Universities spend a lot of money on journal subscriptions. A lot. My institution budgets millions of dollars annually for this. Multiply that by all the major universities in the world and it's clear that academic publishing is a multi-billion dollar industry.
In this era of declining university budgets, however, libraries are struggling to keep up with the costs. A few years ago my university's library was forced to cancel hundreds of print journal subscriptions (it kept electronic access, which is a bit cheaper). There was talk of dropping some of the more expensive electronic bundles; that didn't happen, thankfully, as it would have had a devastating effect on the research enterprise.
High journal prices are especially galling to mathematicians and physicists. We use the technical typesetting program LaTeX almost exclusively (a word of advice: never ask a mathematician to deal with Microsoft Word). The journals we publish in (and this is true of physics, too) use LaTeX as well. Publishers even provide template files for authors to use as they write up their work so that it will fit the particular journal's formatting. Yes, you read that correctly: mathematicians and physicists do the bulk of the typesetting of their journal articles. The publishers do employ copy editors, but they really don't have to do much provided the authors use the templates. Add to that the fact that academics referee papers for no compensation (it's considered a service to the profession) and one begins to wonder just why math journals are so expensive.
They're not all that way. Journals published by professional societies such as the American Mathematical Society and the European Mathematical Society tend to be reasonably priced. For example, the Journal of the American Mathematical Society, widely considered one of the top mathematics journals in the world has an electronic subscription price of less than $300 per year. The Annals of Mathematics, perhaps the top journal in math, is published by Princeton University Press and costs about $500 per year. It's not as easy to figure out how much a particular commercial journal costs since publishers offer multiple titles at a bundled price, but they are significantly more expensive. Twenty years ago, Inventiones Mathematicae, a top math journal published by Springer cost more than $2,000 per year (roughly $1.10 per page at the time compared to roughly $0.15 per page in the Annals). Prices such as these drove many leading mathematicians to quit editorial boards in protest.
With the rise of online journal access, publishers faced increasing pressure over subscription prices, especially as demand for print journals has declined (I haven't held a print journal in my hands for years). One response has been to allow authors to choose to let their articles be open access. Such an article is available to anyone free of charge. Should authors wish to do this, however, the journal charges them for the privilege, often $3,000 or more per article. Even if you have a large research grant it's difficult to justify such an expense.
Mathematicians and physicists also use the arXiv, hosted by Cornell University, to post preprints of their work. In fact, this is how we find out about new results since the editorial process for journal publication can take years from paper submission to its appearance online or in print. Science moves more quickly than that, and the arXiv is so useful for the dissemination of information that the role of academic journals these days is to properly vet papers via a rigorous refereeing process (and to provide documented, dated evidence of productivity for tenure and promotion committees). Moreover, the arXiv is a stable, completely open access platform supported by the Simons Foundation and a consortium of universities. Authors upload their articles free of charge to the arXiv servers where they are housed in perpetuity.
Enter Sir Timothy Gowers. A 1998 Fields Medalist, Gowers is a professor of mathematics at the University of Cambridge working in functional analysis and combinatorics. Earlier this year, he led the launch of a new journal, Discrete Analysis, with a new publishing model. The interface is very sleek, with a modern look and feel powered by Scholastica. The journal's website has editorial introductions to the papers "published" by the journal, but the articles themselves are housed on the arXiv and are therefore freely available to anyone. The submission and refereeing process functions much like a traditional journal, but the costs are minimal. For now the article charges are being covered by funds procured by Gowers, but should it become necessary for authors to cover the cost it will be two orders of magnitude less than what commercial publishers charge for open access.
I asked Gowers some questions about this new venture via email. Here is the transcript of our communication.
KK: Tell me a bit about the aims and scope of Discrete Analysis. Does it fill a gap in the literature, or do you anticipate that it will replace one or more existing journals?
TG: It is a bit of both. I work in an area of mathematics known as additive combinatorics, which does not fit very well into the standard categorization of the subject, as it lies at the interface of several different branches, such as harmonic analysis, analytic number theory, dynamical systems, and extremal combinatorics. So it seemed like a good idea to have a journal where it would be natural for additive combinatorialists to submit good papers. However, when we thought about how broad to make the scope of the journal, we decided that it would be best to make it quite a bit broader than just additive combinatorics, so that people could submit papers in the areas that feed into additive combinatorics even if they weren't particularly additive-combinatorial papers. So I hope that we are both filling a gap and competing with certain existing journals.
KK: What was your motivation for launching the journal? Why did you choose this particular publication model?
TG: I have been concerned about journal prices for a long time. However, it is not realistic to ask the commercial publishers to change the system in a way that would drastically reduce their profits, so it is necessary, it seems to me, to create an alternative, much cheaper publication system that can build up a reputation to the point where there is no longer any point in using the much more expensive system. That could take a long time, but it will take even longer if we do not even start.
A secondary motivation was that I wanted to challenge the received wisdom about "gold" open access -- that is, the model where authors (or more accurately their institutions) pay a charge to publish their papers and readers can then freely read them online. I got tired of hearing from publishers that it was unrealistic to expect the charge per paper to be substantially less than £1000, and that academics did not understand all the services that publishers provided, and that they were necessarily expensive. The best way to counter such arguments was to start a new journal that provides all the services that authors and readers actually want and that is run at a tiny fraction of the cost: I'd estimate that it costs us an average of around £30 per published article to run Discrete Analysis, a cost that we do not pass on to authors or readers.
The fact that it is an arXiv overlay journal (that is, papers live on the arXiv rather than being hosted by us) is less fundamental, since it would not cost us more to host the papers ourselves. But it is meant to make the symbolic point that mathematical dissemination tends to take place via the arXiv these days, and publication in a traditional journal, often quite some time later, is almost irrelevant.
You might ask whether publication in Discrete Analysis is therefore irrelevant as well. I think not, because we do something that traditional journals do not do, which is have a beautifully designed website (designed by Scholastica, the company we have used for creating the journal) with "editorial introductions" to each paper. These provide people with useful information about the papers we have published: we don't just give a bare list of links to the arXiv.
KK: Journal prices are a real concern for universities these days. Do you think the model used by Discrete Analysis can spread to other branches of mathematics and to other disciplines? What are the barriers to this?
TG: I very much hope that the model will spread to other branches of mathematics and to other subjects that make heavy use of the arXiv. Unfortunately, there are several barriers, of which three are the following.
1. People imagine that it takes a lot of work to set up a new journal. Actually, it has taken remarkably little work to set up Discrete Analysis, a message I want to spread as much as possible, but I have been quite surprised by this.
2. Remarkably, given how thoroughly discredited a notion it is, people still take impact factors seriously, and it takes time for a new journal to build up an impact factor. While it does not have one, authors may be reluctant to submit to it. More generally, in many subjects the perceived reputation of the journal you publish in is far more important to your career than the actual quality of what you publish. This is a ridiculous state of affairs, of course, but it exists and it gets in the way of switching to a cheaper and more sensible system.
3. Because of bundling, the practice whereby publishers sell huge packages of (electronic access to) all their journals to libraries, even if all mathematicians were to stop using commercial publishers, it would make no difference to what the publishers would be able to charge for their bundles. It takes concerted action on the part of all subjects, and some big subjects, such as biology and medicine, seem much more wedded to the old system and less ripe for change.
KK: Do you serve on the editorial board of a for-profit journal (as opposed to one run by a professional society, say)? If so, do you notice a difference in the editorial processes?
TG: I am on the board of various journals. The only one I can think of that is run by a big commercial publisher is Combinatorica, which is published by Springer on behalf of the Janos Bolyai mathematical society. I have never been asked to do any work for it. I think it is probably more common with commercially published journals to have long lists of editors of whom only a few are active. I am also on the boards of a few journals run by university presses or mathematical societies. Although most journals at least nominally have a process where the final decisions are taken collectively, I find that there is a noticeable difference between journals where I feel I am working on my own and that whatever I recommend will be what happens, and journals where there is a genuine dialogue between editors. The latter are, obviously, much better. One journal for which I am an editor, Forum of Mathematics, a recently founded open access journal published by Cambridge University Press, is particularly good in this respect, and I have used my experience with that journal as a model for how Discrete Analysis should be run.
KK: Terence Tao rather famously submitted his solution to the Erdos Discrepancy Problem to Discrete Analysis. He could have sent it to one of the top journals (perhaps the Annals or Inventiones), but he chose your new journal. Did you ask him to do this, or was it his idea? Do you think it has helped with the journal’s launch?
TG: That was entirely his idea. If I had to guess at his motives, I would say that he too is very committed to making Discrete Analysis work (although it has been presented as my initiative, it is actually a collective effort of the entire editorial board -- that is a point I have tried on a number of occasions and without total success to make clear), and that he knew that his reputation, and that of the paper, would not be in the slightest bit affected by which journal the paper appeared in. Also, the proof of his amazing result came in two parts, and the other part was submitted to Forum of Mathematics, Pi, which is one of the top journals.
All that said, it was a remarkably generous thing for him to do, and it has indeed been extremely helpful to us, enabling us to position ourselves almost immediately as a very good journal and one that people are happy to submit excellent papers to.
I wish Professor Gowers good luck in this venture, and I hope it's a model that can spread. Commercial publishing will always have a place in academic life, but this sort of DIY enterprise could help universities and faculty manage costs while maintaining quality. Discrete Analysis is certainly off to a good start and I expect similar ventures to launch soon.
