its next round of funding will come from — enough to make it as an independent non-profit — remains to be seen, but all the current funds less some research funds are going along with arXiv.

• Given an annual run-rate of ~$5M, the $17M since 2023 would be just about to run dry (assuming $5M in 2023, 2024, and 2025 spent).

Expenses have long been a concern for the service, and new policies requiring greater screening will only add to them. I’ve covered this for years (here and here). Cornell has been covering expenses for arXiv and may simply have tired of it. The lack of an obvious upside in the near-term, increased expenses, and a complex and unpredictable funding model — consisting of grants, membership dues from libraries, labs, and academic institutions, contributions, sponsorship, and gifts — may have also made arXiv look like too much trouble. And anything that’s a boondoggle when university and science funding is being slashed by the goons at the NIH is by definition on thin ice.

The change also involves a new board of directors being selected.

Cornell is using a top-tier CEO search firm, SpencerStuart, to conduct the search. They may also be involved in filling out the board, as it is a specialization they also list on their site.

The divestiture of arXiv from Cornell seems like part of a larger pullback from techno-utopian investments in magic information servers at academic institutions, with Cold Spring Harbor Labs letting bioRxiv and medRxiv go to CZI LLC last year via openRxiv. Now, the progenitor of preprint servers itself is leaving government and academic shelter to try to make it on its own.

Speaking of bioRxiv and medRxiv, there’s a whiff of desperation coming off them, maybe because they are becoming superfluous. Opinions of the initiatives have been negative for years, but now there’s apparently no need given the amount of nonsense that can be put into PMC and LLMs via pay-to-play and “gaslight” journals.

• Note: As anticipated, one of Harlan Krumholz’s bogus preprints on medRxiv published with React19 MAHA people seeking to recreate Andrew Wakefield’s initial “vaccine injury” litigation and product scam from 1998 has been cited in a recently unearthed confidential report from the less-than-legit modern ACIP.
  • Krumholz is truly a two-faced cyberlibertarian opportunist who should not be at the head of a prestigious journal, in my opinion. You can’t trust him.
    • I told you so.

But back to the whiff of desperation. Last month, for no apparent reason, bioRxiv posted a self-promotional preprint (a PRprint?) and began touting it on social media. Of course, Nature’s reporters swallowed it hook, line, and sinker, failing to do 10 minutes of analysis to see if the shtick actually merited coverage or if bioRxiv‘s narrative was just plain hooey.

Let’s see what 10 minutes of data analysis shows us about how these servers are actually doing, and why that whiff of desperation may indicate deeper problems.