Permanence of open access publication

Both email accounts I use have rather zealous spam filters, and from time to time I have to check what they have filtered out. (An excuse for slow response to email?) The last time I looked I found a copy of the International Mathematical Union’s Bulletin. Included in it, along with the sad news of Vaughan Jones’ death, I found a discussion and link to a paper on the arXiv documenting the authors’ research on the disappearance of open access journals over the last two decades.

Back in pre-Internet days, libraries subscribed to journals, and you could go to the library and read a back issue (except that, if it was very old, you might have to order it from the stacks). Sometimes, scandalously, the library would pulp its old journals on the grounds that it could no longer afford the cost of storing them. But there was also the option of Inter-Library Loan: if your library didn’t hold the journal issue you were after, you could order it from another library (at a small charge, which the University would pay).

Now all is different. The existence of the Internet encourages small groups to start up journals. Usually these are run by volunteers. (Even the big publishers get the editorial and refereeing work done by volunteers, so this is not such a big difference.) Also, they are free, both to authors and to readers: “diamond open access”.

So the study in question, by Mikael Laakso, Lisa Matthias and Najko Jahn, asked what happens if you need to refer to a paper in one of these open access journals and find that the journal is no longer in operation. This might happen for many reasons: someone dies, retires or loses interest; something similar happens to an organisation; a change in technology has made the papers unreadable or unfindable; and so on. Your local library will not have hard copy, and maybe Google is unable to find the paper from the incomplete information you hold.

Clearly this is a serious problem.

One of the first organisations to address this problem had the acronym LOCKSS (“Lots Of Copies Keep Stuff Safe”). Now there are several others, all operating a little differently. The idea is that publishers agree that they can keep copies of the papers; if the copyright lapses for any reason, they can make their copies freely available. But of course this costs money. Small groups of volunteers running a journal may not be able to afford the cost; and, in any case, they are more focussed on getting papers published than on their long-term security.

By various ingenious methods, including using the Wayback Machine (which provides snapshots of the internet in former times), Google searches, and information from friends and colleagues, they were able to identify 176 open access journals that have disappeared in the last two decades. The dataset is here. They point to the urgency of the problem and encourage use of the data for further research. According to the IMU, two of the 176 journals are in mathematics. (I have not checked this.)

Actually I think there are wider issues.

First, a policy issue. It seems reasonable to assume, though I have no data on this, that open access journals are more likely to disappear than subscription journals. While the papers are behind a paywall, they are a source of income for the publisher, who is thus likely to ensure their continued existence. They may be less inclined to support something which brings them no further income. On the other hand, funders are putting very strong pressure on us (for good reason) to publish in open access journals. Our research is publicly funded, so the results arguably belong to the public, who should have free access to them.

Second, not every document worth preserving is published in a journal, open access or otherwise. The pressure of research assessment is forcing us to submit our papers to higher-prestige journals (since our administrators find it easier to judge their quality by looking at the journal impact factor). Thus, even good papers get rejected, and after a couple of rejections the author may get discouraged and the paper languish unseen. Nowadays we are likely to put the preprint on the arXiv, but many of us have various papers from earlier times which may still be of interest, either on personal web pages or in our filing cabinets.

I have many such files. A few of them have seen the light of day, mostly on the arXiv, such as this one, a paper I wrote in around 1980 showing that most Latin squares and Steiner triple systems have trivial automorphism group (I was scooped by Laci Babai but later he encouraged me to make my version public); this one, on some group invariants related to permutation bases; this one with Sam Tarzi, on the outer automorphism group of the automorphism group of the random m-coloured countable complete graph (no, that is not a typo). As well I have several other papers on the arXiv which were never published because they were rejected by journals, such as this and this. And elsewhere there is this essay I wrote for John Cannon when he was developing Magma.