This is a short version of a rant that many of my friends have been patient enough to listen to.
Here’s a short version of how scientific papers get published:
- Scientist writes a manuscript. (I’m skipping a lot of steps here, of course. There’s research and whatever that comes first. But this is, in a strict sense, the start of the publication process.)
- Scientist submits the manuscript to a journal.
- An editor at the journal decides if the manuscript “fits” the journal. This has to do with the journal’s content (e.g., a physics paper typically wouldn’t fit well in a molecular biology journal) as well as the journal’s rigor or prestige (e.g., the first publication of the human genome was in the top-prestige-tier journal Nature, while my paper about how to analyze a fairly obscure kind of experiment is in PLOS ONE).
- If the paper “fits”, then the editor asks other scientists to review the paper. This review checks to see if the paper is technically correct, to ask if the paper “fits” the journal (again, mostly in terms of prestige or rigor), and to suggest improvements to the paper. The scientists are not paid for their work.
- Depending on the reviews, the editor will accept the paper as is, ask the authors to make changes based on the reviews, or reject the paper. A paper can go through many rounds of review at multiple journals before it is accepted.
- Once the paper is accepted, the journal publishes it. A fairly lengthy process of typesetting is done so that the paper looks nice on the journal’s website and has a beautiful downloadable pdf version.
Journals come in two basic flavors: subscription and open access. In a traditional subscription journal, readers must pay a subscription fee to access the papers in that journal. In an open access journal, authors pay a fee to the journal upon their paper’s acceptance. The paper is then freely available to anyone.
There’s another wrinkle to this process, called preprints. A scientist can submit their manuscript to a preprint server. The servers accept almost anything that gets submitted, and anyone can view the manuscripts posted there. Preprints can be posted before, during, or just after the peer review process. This means that, in many cases, there is a paper in a subscription journal, and there is also a freely-accessible version, almost identical in content, on a preprint server.
There’s a lot of debate about whether open access is better for science. Europe is moving ahead on Plan S, which would basically require many European scientists to submit their papers only to open access journals. Some people think this is great; some think it’s terrible.
I think this single-axis debate —subscription vs. open access— is short-sighted. I’ll say why.
As outlined above, journals do three things:
- Journals filter papers based on their area of contents and their prestige/rigor. Because I know that many of the most interesting papers in infectious disease medicine will be in The Lancet Infectious Disease, I don’t have to look through every preprint tagged as being about infectious disease medicine and decide if it’s good or important or whatever.
- Journals coordinate peer review. Peer view is not that old (amusingly, Einstein was annoyed when one of his papers was sent through this then-new process) but it’s now considered a critical part of the scientific process. Note that journals do not do peer review. Scientists do that, for nothing. Gratis.
- Journals store manuscripts and make them accessible.
Journals also do other stuff. Many of the top journals have really good science news sections, policy forums, etc. Top journals can also set standards for good research practice by setting up policies that then will only publish papers that meet certain criteria (e.g., that report certain experimental and ethical information). But the most important content is the papers.
The interesting thing to me is that preprints store manuscripts and make them accessible, without doing either of the other functions. I can therefore imagine a world where the other two functions are also conducted by separate organizations:
- Things like preprint servers hold the manuscripts. (If scientists want beautifully typeset manuscripts, they can either do that themselves with LaTeX or pay someone to do it for them. Journals’ copy-editing departments are basically this service.)
- Scientists ask peer review panels to evaluate their work up to some standard. (Something like this exists in IRBs, the groups that verify that scientists’ work is ethical. Scientists submit proposed research protocols to an IRB at their institution, or they pay an independent IRB to review their protocols. The IRB approves, demands changes to, or rejects the protocol.) Right now there are different de facto standards of peer review that are tied up with the journal prestige, but I think we as a scientific community could unpack that.
- Blog-like entities pick the best manuscripts that have come out in some topic area over some period of time. From my angle, the main function of Nature is to aggregate a list of links to the most interesting papers that have come out in the past week. It’s mostly immaterial to me that Nature also happens to store the manuscript and also coordinated the peer review. (I do care that Nature papers are subjected to high levels of scrutiny in peer review, but I don’t think Nature has a particular standard of peer review scrutiny that is distinguishable from the one you would find in other top journals like Science or Cell.)
The main advantage of this separation is that is makes the movement of money and the division of labor. Rather than paying a subscription journal some amount of money that gets split between typesetting a manuscript, hosting a website of papers, coordinating peer review, doing science journalism, and also repaying corporate shareholders, instead you see what money is going to each of those functions as its performed by different parties.
The main reason none of this will happen on anything approaching a short timescale is that academia is a highly conservative institution. It is a highly conservative institution in large part because success in the system depends strongly on your ability to understand and play to specific rules. For example, to have a successful scientific career, it’s helpful to have a lot of money, which means you need to win grants, which is easier if you’re at a top-tier institution, at which you can only get a job if you have prestigious publications. So folks are very nervous about anything that could affect their ability to perform well in the papers game. In other words, a modular system like the one I propose might be better in the short run, but the important actors are mostly afraid to make any change to the existing system, so I’m guessing it won’t happen soon.