Bob Dylan said,
When you think that you’ve lost everything,
You find out you can always lose a little more.
Open access for the REF is not in that league, but when you think you have plumbed the depths of HEFCE policy on open access, someone pops up to tell you that you haven’t got to the bottom yet.
I am going to tell you two entirely realistic scenarios which could lead to papers being judged ineligible for the REF despite your best intentions, or at least to a degrading of the quality of research in the name of “excellence”. But a couple of preliminary remarks first.
Recall that you must put a preprint of your article (in the final accepted form for publication) in a public archive within three months of acceptance, from 1 April 2016. (Incidentally, it is not entirely clear that this start date refers to acceptance of papers. We know how slow mathematical publication is, and it is quite possible that a paper accepted years earlier only comes to published form after that date.)
During my career, I have been fortunate to know several mathematicians of the highest level of creativity, including Graham Higman, John Conway, Paul Erdős, and Ian Macdonald. I can imagine the reaction that these people would have had if someone tried to impose the current HEFCE rules on them. Some of them would simply not have complied. So then the HEFCE bureaucrats, who after all know what research excellence is since they invented the concept, would decide that these people were not up to scratch.
On that theme, my current contract ends one month and a day before the new HEFCE rules come into force. I hope it will be renewed; but if it is not, the silver lining of the cloud will be that I will no longer be bound by these silly rules. I will be able to do research, post it on the arXiv, and if I am really proud of it, submit it to a diamond open access journal, and that will be that.
And further diverting on that theme, it really seems that neither HEFCE nor one of the commenters on my previous post realise that there is any alternative to gold or green open access.
Back to general issues. How do you prove acceptance date of a paper? By the date on the editor’s letter notifying acceptance, apparently (with some exceptions: I found one journal which included an official acceptance date in the letter). So you have to keep this letter. It probably came by email. The two University email systems I have to deal with are Outlook and Office365 based, which means that all my mail is stored in a cloud under the control of Microsoft. It seems more than a little naive to assume that it will still be there in 2020 when you might need it. Moreover, these systems do not allow you to save emails as local files, unlike the old workhorses I used to use such as mutt or squirrel mail. I have asked various systems managers for a way round this, but nobody has been able to help. So I have resorted to copying the text (including headers) into a text file and saving that somewhere that will get backed up.
And, while I am on general things, Martin Eve said,
If your institution isn’t allowing you to use arXiv to fulfil the requirements, that’s not HEFCE’s fault, it’s your institution being over-zealous. The policy explicitly allows arXiv: “a subject repository such as arXiv”.
I mentioned the fact stated in the last sentence in my original post. But the whitewash of HEFCE doesn’t hold, since it is their policy which has driven universities into this over-zealousness.
In pre-Internet days, we had a system which worked well for making papers public. Journals would provide a number, typically 50, of “offprints” of a published paper (printed documents identical with the published version). Anyone could then write to the author asking for an offprint, which would be sent provided that the paper was not so popular that the supply had been exhausted. Departments usually had a supply of request cards which could be filled in and posted. The analogue of gold was the facility to buy extra offprints at your own or your university’s expense.
Right, down to business …
I write a paper, prepare it very carefully, and post it on the arXiv at the same time as submitting it to a journal. Let us say, either a diamond journal, or a gold journal for which my university is prepared to stump up a huge sum of money. Back it comes with referees’ reports pointing out that there is nothing wrong with the mathematics, but asking for small changes in grammar or style. I happen to think that these changes degrade the paper, making it less precise or harder to understand; but I want the paper published, I am busy, so I swallow my pride, make them, and send the paper back to the journal, who tell me it is now accepted. Now there is a superior version on the arXiv, but I can’t count the paper for the REF unless I post an inferior version. Moreover, the over-zealous administrators will probably expect me to post this inferior version on the institutional repository as well. Then finally it will appear in the journal.
So now there are four copies of the paper out there. The arXiv never deletes anything; if you update a paper, the old copy is still there. If you think about arXiv submissions of controversial papers claiming to solve big problems, you will quickly realise that this is the correct, indeed the only possible, strategy. (I don’t know if this is also true for institutional repositories.) So the “good” version is still there, but it is no longer the default, and you will only get it if you ask for it.
Is there anywhere a mathematician who thinks this scenario is not realistic?
Of course, if I decide to fight against this proliferation and depression of standards by not posting the second version, I can’t put the paper in the REF. So, even leaving aside the possible depression of standards, the effect of HEFCE policy is to fuel a big increase in the number of copies of a paper on the web, making searching for the “right” one virtually impossible.
I write a paper with an author in a different country. She does not realise the stupid bureaucracy that UK academics suffer from, and she knows that I am busy, so when the acceptance letter comes, she does not bother to forward it to me.
So how do I find out that the paper has been accepted? Maybe when it appears in the journal (or on their website), or maybe when I wonder why I have heard nothing and ask my coauthor. In either case, by this time it may be too late to satisfy the HEFCE requirement (even if the version on the arXiv may, as in the first scenario, be as good as or better than the published version).
I hasten to say that none of my co-authors would behave like this; they are without exception more conscientious than I am! (Not difficult, actually!) But it may be a bit intimidating or embarrassing for junior academics to have to nag senior colleages in other countries for these bureaucratic details.