As I wrote about previously (Preprints and the pandemic), preprints have gained in popularity during COVID-19 as researchers either attempt to preempt or bypass traditional journal publishing venues. Notably, however, preprints are not vetted through peer review, and as I wrote in that previous post, this brings with it a whole range of issues. Medical research is extraordinarily sensitive in the way it can affect, for better or for worse, human health and lives. In response, some researchers have begun checking COVID-19 related preprints: for example, the Sinai Immunology Review Project (SIRP) regularly writes reviews for preprints posted to bioRxiv and medRxiv about COVID-19. You can read about their efforts here.
Such initiatives are extremely interesting to me and reminded me of an emerging model of publication in mathematics: the arXiv overlay journal. This mode of journal publishing only arose in the past few years as discussion in the mathematical community about what role a journal actually fulfills gained a lot of attention. Mathematics was, I think, uniquely positioned to have these discussions, for a few major reasons.
There are perhaps three major roles a journal fulfills: peer review, copyediting, and dissemination. In mathematics, posting preprints on arXiv has long been standard, and it is common for researchers and research trainees to keep up-to-date with arXiv categories that are relevant to their interests. Much like how researchers regularly browse their favorite journals, mathematicians (and others) stay up to date on arXiv either by going on the website or subscribing to arXiv emails. So, the dissemination function was already being fulfilled outside of journals.
Secondly, copyediting was already being handled by using the typesetting language, LaTeX. LaTeX is the standard for writing papers in fields like math, physics, and economics; indeed, as the mathematician Scott Aaronson has written, a common rule-of-thumb to tell a “crackpot” math paper is to see if it’s not written in LaTeX. It is an extraordinarily powerful language that provides a consistent and professional look (either by using the default theme or the American Mathematical Society’s) to papers. As well, it can handle drawing graphs and charts, drawing graphics, importing graphics from R and other programs, etc. With authors already handling all that, the copyediting function was already being fulfilled outside of needing journals to do it.
Recognizing the fact that the dissemination and copyediting roles were already taken care of by arXiv and LaTeX, the arXiv overlay concept sought only to fulfill the peer review function. The journal Discrete Analysis, founded in 2016 by the esteemed mathematician Timothy Gowers (who is perhaps best known in general academia for his crusades against Elsevier and other huge publishing corporations at www.thecostofknowledge.com), is one of the earliest such examples. Essentially, the articles simply live on arXiv. Submissions only consist of inputting the arXiv URL to a submission portal. After that, the typical journal process ensues: editors find referees and the usual process of revisions, rejections, or acceptances occurs. Once accepted, the journal simply posts a link on their website forwarding the reader to arXiv, and the article is given a DOI just like in any journal. The arXiv entry is updated to indicate it has been published in Discrete Analysis (or, whichever journal). This is why such journals are called arXiv overlay journals: they simply overlay that peer review “layer” on to the arXiv system. In effect, the journal simply provides the seal of approval of peer review.
An aside: given that Gowers was the founder of Discrete Analysis, it is not so difficult to believe that part of his motivation in founding the journal was as a response to the enormously high profit margins that journal publishers like Elsevier make, whom he has ardently criticized. Given that peer review is done for free (ie. reviewers volunteer their time), really the only service that journals were providing through their own investment was that of dissemination and copyediting. But, if that function is already being fulfilled, then why not just have a truly free journal (free to submit and free to read) using reviewers as volunteers, as they already are. It’s an enormously interesting idea of Gowers that ties into numerous debates about academic publishing, and, in my opinion, demonstrates why Gowers is one of the most talked about mathematicians today.
All that said, could the preprint overlay concept be introduced into medicine? bioRxiv and medRxiv are increasingly taking on that dissemination role: their preprints are often tweeted about and discussed online. The one stumbling block is the lack of standardization of how papers look on those servers, but perhaps this could be remedied by introducing some common requirements for certain aspects like how graphs should look. A more long term solution would be for LaTeX to be adopted in medicine and the life sciences as well – as a LaTeX fan, I would love to see such a day.
Considering the growing use of preprints in the biological and medical sciences, there is perhaps the potential to extend or build on initiatives like the SIRP, which I mentioned at the beginning of the article, to implement a preprint overlay model for other scientific fields. It is not much of a leap to go from a program like SIRP, where people are writing reviews of preprints, to a preprint overlay concept. If the review of the paper is good, just post a link of it on a website. When you strip it down to its essence, as Gowers demonstrated with his journal Discrete Analysis, isn’t that all a journal is – a compilation of papers deemed “worthy” by an editorial committee?
I would be interested to see such a publishing model in other scientific fields as well, including medicine, thus helping to maintain the standard of peer review while recognizing the proliferation of preprints.
Maaz's Notes 