mixedmath

Some may have heard me talk about this before, but I've caught the open source bug. At least, I've caught the collaboration and free-dissemination bug. And I don't just mean software - there's much more to open source than software (even though the term open source originated in reference to free access to source code). I use open source to refer to the idea that when someone consumes a product, they should have access to the design and details of implementation, and should be able to freely distribute the product whenever this is possible. In some ways, I'm still learning. For example, though I use linux, I do not know enough about coding to contribute actual code to the linux/unix community. But I know just enough python to contribute to Sage, and do. And I'm getting better.

I also believe in open access, which feels like a natural extension. By open access, I mean free access to peer-reviewed scholarly journals and other materials. It stuns me that the public does not generally have access to publicly-funded research. How is this acceptable? Another thing that really gets to me is how selling overpriced and overlarge calculus textbooks can allow the author to do things like spend 30+ million dollars on his home? This should not happen. At least, it shouldn't happen now, in the internet age. All the material is freely available in at least as good of a presentation, so the cost of the textbook is a compilation cost (not worth over $100). But these books are printed oversize, 1000+ pages, in full color and on 60-pound paper. That's a recipe for high cost! It's tremendously unfortunate, as it's not as though the students even have a choice over what book they buy. But this is not the argument I want to make today, and I digress.

Recently, I was dragged down a rabbit hole. And what I saw when I emerged on the other side made me learn about a side of math journals I'd never seen before, and the fight over open access.

A while ago, mathematician and Fields Medalist Timothy Gowers took up a standard against the Dutch Publisher Elsevier. In the world of math research journals (and other research journals), Elsevier has become a symbol of The Man. Elsevier has many world class journals, many poor journals, and many in between. The problem with Elsevier is that when an institution such as Brown University wants the world class journals, Elsevier only sells their journals in bundles - big bundles. So Brown has to pay for many many times the number of journals that it wants to pay for - and this is just sort of how Elsevier does its business.

It is behavior like this that angered people such as the late, great Aaron Swartz, who has a very interesting story and past (which is too much to fit in this post). But in short, Aaron Swartz championed open access, and fought vehemently against those who he though cheated the system such as Reed Elsevier. He thought this so strongly that he also championed what he called 'Guerilla Open Access' (as described in his Guerilla Open Access Manifesto), wherein people with access to research hidden to others behind paywalls should grab and rapidly distribute the closed off research. While I do not propose guerilla open access, I do think it's stunning that most research is publicly funded, yet very little research is public. Further, Elsevier has been generating over a billion dollars of profit per year.

I'd like to emphasize one aspect. The authors of papers do not receive any money from the journals for submitted and accepted articles. Nor do the expert reviewers who analyze the articles for scientific and/or mathematical integrity. So all the costs that go towards journals go entirely towards the journal and its overhead (and to its profits).

Some people kept their suffering in quiet, or developed an academic black market distribution networks. But other than a few like Aaron Swartz, protesters tended to protest privately. Tim Gowers was one of the people who decided that this was insufficient. So he helped start up TheCostOfKnowledge, a site where academics publicly sign a statement of boycott of Elsevier. (At the time of writing, 13164 people have signed the document, many of whom are quite high profile). For those who are interested, Gowers is also involved in this list on the polymath site, all about publishing reform. There's a lot on that page. I'd like to point out an interview with Gowers and Neylon, the two most responsible for TheCostOfKnowledge.

What happened afterwards? Well, Elsevier promised to drop prices on their math articles to below $11 per article, opened up a few journal's histories to the public, and withdrew their support for the Research Works Act, which essentially would have prevented any statement that publicly funded research had to be publicly viewable. Since then, the boycott has gained roughly 6000 academics, so it would seem that the boycott isn't over.

So what should be the future of math research and math journals? In 2003 a journal collection known as the Public Library of Science (PLoS) was born, and began publishing a Biology journal. It has since grown to a collection of 7 journals, all peer-reviewed, and all completely open access. Anyone can read them if they want at their website. But they do have operating costs - where does the money come from? PLoS uses an author-pays model, where authors whose papers are accepted must actually pay for their submission. Right now, authors writing for PLoS ONE must pay $1350 per article. This might seem a little nonsensical, but this type of model is often rationalized by saying that the money must come from somewhere. Authors ultimately benefit by having more articles published - they look like better candidates for better university jobs. Further, having an article published in an open access journal means that it is easier for people all over the world to cite your paper, which some consider to be a far better measurement of the importance of an article. Research that inspires more research is good. Is that so bad? Well, it's not so bad as long as authors don't mind paying - but that's a lot of money. One can argue that this disadvantages poor or independent authors (PLoS has been known to meet in the middle in some cases, but other journals on the same model do not).

For some, it is so bad. You might notice that none of the 7 journals of the PLoS are dedicated to mathematics. And I suspect that none of them will ever be dedicated to math. A big problem with the PLoS model is that journals are given a direct incentive to accept more articles - more articles means more money. This means that articles of lower quality might muddy the waters, lowering the prestige of the journal. PLoS does not have this problem, and is currently financially independent and okay (they were supported by grants at first). A second problem is that this model is prone to spam: suppose that you get an email from The Public Journal of Math and Number Theory (I made up this name), inviting you to write an article for their 2013 spring issue celebrating the life of Gauss. You write the article, submit it, and they inform you that they're an author-pays journal. You pay. Maybe they publish, but they publish a terrible compilation of nonsense articles that weren't actually peer-reviewed. Or perhaps they don't even publish - they just take the money and run. Even worse, this might seem like a really nice opportunity to young academics or grad students desperately trying to prove themselves or improve their CVs. There are apparently many scam journals like this. So every such journal must built a reputation before anyone reputable will submit to it.

This is a proper chicken-or-the-egg scenario, and it makes it challenging to start such a journal. And even then, you'd need a willing group of academics to submit to the journal despite the cost of submission, when they can already submit to journals such as those owned and operated by Elsevier at no cost to the author. Even worse is that a few academics starting to submit to author-pays journals do not carry enough weight to cause their respective universities to stop ordering the Elsevier Bundles. Altogether, this requires a strong moral backbone and morals leaning towards open-access. And thus far, these haven't really taken over in mathematics.

Here's a short anecdote of why: There's a very humorous site called mathgen, where you insert 'author names' and it spits out a fake math paper, absolutely full of upper-level math mumbo-jumbo. The output of mathgen is utter nonsense, but a non-mathematician might honestly not be able to tell the difference half the time. (My first mathgen-erated paper sounded very suspiciously like one of my grad algebraic topology exam questions, with loopspace functors and homology groups, etc. I've noticed that it tends to use a lot of category theory, probably because category theory tends to always sounds magical, even to a trained ear). Often, there are numbered lemmas whose statements are nonsense, and for a proof there is the all-too-common "The proof is trivial." You might get a definition like: A Newton, `-globally Hermite, everywhere normal factor M is Artinian if Liouville's criterion applies. In fact, this was an exact definition in the mathgen paper Independent, Negative, Canonically Turing Arrows of Equations and Problems in Applied Formal PDE, that was submitted to the author-pays journal Advances in Pure Mathematics. The great tragedy is that this paper was accepted as well, claiming to be peer-reviewed. More on the paper, the submission, and a brief interaction between mathgen and Advances in Pure Mathematics can be found on that's mathematics. Unfortunately, the author-fee was never paid, so it never appeared in print.

There's also this article and this article, each submitted to Applied Mathematics Letters (which happens to be an Elsevier journal), which were retracted after terrible public response. Both the articles are rubbish, and one ends with:

Gödel’s incompleteness theorems put an upper limit to scientists in knowing the ultimate reality of the nature. The theorems express and explain that mathematics cannot solve everything. The problems will remain and remain for ever. Taking this logical fact into account both possible and impossible dominate mathematics. The one side of the coin states that a particular statement is valid but the other side demands and deduces that the same statement is invalid. This is a peculiar truth. Both science and spirituality came from space. Science is based on equations and experiments whereas spirituality relies on beliefs. The spirituality promises that everything in this universe was created in pairs but in opposites. For example, origin and end, man and women, light and dark, day and night, sorrow and pleasure, loss and gain, God and devil, ugly and beauty, good and bad and so on. Similarly, possible and impossible are consistent in mathematics.

How could that get through a review process? Applied Mathematics Letters is not an author-pays journal, but it is one of the journals that Elsevier was so kind as to release to the public.

I got side-tracked. The problem is that there are many scam, nonsense, or low-quality journals that are ruining the growth of the PLoS author-pays model. So what now? Perhaps some author-pays journals will come into existence, waive the fees for the first few years (living off of grants, maybe) to establish a reputation, and then start the fees after they've built a reputable name.

Or perhaps Gowers will step in with his next grand idea. As Gower's says, he has 'Joined the Bad Guys.' But by bad guys, he doesn't mean Elsevier, but instead the general idea of Gold open access journal providers. I use Gold open access (Gold) to refer to the idea upfront payment for the costs of dissemination, such as the author-pays model. (As an aside, this shouldn't be considered too foreign. It's essentially what radio and tv providers do, as broadcasts are paid for upfront. But they recoop the costs in ads, etc.). Gowers has been fundamental in the establishment of a new Gold math journal called the Forum of Mathematics. It's a slightly different model, which Gowers first talks about and later clarifies. If I can briefly paraphrase, the idea of the Forum of Mathematics (FoM) is not to be author-pays, but rather author's institution pays if possible. So if I, as a graduate student of Brown University, were to have a paper accepted to FoM, then I would ask Brown University to pay FoM $500 as my submission fee. In theory, if the university declines, then there will be no fee. Discussion about this can be found in the thread 'Why I've Joined the Bad Guys.' The goal is to not charge the authors - as this receives a lot of flac.

Why would universities pay? This is a harder question. Departments might realize that open access is good for everyone, and could lead to cheaper models of research dissemination in the future. In some cases, there are open-access mandates that force universities to cover the costs of certain journals. This is an idea growing in popularity.

I should take a moment to plug FoM, as they're now accepting submissions. Really, FoM is a pair or forums. There is a general interest forum and a special interest (e.g. number theory, differential geometry, etc.) forum. More details can be found at its site. Gowers has said that these journals intend to rival the top-quality journals in the world, such as the Annals of Mathematics. That would be something. In addition, FoM will not be released issue-by-issue, but instead each article will appear as soon as it is ready and reviewed. The first set of articles should appear any day now.

You might ask - how is FoM going to get over the initial hurdle? For the first three years, there will be no cost of submission. So it will build up a reputation. Best of luck.

A pointed response against FoM is given in the post Worse than Elsevier, on the blog noncommutativeanalysis. It raises some issues that I haven't mentioned, but which are mostly about conflict of interest. Gowers does a remarkably good job of responding to this in his post "Why I've joined the bad guys" (which I've linked to several times already, so now I'll refrain). But Gil Kalai, another mathematician who pops up all over the place and who I admire, has given a pointed response in a comment on Gowers' blog with both statements of support and hesitation. It's a much better thought-out response than that given in noncommutativeanalysis, in my opinion.

I hope that FoM becomes successful, as I too believe in open-access and open-source. A bit of momentum will benefit the movement, I think. And if an open access journal grows to rival or even compare to the Annals, all the better.

Much of the discussion I've linked to has taken place since the start of the year, and there's one big idea left. But I also wanted to point out that open-access is growing. Roughly 10 days ago, JSTOR, one of the largest journal archivers, has released many of their past archives (including many math journals) open to the public in limited quantities per time access. I think this is amazing. JSTOR archives tend to cover all but the last 6 years or so, which is quite a lot (but not the most recent). I've used resources from JSTOR more times than I can remember.

So this all sounds exciting. What more could there be? Gowers has just released that he has also joined 'The Good Guys,' by which he means that he is also supporting the establishment of a Green open-access set of journals (He calls it a Diamond open-access journal; but this should not be so. 'Diamond' indicates as clear, free, and transparent as possible, but the arxiv limits who can submit to it. Thus there are limits. In the comments, Gowers mentions calling it Platinum open-access, perhaps. But to be clear, it's more than classic Green, but not Diamond). Well, sort of. He's developing a platform that would make it very easy to set up Green journals. By Green open access, I mean that authors post copies of their articles on the web, and thus give open access. In other words, authors self-archive. This is already done all the time in the arxiv, but the problem is that there is no good method of peer-review handled routinely on the arxiv. So Gowers is part of a group, the Episciences Project, that acts as an arxiv overlay. (Really, the arxiv does such an amazing job that there is little reason to try to reinvent that wheel). The idea is that mathematicians will do the editing and refereeing freely, as they normally do, and will do without the copy-editing done in pay-for journals. The project is still in development, and Gowers indicates that it will come to fruition in April or so. (At that time, both Gowers and Terry Tao, who is aligned with Gowers in everything that I've discussed today, have indicated that they will likely serve as editors on one or more of the newly formed journals. Both are on committees for the Episciences Project and its development).

I am very excited about this. It is a good time to be a research mathematician.

This sounds almost like a higher MathOverflow - papers get 'submitted' for review, mathematicians voluntarily review them, and hopefully each will get a separate page for unified discussion and comments. This has the potential to be highly cooperative, and essentially free other than hosting services (the arxiv is supported by many institutions and is hosted by Cornell). An example of an arxiv-overlay journal is Electronic Proceeding in Theoretic Computer Science (which is a good start, but which could obviously be improved).

I can only hope that open access will grow. I'm looking forward to these new journals, and the prospect of quicker and facilitated peer-review. To conclude, I'd like to point out an article that talks about why MOOCs (e.g. see my previous posts) grew so much faster than open access.

Leave a comment

Comment via email