A Scenario for Publishing Mathematics in the Future

Abstract

Once the electronic version exists, it should be sent to an e-print archive, the obvious one, at this time, is xxx at Los Alamos (http://xxx.lanl.gov). This follows the time honored tradition of mathematicians sending out paper preprints of their work. These preprints enlightened the recipients long before the paper appeared in a journal (usually years, not months later), and sometimes established priority. The defect is that sending out paper preprints is somewhat laborious, and they reach only the "right" people. The electronic version is simple and democratic (once the e-print is at xxx, then it is available to the world).

We mathematicians have a long and valuable tradition of putting papers through a refereeing and editorial process. This is fundamental to getting (usually) papers whose accuracy we can trust, and which meet (usually) some minimal standard of readability. This tradition should not be weakened, so the second step after the e-print archive is to send the paper to a journal. The journal should continue to perform its usual function of refereeing, editing, and accepting or rejecting. If it accepts, it publishes.

This is a simple scenario, involving only one simple extra task (which should be considered in the author's self interest), namely, sending the electronic version to the e-print archive. But it has some interesting consequences.

First, the author should retain the copyright, allowing only a license for the publisher to print its version and do whatever they do currently with that version. Or, at the very least, the author should retain the right to keep the paper at the e-print archive, which after all, is an archive. What is the point to making all or almost all of mathematics available quickly in one central location (the e-print archive) if one is then going to take away some of the papers?

It is then easy for an electronic journal to exist. It need only create an overlay, rather like a web page, which describes the journal and gives a table of contents of published papers; if the reader is interested in a certain paper, he/she clicks on the paper which is then retrieved, not from the journal, but from the e-print archive where it will reside for all time. Thus a fledgling electronic journal does not need to set up its own smaller version of xxx, but only needs to piggyback on the existing e-print archive. Incidentally, the archive can include the additional information of which journal has accepted the preprint.

The point to this electronic journal is that it has given a "Good Housekeeping seal of approval" to its papers. This is just as important in the future as it has been in the past. This work has always been done essentially for free by mathematicians acting as referees or editors, and it would continue so. How "good" the "approval" is depends, as in the past, on the standards of the journal.

What about paper journals? Certainly the very best journals will continue to exist in paper form. Probably some of the weaker journals should stop bothering with the cost of creating a paper version and be electronic only. What decides this will be up to libraries and individuals who will vote with their pocketbooks as to whether they want a paper copy of a journal badly enough to pay for it.

Journals will be forced to be very efficient in producing a paper version. Otherwise libraries and individuals are likely save money by skipping the paper version and relying on the e-print archive. The journals may only publish a yearly volume, rather than monthly, quarterly, or whatever, simply in order to reduce mailing and other costs. Libraries may prefer a yearly volume, for timeliness is no longer as important since the papers reside at the e-print archive for those who need them before the end of the year.

Journals are likely to require some standardization in the authors' TeX files, and also to relax their own standards of beautifully and uniformly typeset journal pages. This is one of the big costs for journals, and it is not at all clear that readers would pay for it if it came out of their own pocket. On the other hand, authors, after having labored over a paper, like to see it beautifully reproduced in a handsome journal--it validates their work to some extent. So it is not clear how this aspect will evolve in the future.

But I would guess that most journals will contract with very efficient printers and distributors so as to keep their prices low so that libraries will continue to subscribe to their paper journals. I would expect to see vendors arise that contract with a number of electronic journals to efficiently print and distribute the paper version. Competition between vendors should keep prices low, for the mathematicians would be running the electronic journal and would (being non-profit) want to see their journals on many library shelves.

Where does this leave current commercial publishers like Elsevier or Springer? It is hard to see them surviving in mathematics unless they change dramatically. For example, any list of the best math journals is likely to include the Annals of Mathematics, Journal of the American Mathematical Society, and Inventiones Mathematicae. They cost, respectively, 15 cents/page, 15 c/p and 110 c/p. You can guess which is published by Princeton University Press, the American Mathematical Society, and Springer. The subscription base for Inventiones is roughly half that of the Annals, and once most papers in Inventiones appear on the e-print archive, how many libraries are going to continue to subscribe to Inventiones when it costs seven times as much?

It is claimed that there are hidden subsidies to the non-profit publishers which make these comparisons unfair. The only source of subsidy for the societies is dues, but many societies, including the AMS want to subsidize other activities through their profits on journals and books. University press journals do get modest hidden subsidies (e.g. university staff sometimes get part salary for journal work from the journal, but work in university offices on university equipment). But it is hard to imagine these subsidies even doubling the cost of the journal, and if Annals of Mathematics cost 30 c/p, how much difference would that make in the comparison with Inventiones?

Here is a recent anecdote: a colleague, X, recently edited a special issue of of Elsevier's journal Chaos, Solitons and Fractals devoted to papers in the area of colleague Y. Y was interested in buying that issue, called Elsevier for the price, and was told $346. He didn't expect cheap, but is "still reeling".

It takes time, for most mathematicians don't pay too much attention to these matters, but there is a growing list of us who are coming to realize that the high priced commercial publishers are no friends of mathematics. More and more of us are refusing to submit papers to, or referee for, or edit journals with high subscription rates. Those who edit conference proceedings only need a couple of their authors who are aware of prices, to convince the editors to seek out low cost publishers. The electronic ways of distributing math research are attractive in their own right, and the changeover from paper to electronic distribution is only spurred on by the high prices of some commercial publishers.