30 December 97

Mr. Nigel Stapleton, CEO, Reed Elsevier

Mr. Russell White, President, Elsevier Science Inc. Dear Mr. Stapleton and Mr. White: I am a mathematician (in particular a topologist) writing to you about scientific publishing (particularly in math, and more particularly about your journal Topology). During the past year I have been gathering information, thinking, and writing about the high cost of commercial math journals and alternative methods of publication. Let me explain my view of the future of math publishing and how it relates to Reed Elsevier. The latest issue of Topology shows that it takes, on the average, about two years from the time a paper is submitted to the time it is published. This is typical in mathematics, and probably hasn't changed much in decades. Thus, in order to disseminate their mathematics quicker and to establish priority, mathematicians send out preprints. Mathematicians want to receive preprints so as to keep up with the field without having to wait for the journal to appear years later. For these reasons the preprints are crucial to our profession. The journals are crucial because of their role in refereeing, vouching for the worth of a paper, and as an archive. By now, almost all mathematicians produce TeX versions of their papers, and distribute an electronic preprint. This makes distribution easy, and it would be still easier if mathematicians were more willing to standardize their personal versions of TeX. What I would like to see happen, and is beginning to happen, is the following: there would exist preprint servers in each subfield of mathematics (a subfield being 2 to 5% of math, defined in a natural way, for example mine would be low dimensional topology). Once an author has produced a suitable TeX version of a paper, s/he would send it to one or more preprint servers which would list the paper on-line for eternity. I could go to a server that interested me and ask for a specific paper, or all papers by a certain author, or all papers in a certain subject since a certain date, or all papers with certain code words; I might have a subscription which sends me an abstract each time a paper is listed. Such a system has the following advantages: (1) assuming almost all mathematicians join this system, I have quick and efficient access to papers that interest me, which I can peruse on my computer screen and print if I wish (contrast this with waiting for publication and then hoping my library subscribes so that I can laboriously xerox page by page); (2) I can easily distribute my own papers to any interested reader (contrast this with duplicating a paper preprint, stuffing envelopes and addressing, and then missing part of my interested audience). Such preprint servers already exist. One in high energy physics is working splendidly, run by Paul Ginsparg at Los Alamos National Lab. Some servers exist in math, and we are in the midst of organizing a more coherent system to cover all of math, done through Ginsparg's operation. I think we are off to a good start in "Geometric Topology" (see http://front.math.ucdavis.edu/ ). These preprint servers are only meant to be a better way of maintaining the old preprint system. It is still vitally important, in my opinion, that we mathematicians retain our tradition of having papers refereed and accepted (or rejected) by journals of varying reputation. This should continue, and can also be done electronically. We have started an electronic journal, Geometry & Topology, based at the University of Warwick (see http://www.maths.warwick.ac.uk/gt/ ). It is free, at least in the foreseeable future. It is off to an excellent start with a large distinguished board of editors and standards at least that of Topology. We expect to offer a paper subscription at a very low price, determined by the cost of a commercial printer using our TeX files. Of course, there are the hidden subsidies to G&T by Warwick, such as supporting the computers that G&T uses and some time by the editors. Where do the commercial publishers come into this picture? We mathematicians write and TeX our papers, we referee and edit for the journals, and we are well paid by our universities for these tasks. We then turn over the papers, with copyright, to the publishers who add relatively little value in producing the paper volume from our TeX files, and who then turn around and sell the journal to the university libraries at what is, in some cases at least, an exorbitant price. Last spring, I collected some data on prices of journals which can be found at /~kirby/journals.html in the appendices. The data is given in price per page and price per 10,000 characters. The latter is more accurate, but the former is easier to understand. Briefly, an efficient and cost conscious journal like the Pacific Journal of Math can be sold at about 13 cents/page. It is a non-profit company, with only minor subsidies from a few universities, and a 40+ year history of low costs and good mathematics. Perhaps the three most prestigious math journals are the Annals of Math (Princeton Univ Press) published at 15 cents/page, the Journal of the American Math Society at 15 cents/page also, and Inventiones Mathematicae (Springer) at 110 cents/page. Your journal Topology costs us 81 cents/page. Berkeley subscribes to 13 of your math journals, and the average cost is 73 cents/page, ranging from a high of 135 cents/page for Nonlinear Analysis to a low of 47 cents/page for Linear Algebra and its Applications. Wolters Kluwer averages 67 cents/page for 5 journals. Springer averages 82 cents/page, Academic Press 40 cents/page, and Birkhauser 68 cents/page, for comparison. The non-profits are usually less. So this brings me to the main point of this letter. I read in yesterday's New York Times about Reed Elsevier expanding dramatically and having high margins in its scientific publishing. Yet it looks to me as though you are on your way out in scientific publishing, at least in math. How can you stay in business offering a topology journal at 81 cents/page when you have competitors offering roughly the same product at under 20 cents/page? What's going on here? I know part of the answer. There is a time lag. Your prices have been going up steeply in recent years. Mathematicians pay little attention to these things for we subscribe to few journals as individuals and in those cases get much lower prices if our libraries also subscribe. Our librarians have noticed the higher prices, but it has taken a while for them to get our attention. But many of us have woken up by now. (Furthermore, publishers added much more value in the old days of typesetters and galley proofs.) But now that we are awake, we also notice the alternatives. In my field, we have a free electronic journal and are on the way to having a preprint server. By the time our library decides on which journals to subscribe to this summer, I expect that much (most?) topology will be on the preprint server, some good topology papers will have shifted away from Topology to lower cost journals, and I will be able to recommend to my library that they drop Topology (and I will, alas, drop my personal subscription). Perhaps this will be delayed a year. When Berkeley drops Topology, and editors, referees and authors begin to desert Topology, how long will it continue to exist? An interesting issue here is copyright. I understand that, at the moment, you allow authors to list their paper on their web site, but do not allow (once the copyright form has been signed) authors to put their papers on preprint servers. I'm not surprised at this restriction, given what I wrote in the last paragraph. But why should authors give up the right to use a preprint server when it is obviously so much in their interest that everyone use preprint servers? You might win that battle if Topology were unique enough, but if you demand a restrictive copyright, authors are likely to go to your competitors. So you see boom and I see bust. Your markets in other areas such as law may be in good shape; things vary widely between academic disciplines. But I think I know math pretty well, and I don't see much of a future for high cost commercial publishers. But I would be sorry to see Topology disappear. It's an old friend. I published my thesis there in 1967. From my desk I see all the back issues, and know many papers therein. The editors at Oxford are friends and have worked hard from the beginning to make Topology a journal of stature. But all that sentiment can't overcome a price that's too large by a factor of 4. It is presumptuous of me to tell you how to run your business, but here are a few possibilities anyway. I'd guess that you could cut the price of Topology in half, and probably more, and still make a profit. After all, there are economies of scale and you have much experience and a tradition to help. American automobile manufacturers who thought they were pretty efficient in the 70's found they could double their efficiency when they ran into Japanese competition. Also, you apparently have high profit margins in scientific publishing which could come down some. Announcing a price cut to 40 cents/page, with promised cuts in the future, would probably save Topology, at least for quite a while. But this doesn't seem too likely when I read about 3 to 5 year contracts with 9.5% increases locked in, and only some electronic enhancements to offer us (those electronic enhancements look modest compared to what we are doing ourselves with preprint servers and electronic journals that are free). Best for us would be for you to simply give the journal Topology to its editors in Oxford, to run as a non-profit as the Pacific Journal is. All you'd get is some good will from topologists. I think that proceeding as you are will just run journals like Topology into the ground. There may be profits there for another few years. But then all that remains is ill will from some academics who didn't want a favorite journal to die. I'd also encourage you to talk to us. We mathematicians produce the math for your journals, and we urge (so far) our libraries to buy your journals. Of course it is costly in time to talk to us. I will circulate this letter to other mathematicians, and will be happy to circulate your reply in the same way. That is one way to reach us. Yours, Rob KirbyHere is the NY Times article which inspired my letter to Elsevier New York Times, 29 December, 1997, page C2