On refereeing

David Fremlin

I have recently been refusing most requests to referee papers, and perhaps I should explain my reasons in a place where there is a chance that editors will notice them before wasting their (and the authors') time.

The present system of publishing in academic journals is not exactly broken. But it is extremely expensive. (The University of Essex spends about 600,000 p.a. on journals. That is something under 1% of the University's total expenditure.) I believe that most of this expenditure is unnecessary; let me explain.

Traditionally, journals provide three services:

  • dissemination --- making new facts and ideas available to those who may be able to use them;
  • archiving --- providing permanent date-stamped records;
  • validation --- confirming that published material is new, true and interesting.

    All three of these are essential. But the first two can now be done better, and very much more cheaply, by electronic means. I do sometimes, even now, cast my eye over the contents page of a journal. But the overwhelming majority of the new ideas which come my way reach me by relatively informal routes --- usually personal recommendation --- and I then go to a journal only in order to check that the printed version coincides with what (with luck) I have already been able to download. Printed versions are, indeed, the currently best-organized archive; but with a very small collective effort the Mathematics ArXiv could surpass them. There are questions about the permanence of electronic storage systems --- but then there are questions about twentieth-century paper, and electronic storage is so cheap and easy that (again, with a tiny collective effort) far more copies can be preserved and made generally available than is possible with printed matter.

    For validation there is indeed no satisfactory alternative, so far, to the present system. (Wikipedia is a most interesting innovation, but depends critically on the character of its editors, and I am not sure that it offers a prospect of long-term reliability.) I do not see how an academic contribution can be assessed without concentrated effort by an appropriate expert, and I do not see how to organize such efforts except through a network of editors and referees very much like the present one. This is work which needs to be done.

    However, it is exactly this necessary part of the process which is not being paid for. It is a service offered, willingly if not always exactly gladly, by individuals who are giving their time (their personal time, if they are British, since no recognition of this work is incorporated in the Research Assessment Exercises on which our departments' prospects depend) for the benefit of the community. Certainly I have sometimes profited from editors' being on the lookout for papers which might interest me. But I have come to seriously resent the way in which my work is being used to prop up a system which is exploiting the taxpayers of rich nations and depriving the scholars of poor nations. I note that this system is actually very remunerative. Elsevier, as a public company, have to say what they are up to, and they are doing rather well out of it. The big mathematical societies, under the magic `not-for-profit' label, are allowed to be less forthcoming; but a lot of people are in reasonably comfortable jobs processing trees which should have been left standing.

    For this reason (I come finally to the point) I normally do not referee for journals which do not have free electronic versions. By `free', let me say, I mean something which I can get on my home computer. I am prepared to be open-minded about delayed access; but I think that the group at the University of Nepal who have been trying to set up a functional analysis seminar ought to be able to see something of what has been done in the current decade.

    Mathematicians are healthily resistant to claims of moral virtue, and I do not ask anyone else to do anything more than consider the arguments above. But I note that they apply at least as strongly to submitting papers as to refereeing them. So if you find that what I have written is persuasive, I suggest that you should look at the electronic-access policies of journals before submitting papers to them.

    Back to home page.

    Revised 29.9.06