Suggested Changes to the Copyright Transfer Practice

I want to present gere some of my thoughts and ideas on the current relations between mathematicians and journal publishers.

In this context, the main role of publishing companies is to disseminate mathematical knowledge. It is reasonable that they charge for their services, or even make considerable profits, provided this does not impair the quality of their service.

However, we see some quite alarming trends in recent years. It seems that the high prices on many paper journals (which often exceed by far any reasonable estimates of the publication costs) force academic institutions to opt for a temporary access to the online versions. For example, even Carnegie Mellon University, one of the most prestigious technical universities in the USA, stopped (in 2004) the paper subscription to Discrete Applied Mathematics, Discrete Mathematics, Journal of Combinatorial Theory (both Series A and B), and many other journals.  Thus the access to these issues for the future CMU researchers will completely depend on the publishers' providing the on-line service at an acceptable price.

At the same time, a typical copyright transfer form that the publishers currently require the authors to sign gives the publishers a practically unrestricted control over the contribution. (For example, do mathematicians have any, even slightest, control over the journal prices now?) Essentially, the only legally allowed ways for disseminating the article without the publisher's permission are limited to the authors' personal web-pages and sharing (p)reprints with colleagues. This relies on the authors' willingness and possibility to do so; thus in no way it can serve as the primary source of scientific reference. (Especially that many publishers explicitly forbid putting the final version on web.)

There are could be many reasons why an article cannot be obtained from the author:

1. the author is dead either physically or in the meaning used by Erdös (that is, not doing research anymore);
   
2. the author is not so impeccably organized as to have copies of all papers readily available;
   
3. the author  is too busy to put preprints on-line or post copies on request.
   

Hence, in 10-20 years private publishers will hold exclusive rights to a large share of the mathematical research published today. So, if huge prices will happen to generate more revenue (even if fewer copies are sold), it is very likely that the publishers will take the advantage of it. (This seems to happen already.) It is obvious who will suffer most in this scenario.

In my opinion, a possible solution to this problem is to change the terms and conditions of the agreement between the authors and  the publishers. In a nutshell, I suggest to use the essentially the same copyright form, except for the following clause (which is also included into the text of the article):

"After [DATE]  this article can be freely copied in any quantities, on any media, and by any parties, also for the purposes of selling. The content of the article, including this notice, must be passed unaltered at all times."

This means that for a certain period of time (say, 10 years after the publication date) the Publisher holds the rights comparable with the current practice. This is a sufficiently long period to make institutions buy either the paper version or the electronic access. At the end of this period, the published article can be freely duplicated, on any media, in any quantities, by anybody, provided its content is not altered in any way. Similarly to the GNU General Public License, it is reasonable to allow anybody to resell the article after the expiration date. The clause stipulating this and stating the expiration date is to be included also into the content of the article and, therefore, must be always passed unaltered.

The advantages of this proposal are that publishers can still make profits (and thus will be interested in maintaining their services) while mathematicians, especially those from poorer institutions, will clearly benefit. The examples of some successful journals, such that Annals of Mathematics or Electronic Journal of Combinatorics, show that it is possible to combine a high scientific level even with the immediate free full-text access. Also, the articles bearing such a clause are clearly identifiable.

Suppose (as a wishful dreaming) that all major publishers announced that all pre-2000 articles can be copied without any restriction. It is likely that almost all important articles (e.g. the original Szemerédi's regularity lemma) would become available for free somewhere in the cyberspace fairly soon. This is so because the existence of just one person who scans the article into a pdf-file and puts the article's authors and title inside the link (so that google can index it) would be enough here. A few enthusiasts should suffice for starting a repository of published articles with expired copyright. Probably, there would be enough volunteers to develop a free database, not inferior to any of those currently offered by publishers. The examples of e.g. Linux or the current refereeing process, both based on people's willingness to contribute their effort and time to a good cause, make me believe so.

Of course, there are drawbacks too. One of them is that, even if a precedent is set, it is not clear if the other journals will endeavor to follow it. Also, the transition to new terms may disrupt some established routines. But, in my opinion, the mathematical community would overall benefit from publishing contributions under new terms.

Obviously, we cannot expect private publishers to embrace this initiative, since it will eventually cut down their income. The most effective way would be for a large group of mathematicians (e.g. AMS, ICM, etc) to organize a referendum on these issues and, if the outcome is strongly in the favor of changing the current system, to publicize widely the results. (So that academic publishers would have to address this issue somehow or risk losing their reputation of being supporters of science.) Unfortunately, this does not look very likely to happen in near future.

My next step would probably be to ask the publishers to publish my accepted papers under the proposed new terms. This approach is extremely unlikely to succeed, but my modest aim at this stage is to draw the attention of other mathematicians to this problem (and to a possible solution). I'll try to record here any failures or successes in this direction at this webpage.

I won't be collecting any signatures or maintaining a discussion group, but I'll be very glad to hear any suggestions or comments on the topic. 

Thank you.

Oleg Pikhurko

New Developments

September 2007:

Recently, I learned of an earlier article by John Ewing in the Notices of AMS who made similar suggestions (but much better argumented and designed than mine). As of now, his ideas are implemented by the American Mathematical Society. Namely, the authors of journal articles have the option to dedicate their article to public domain 28 years after publication: if they opt for this, then the corresponding statement appears on the title page of the published article. So far this scheme seems to work and not to cause any legal problems for the publisher or the authors.

Therefore, I thought this to be a good time to raise the issue again with the Elsevier (see here) and to see if Springer, another major publisher, would be willing to implement these changes (see here).

Also, the Australasian Journal of Combinatorics seems to give a free on-line access to papers that are 4 or more years old. Very nice of them!

October 2007:

Unfortunately, the experiment with the Springer was unsuccessful: I was faced with the alternatives: either to sign the stardard copyright transfer form or never ever publish with "Graphs & Combinatorics". (So my decision was obvious.) It is unfotunate and rather surprising, since I have always regarded the Springer as the friendliest one among the major scientific publishers (because, for example, they have the cost of many textbooks and monographs relatively low when compared with some competitors). Yet,  their reaction to my request was the most negative so far. Although it formally came from the editors of the journal, one can notice by looking at the correspondence that Mrs van Straalen, Head of Rights, Rights and Permissions Department of Springer chose not to reply to my emails or to voice any opinion on the matter at all, thus implicitly endorsing the decision of the editor of "Graphs & Combinatorics".
 
February 2008:

As it turned out, my papers with "Graphs & Combinatorics" did come out with the "Reverts to public domain 28 years from publication" statement on the title page :-) I do not know whether this was an overlook or intentional, but this happened. 

Also, I discovered that NIH (the National Institute of Health) will require from April 2008 that the scientists supported by its grant are required to provide a preprint of any article published to be included in a free-access server (read more).

I was involved with a similar correspondence with Wiley (in January 2005) about our paper "How Complex are Random Graphs in First Order Logic?", where we tried to do what Reinhard Diestel suggests. The publisher (Wiley) outright refused to published our paper unless we sign the unaltered copyright form, so we, the authors, had to agree. (Unfortunately, I have not made any copies of emails from that discussion.)

No answer from the Elsevier...

Conclusions

1. The current copyright transfer system does not work well.
   
2. Most of the major journal publishers (Elsevier, Springer, Wiley) are very reluctant to change their copyright policies. 
3. Also, not all scientists agree with the arguments put forward here.
   
4. There are viable solutions to this problem, the best one probably being John Ewing's suggestion.
   
5. There are some big organizations that are in favor of free availability of scientific publications (AMS, NIH, SPARC).

Related Pages

Correspondence with the Elsevier

Another attempt with Elsevier

Correspondence with Springer