Matilde 6, 16-17, 2000
During AMS Scand 2000, the Danish Mathematical Society hosted a panel discussion on the enhanced availability of mathematical databases and the changes in the ways of publishing mathematics. The panelists were Donald G. Babbit, AMS Publisher and Bernd Wegner, Editor in Chief of Zentralblatt für Mathematik. The discussion was chaired by Bodil Branner, DTU, President of the Danish Mathematical Society. Here we summarise the issues and opinions covered by the presentations of the two panelists and the ensuing debate.
With the large number of papers being published it is increasingly important to have tools to help navigate the mathematics literature. Evaluative reviews play a useful rôle in this, but both articles and reviews have much to gain from modern technology.
Currently more and more of the refereed literature is available on-line. Older journals are being scanned, digitised and made available by organisations such as JStor [4]. Traditional journals are introducing electronic versions. New, and often free, electronic journals are being created; a few of these, built on top of preprint archives, e.g. [1]. These developments open up the possibility of having direct links from indexes and reviews to the original articles. MathSciNet, the on-line version of Mathematical Reviews, now contains a large number of such links.
For truly electronic articles, one can and should go further and provide links from the list of references to the reviews of the referenced articles. Merely putting up postscript or pdf files on the web is not taking full advantage of the technology. The AMS has introduced the free tool MR-lookup [5] that provides a simple search interface to reviews and bibliographic information, including a permanent URL for the review. Such links are now being included in on-line versions of articles published by the AMS, and other authors and publishers are free to include these URLs in their articles. This method supplies a two step link from reference lists to original articles via reviews. If user does not subscribe to MathSciNet, then the links are provided via bibliographic information instead.
Direct linking from references to original articles is another possibility that is being pursued by some publishers. The critical piece of information here is a unique persistent Digital Object Identifier (DOI) for every article [2]. This is a number of the form publ.xxxx. The first part is a prefix specific to the publisher, and is obtainable for a fee (currently 1000USD) from the International DOI Foundation. The second part is assigned by the publisher to the individual articles. DOI's are stored in a central registry together with information such as the author, title, journal, first page, etc., and a URL for the article. This registry can then be searched by publishers who wish to include links in their articles. A version of this system was originally proposed by the IMU. However, the library community seems to be very wary of the current form of the proposal: the main participants are the large commercial publishing houses, and the fees involved are potentially a burden for smaller publishers. The AMS has withdrawn from the DOI scheme and now supports the review based model described above.
Providing good search facilities for the literature is not as easy as it might seem. Mathematical Reviews spends about half a million dollars a year on author identification alone. The problem is that the name of a given author appears in different forms on different articles. The simplest case is when first names are given in full or just as initials. Much worse is the many possible ways of transliterating e.g. Russian names. Babbit gave the example of I. M. Gelfand who has 477 publications listed in Mathematical Reviews containing 20 different variations of the author's name. There can be similar problems with identifying journals from abbreviations.
For the future, the AMS would like to add forward references to Mathematical Reviews. Currently, the review of an article A contains links to all other reviews that cite the current article. But ideally this should be extended to include links to (reviews of) all other articles that have article A in their list of references. Babbit suggested that this could be provided for articles in the top 100 journals in the next 2 to 4 years.
All in all, Mathematical Reviews is an expensive service to create and improve. The AMS wishes to have the widest possible access, but money is needed to maintain and update the services. For example, keyboarding the old issues of Mathematical Reviews from 1940 to 1970 in to the MathSciNet database cost about a million dollars. One method of increasing access is creation of consortia, whereby a number of institutions subscribe collectively at reduced rates. The first of these agreements was made with Denmark, via the Euromath Centre, in 1997, and there are now about 260 consortia. The size of the saving is determined partly by the number of new subscribers the consortium brings.
The European Mathematical Information Service [3] provides documentation of the mathematical literature, searchable databases, a publishing portal and access to electronic journals. There is some overlap with the services of the AMS, but two distinguishing projects are the Jahrbuch database, which contains reviews of mathematics papers published between 1868 and 1943, and the MathDI, a preprint database for mathematics education.
Improvements to the Zentralblatt MATH Database with better hyperlinking and references to on-line versions of articles are being funded by LIMES, a European Framework V project. There are currently 9, soon to be increased to 14, editorial units spread across Europe, in contrast to the centralised Mathematical Reviews structure. This enables regional and national information to be made available and helps to reduce costs. There is also on going support for smaller publications to be improved and made available electronically.
The journals, proceedings and monographs made available on EMIS have their full content mirrored at 40 different sites. They are supervised by a board appointed by the European Mathematical Society. This board provides quality control for the journals and proceedings and aims to create and give access to freely accessible mathematical publications.
Currently there are about 1500 mathematics journals worldwide. Of these about 400 may be regarded as core journals. They are poorly covered by the Science Citation Index and there is not much reliable information about the quality of individual journals. However, it is estimated that 200 to 300 hundred could publish electronically using their current material. EMIS is actively encouraging and supporting current academic paper only journals to start up electronic versions.
Rolf Jeltsch, President of the EMS, told how the European Mathematical Society is considering setting up a publishing house to combine resources within Europe, help to keep journal prices reasonable and to start up own journals (such as the Journal of the European Mathematical Society). This will be discussed at the Third European Congress of Mathematics in Barcelona. However, the EMS is not rich and there needs to be some mechanism to ensure financial stability.
Growing journal prices are currently a particular problem, since library budgets are not increasing at the same rate and many are being reduced or at best held stable. From the publishers there are two contrasting ideologies regarding journal pricing: i) take as much money as you can get from the market; ii) balance income against costs.
What are the production costs for a journal? The scientific publishers tend to be very open about this, whereas it is hard to get information from the commercial publishers. In mathematics many services are provided free: authors of manuscripts receive no payment, referees do their work without remuneration, much work of the scientific editors is done on a voluntary basis. The costs are thus mainly associated to printing, distributing and providing electronic enhancements. Presumably, these costs are lower for electronic only journals.
Wegner suggested some ways individuals can help to influence pricing policy: restrict our voluntary services to fairly priced publications; only submit our papers to such journals; support publication through scientific publishers; organise cooperation between university publishers; support projects to establish free publication.
Babbit told of one particular initiative in the US. There, an organisation called SPARC [7] has a project to target expensive journals and try to get them to reduce their prices. One mechanism they have used is to go to the journal's editorial board and offer to start up a rival journal with the same coverage but at a lower price. SPARC has many American libraries as members and these guarantee to subscribe to the new journal. This gives a sound financial basis for the new publication with at least 200 subscribers from the outset. (Some of these initiatives are described in a recent article by Joan Birman [8].)
The AMS itself is not at present keen to start up new journals. They currently have two non-free journals that are only available in electronic form. However, these have not attracted many subscribers, despite their low cost and offers of free subscriptions for those who already take other AMS publications.
There is a perceived problem with the authoritativeness of electronic journals, despite the fact that they are often operate under the same rigorous standards as paper journals as regards peer review of articles. This can be a particular difficulty for young mathematicians seeking jobs and is an attitude that we need to change.
Archiving: There is confusion about who is responsible for archiving materials. The AMS archive their publications and promise to convert TeX files to new language if this becomes necessary. A large proportion of electronic journals also have paper versions that may be archived in traditional ways. Many libraries see it as their responsibility to keep archives and this attitude is supported by the EMS. Journal publishers come and go, and if they hold the only archived copies of papers, then access is hard to guarantee. It is important that libraries also take care of non-text material. Here, there are even greater problems of ensuring that access is maintained as technology changes.
Individual copyright: This does not seem to be a big issue at the moment as the numbers are small. There has been little problem with authors who simply delete the parts of the copyright agreements that they do not like. After some pressure, the AMS has agreed that individual authors may choose to retain the copyright of their articles, but very few (less than 1%) have taken up this option.
Personal home pages: are not reliable places to link to, and are not refereed. MR and Zentralblatt concentrate on the refereed literature. Under EMIS, there is a preprint database MPress. This is based on a worm that collects data from certain other preprint databases around the world that attach suitable metadata to their papers. Systematic links to local holdings are not yet available, although lists of department home pages can found at several sources.
The panelists were interested to see how the discussion has changed and matured from four years ago. Comments are now at a much more advanced level and the need for taking electronic publishing seriously is widely recognised. This area is rapidly changing and it is important that mathematicians work together to ensure continuing affordable access to the research literature.
[1] Advances in Theoretical and Mathematical
Physics: <http://www.intlpress.com/journals/ATMP>
[2] Digital Object Identifiers: <http://www.doi.org>
[3] European Mathematical Information Service: <http://www.emis.de>
[4] JStor: <http://www.jstor.org>
[5] MR-lookup: <http://www.ams.org/mrlookup>
[6] AMS Journal price survey: <http://www.ams.org/index/membership/journal-survey.html>
[7] SPARC: <http://www.arl.org/sparc>
[8] Joan Birman, ``Scientific Publishing: A Mathematician's
Viewpoint'', Notices of the AMS, Volume 47, Number 7, August 2000,
pp. 770-774