There’s been a lot written recently about academic publishing, in the kerfuffle over the “Research Works Act”– John’s roundup should keep you in reading material for a good while. This has led some people to decide to boycott Elsevier, including Aram Harrow of the Quantum Vatican. I’m generally in favor of this, but Aram says one thing that bugs me a bit:
Just like the walled gardens of Compuserve and AOL would never grow into the Internet, no commercial publisher will ever be able to match the scope and ease of access of arxiv.org. Nor can they match the price. In 2010, there were about 70,000 new papers added to arxiv.org and there were 30 million articles downloaded, while their annual budget was $420,000. This comes to $6 per article uploaded (or 1.4 cents per download). Publishers talk about how much their business costs and how even “open access” isn’t free, but thanks to arxiv.org, we know how low the costs can go.
This is very nice, but has one major problem: The arxiv is not a journal.
The arxiv does do one of the things that we associate with academic journals, which is to distribute papers to a broad audience. But that’s only one thing out of the list of jobs performed by a typical journal. The arxiv does no peer review– as I understand it, new submissions get a quick glance to make sure they’re not barking mad, but that’s it. There’s no copyediting or layout work done on the submissions, as anybody who’s tried to slog their way through some of the confusing crap that’s up there ought to know. And those services cost money.
“But referees work for free!” hard-core partisans break in. Which is true– academic journal referees are not paid for refereeing papers. So, technically, it’s true that referees work for free, to the extent that they work at all.
I mean, what’s the usual refereeing process? You get sent a paper, glance at it quickly, and set it aside. Two or three (or eight or ten) weeks later, you get an email from the journal editor reminding you that you were sent a paper to referee a while back, and then you dust it off and read it over and either write a report or write a letter saying that you can’t review it.
I’m exaggerating a tiny bit, but not by much. I’ve been very lucky with the peer review process in my career, but I have never gotten a paper through without at least one email sent to the journal asking what the hell was going on with my paper. The APS journals provide (or did, back in the day) a helpful service allowing you to track your paper through the process, which revealed that mine were not the only emails being sent to nag people.
That stuff doesn’t happen for free. Somebody has to coordinate the sending of papers to referees, so they’re not all going to one person, or randomly going to blood enemies of the first author. Somebody has to keep track of how long each paper has been out with a referee, and send the appropriate nagging emails. Those people won’t work for free– somebody is going to want to draw a salary for doing those jobs.
So, citing the low cost of the arxiv is basically the academic equivalent of citing the low cost of paper and ink for trade publishers. It’s true– I’d be shocked if Elsevier was spending more than $6/paper on web hosting– but not really that significant a part of the overall cost.
(To be fair to Aram, he seems aware of this, and has some suggestions toward the end of that post about what it would mean to make a journal out of the arxiv. I’ve seen the same figures thrown around elsewhere, though, without any acknowledgement of the problem with the comparison, and it bugs me.)
Now, does this mean that Elsevier isn’t gouging people? Almost certainly not. All it means is that the arxiv isn’t really a valid comparison. If you want to get all outraged about their profit margins, there are legitimate comparisons– any of the PLoS journals, for example. And, in fact, if you look up the publication fees for PLoS, you see that they’re not all that far off the $3,000 that Elsevier charges for their open-access options. So when you compare two actual journals to each other, rather than comparing a journal to a hosting service, the difference isn’t quite as dramatic.
(You could, of course, argue that even PLoS is grossly overcharging people for their services, for whatever reason. But while the real cost may not be measured in thousands of dollars, it’s not going to be $6/paper. If you want to try to do a Fermi-ish estimate of the real cost per paper, I’d guess that coordinating peer review of all those articles probably requires a minimum of an hour per paper of the time of somebody who will expect to be paid for their time. Which probably puts you up in the neighborhood of $50/paper, maybe even $100 as a more realistic lower bound to the total cost of running a real journal.)