Reading Time: 3 minutes

Data servers (Unsplash)

In December 2021, eLife received additional funding specifically for preprints.

The new funding will allow eLife to advance its vision for a system of curation around preprints that replaces journal titles as the primary trust indicator of a paper’s perceived quality and impact.

Why not have the preprint as the version of record? Well, apart from a few housekeeping issues, there shouldn’t be any difference. By housekeeping, I mean the small matter of some 50 or 60 separate checks that publishers run on submissions before the proceed to peer review. Many of these checks are trivial, such as making sure that all the references are cited in the article, yet it is checks like these that are the most difficult for a brain to handle. Humans have a very limited attention span. They are good at checking a few things, such as the number of words in the title, but the brain complains when asked, for example, to count the words in an article: it’s easier to get a machine to do it. Why do preprint servers not have an array of automated checks that have to be run before submission? Full disclosure: I work for a company that provides such tools. But in this case, the tool would be run by the author, not by the preprint server (which has no editorial staff). In this way, an effective self-policing operation could be used, by which the author verified his or her own content, and stated publicly (and credibly) that they had checked for, say, conflict of interest.

With a self-policing system, the author would nonetheless have had their journal checked for potential errors. Perhaps, ambitiously, a preprint server would only allow a paper to be uploaded if it had passed all the mandatory checks required. If the author was not happy with the automatic check, they could submit the paper for manual checking – and accept the inevitable delay in publication. Or they could simply accept that a preprint server is run without staff. Either the paper meets the checks, or it cannot be loaded.

In other words, academic publication could move towards flying with Ryanair. You don’t like Ryanair, but you put up with the online options, because you know that any manual operation will cost you £££ more. While complaining, you notice that you have more flexibility, in some respects, using online tools to book and check in, than you did when things were done by phone or in person.

One aspect of using machines to check submissions is that you can check for more things. Excessive self-citation, for example, is something that is simple to check by machine, and the sort of thing that gets forgotten when you are checking by hand.

If preprints become the rule, would such a move weaken the established subscription journals? Probably not. The status of being published in a highly cited journal is the usual justification for continued adherence to journals, but by the time the article has been read, commented and responded to hundreds of times, there appears to be less and less need for a version of record. The analogy with open-access is instructive. In 2020, 578,505 OA articles were published by OASPA members, according to OASPA figures, but it hasn’t weakened the position of the largest publishers; they simply changed how they charged.

In fast-moving subjects such as life science, journals accept that publication of the version of record will never be fast enough to meet the needs of researchers, who start to resemble day traders in their haste. This de facto situation has existed for years, without subscription journals going to the wall.

The most important thing would be to establish the credibility of the preprint system, and a self-service checking system is one initiative that would make sure the author has at least carried out some minimal checks before publication.

So preprints are here to stay. What might tilt the balance away from subscription publishers is if peer review could be moved to the preprint platform while remaining convincing. Among the many new models for peer review, one will become the new default (most likely some kind of open review) and then, perhaps, the big traditional publishers will have something to worry about.