Mathematics Journals: What Is Valued And What May Change .

3m ago
9 Views
0 Downloads
410.50 KB
43 Pages
Transcription

Mathematics journals: what is valued and what may change.Report of the workshop held at MSRI, Berkeley, California onFebruary 14 – 16 2011Mathematics relies on its journal literature as the main conduit for peer review anddissemination of research, and it does so more heavily and differently than other scientificfields. The conflict between universal access and the traditional subscription model that fundsthe journals has been debated for the past decade, while hard data on financial sustainabilityand usage under the different models has been slow to appear. However, the last ten yearshave seen the move from print to the electronic version of journals becoming the version ofrecord, and the workshop took an evidence-based approach to discussing dissemination,access and usage of mathematics journals.The workshop goal was to discuss what is important and unique to the publishing ofmathematical research articles and how we can best ensure that publishing practices supportpeer reviewed research in the long term. Much of the current discussion is taking placebetween funders and publishers, including scholarly societies, but not directly withmathematicians. A second goal was to see if we can find a consensus of opinion on what isimportant about journal publishing to mathematicians, that is, where the balance lies betweenthe need for profits from publishing and the desire for broader dissemination of research.The presentations ranged widely; written reports of the talks make up the body of thisdocument. During the first morning John Vaughn, Sam Rankin and Jim Crowley describedthe way the world works in Washington, leading us to think about the future of mathematicsjournals should new legislation be passed to mandate open access 1 of federally sponsoredresearch in the USA. Interleaved with those talks we had a presentation on the work of theIMU from John Ball and a talk from Jean Pierre Bourguignon that placed journals in thebroader context of the research they publish and the work of a mathematician.We heard talks on how mathematics journals work in practice and saw evidence of the growthof journals and the changing behaviour of readers and authors. Information was provided onthe balance between not-for-profit and commercial publishers; the governance of learnedsocieties; who reads mathematics journals; and the value of the older material to currentmathematics research from the citation records. An unscheduled talk by Kristine Fowler, alibrarian from the University of Minnesota gave some very interesting results from a recentsurvey of mathematicians’ views on open access. David Gabai’s talk on the recent history ofthe Annals of Mathematics provided a fascinating insight to the effect of free open access onthe journal’s subscriptions, along with a description of the low cost of publishing the journal.Talks were presented by a variety of major mathematics publishers, ranging from the AMSand Elsevier to Project Euclid. Finally, new publishing models for changing access werepresented from a variety of speakers: mathematicians, publishers and a new university officeof scholarly communication.Here is a summary of what we learned from the meeting.Characteristics that distinguish mathematics journals from other disciplines:- there are lots of journals in the mathematical sciences – 774 listed ‘cover-tocover’ in the Mathematical Reviews database alone;- they are fully international; one cannot distinguish how a journal operatesaccording to which country it comes from; there are no boundaries to submission1‘Open access’ refers to any research paper that is made freely available in published form at no cost tothe reader; it does not distinguish between funded (gold) and unfunded (green) open access.1

-from overseas authors and no boundaries to the choice of country where anauthor may submit a paper;there are no speed pressures; refereeing is expected to be rigorous and detailed.The average time a paper spends between submission and acceptance is manymonths;published articles form the building blocks of future mathematical research. Aproof, once proved, stands for all time and is cited for as long as the literature canbe found, it is therefore important not to lose the building blocks;evidence was shown for the longevity of mathematics papers in terms of bothcontinued reading and citation of the oldest material;the community calls them referees rather than reviewers; journals frequently relyon a single referee to provide a rigorous check of the work, plus opinions fromothers on the relative importance of the work;data sets and other supplemental materials are rare in pure mathematics and thepaper stands on its own – this means there is no easy way to cheat in terms of theresult presented, apart from direct plagiarism;applied mathematics may include data and other supplemental material, but thedata sets are commonly available and it is not a part of the culture to refuse togive background data; applied mathematics is distinct from applications ofmathematics – both are valid but the relevance of the work is judged on differentcriteria.On the arXiv:Mathematicians recognize the value of having free access to pre-refereed material and thepresence of a preprint on the arXiv (http://arxiv.org/) already fulfils most of the requirementslaid out by the green open access lobby. In view of the long referee times, posting a paper onthe arXiv first establishes primacy of the result in the few cases where this is important tomathematicians. Publishers have learned that they cannot put the genies back in the bottlesand that much of ‘their’ content is already freely available. Instead they work to promote thefinal published version as the ‘version of record’ and distinguish that from the arXiv version.Nowadays publishers encourage authors to post the early versions up to and including thefinal accepted version with a piece of acknowledgement ‘to be published in the Journal of X’.However many authors fail to keep the record updated and there are problems withreferencing an arXiv preprint. This keeps the publishers happy that they still have somethingof value in hosting and selling the final published version in return for the costs of editing anddissemination.For some sampled mathematics journals, as many as half the published papers have preprintversions posted on the arXiv and the percentage is growing. This makes the arXiv by far thedominant preprint repository and it is the first place many mathematicians in certain areas ofthe discipline look for new research. It is supported by the many thousands who choose topost their preprints there; no university or publisher forces them to do this. As a result there isvery little enthusiasm in the mathematics community for alternative institutional repositorieswhich are viewed as self-aggrandising university projects. The prior assertion of copyrightownership made by some universities in order to deposit articles in their own repositories hasthe effect of removing the right of the author to decide where they wish their work to bepublished. In contrast, the arXiv is widely and increasingly used; it is fully international andthe barriers to posting an initial preprint are very low.A problem is that there is no long term economic model for paying for the arXiv beyond therecent plea to major universities to support it through donations. We believe that there is anurgent need for the mathematics community to come up with a truly international solutionduring the next few years and it is hoped that researchers from other subject areas, mostnotably the theoretical physicists, are also looking for a solution. The arXiv may need a fully2

capitalized perpetual fund to be set up; the IMU might consider what it can do to facilitatefurther discussion.On the archive:The switch to online versions as the primary source of mathematics journals has led to aninteresting dilemma. Libraries would like to be the permanent repositories of themathematical literature but have already begun to reduce their paper archives while not takingon the direct hosting of the journals they buy. The publishers are now responsible forarchiving and upgrading the online versions in line with demand for more functionality. Thequestion is what happens if the publisher folds? In the past the literature was scattered acrossmany libraries. Nowadays publishers sign up to archiving services like CLOCKSS but thisdoesn’t meet the desire for upgrades, and storing out-of-date formats has little value. This isparticularly important in mathematics where the rendering of mathematical symbols andformulas remains an issue. The recent development of MathJax is likely to help but mayherald another change in format that will require publishers to charge for futuredevelopments. Libraries may need to review their long-term archiving policies.Open access, green and gold: 2Mathematicians do not like the ‘gold’ open access model although Research Councils aroundthe world are considering whether to fund mandated open access. There was generalconsensus that this model discriminates against unfunded authors, including retired authorsand those from developing countries. The question was raised whether mathematicians shouldbecome involved in the judgement of ‘who pays’ for those papers where the author has nofunding. It would be one more burden on mathematicians to identify the deserving needy butif they are not involved the publishers will make their own choices. If the NSF decides tofund a government-mandated open access policy, the money will go to those publishers whohave set up charges for optional open access. For ‘gold’ open access, there is no embargoperiod and once the NSF has paid the fee, the article is immediately freely available online.Evidence from the Annals experiment in ‘green’ open access was stark; libraries cancelled34% of the subscriptions between 2003 and 2008 when the journal was freely availableonline. The Annals is one of the very best journals in mathematics and one of the cheapestjournals; and so it came as a surprise to many at the workshop to hear that some of the bestfunded libraries in the US had decided to save on the subscription rather than support theexperiment in widening access.On embargo periods: We did not hear anyone at the workshop support the principle of‘green’ open access after a short embargo like the NIH model – a 12 month embargo period(i.e. a manuscript must be deposited by an author in a public access repository within 12months of publication). Many mathematicians voluntarily post their preprints in the arXiv andthis could answer the demand, if there is any, for public access. The window between apreprint being freely available on the arXiv, then again being freely available in publishedform just twelve months later is generally held to be too small given the long life of articlesand the slow pace of publication in mathematics. The fear is that libraries will do as they didwith the Annals, and cancel the journal subscriptions and have their readers look at thepreprint version for an extra 12 months. With no subscription income and no ‘gold’ open2‘green’ is free open access where nobody has paid but the article is made freely available;‘gold’ is where someone, nominally the author but usually the research funder, pays to havethe paper made freely available.3

access fees, many journals will not survive. However there was appreciable support formandating green open access after a period that is more appropriate to mathematics, say afterfive years. This was mirrored by proposals from French and German mathematicians formaking the archives of all journals freely available after five years. Should mathematicians beforced to choose a model for publicly funded future research, we think it likely that theywould see five years as the best alternative even if it were at the expense of the closure of thevery few ‘reverse’ moving wall experiments, such as those operated by the LondonMathematical Society.Other matters: Plagiarism, impact factorsThere was strong criticism of the misuse of journal impact factors to evaluate individualpapers but concern was raised that it may not be possible for the IMU to provide any usefulalternative index. Other concerns about the use of such metrics for quantifying journalquality have been well documented.There was also a discussion on the apparent increase in plagiarism and in multiplesubmissions (where an author submits a paper to more than one journal simultaneously),along with the global rise in the number of mathematics papers being written. It was agreedthat there is a need for societies/publishers to maintain standards. Tools such as CrossCheckhave helped combat egregious cases, but these place an additional burden on staff andeditorial boards. The arXiv is used by some Editors when checking complaints and there wasa discussion on whether its use could be extended to provide a more formal registration ofpapers.ConclusionsThe mathematics research community values its own standards of rigorous peer review,which they call refereeing, and the longevity of its journals. They want access to the oldmaterial and the certainty that it be maintained and remain accessible regardless of themedium. Mathematicians are wary of attempts to change scholarly publishing from a nonscientific political world that does not understand the value and nature of the mathematicalliterature.Many people would like to change the funding model for mathematics journals, arguing thatthey wish to provide public access to publicly funded knowledge. The arXiv already providespublic access but it suffers from having no long-term funding mechanism; we believe themost benefit to the community would come from addressing this problem and providing apermanent solution.There is an argument for letting mathematicians decide what they want to support voluntarilyrather than forcing new business models into the market. We should certainly encourage