Speaker: Bert TePaske King (Mathematical Reviews)

Panel Discussion: Tracking of Papers

Let me first say I haven't been involved in any of the policy decisions about the e-print server. I want to preface this talk by saying that I am representing a rather unique secondary bibliographic review journal that serves the mathematical community, which has very different feelings about refereeing, and about preprints, either paper or electronic, than is the case in the physics community. Having said that, a little bit about Math Reviews for those of you who may never have seen the big orange monster. We add roughly 60,000 new items to our data base each year, which appear in our current awareness journal, Current Mathematical Publications, from around 1,600 journal titles, and around 2,000 monographs and monographic collections. Of these, roughly 48,000 reviews a year are published in Math Reviews. Most are evaluative reviews by professionals, and the discrepancy in these two numbers (60,000 versus 48,000) is often the result of items that are not in final form. That is, they are lacking proofs, or the proofs are sketchy; they are extended abstracts or preliminary versions, etc. The charge for Mathematical Reviews from the mathematical community, researchers and librarians, is to cover the permanent literature of mathematics. That is, to this point, traditionally published monographs, proceedings, refereed journals. Math Reviews editorial policy with regard to preprints is longstanding. Preprints and technical reports are not ordinarily considered part of the permanent literature of mathematics. There are a couple of exceptions. There are items, serials entitled "preprint", mostly from Eastern Europe, especially from the former Soviet Union, that traditionally do not appear published elsewhere and for which Math Reviews may be the only U.S. source. There are also preprint type serials whose publishers have assured Mathematical Reviews that papers which appear in them are in final form and authors are discouraged from resubmitting those papers elsewhere. Given that, the format of a preprint is less of concern to Mathematical Reviews that the state of the content. Paper versus electronic is not a distinction. A preprint is a preprint, whether it is paper or whether it is electronic.

I want to bring up something else that we're currently doing, what we call AMS prepublications. These are papers that are accepted for publication in American Math Society journals. They are sent to us as paper photocopies of the manuscripts. These prepublications are processed and sent for review well in advance of their formal publication. We get the reviews back from reviewers, they are keyboarded, processed. However, the reviews do not appear in Math Reviews until the item formally appears in the AMS journal for which it has been accepted. Are these prepublications essentially different from E prints? In many ways they are not. However, there is a critical difference. These papers come to us already refereed by the editors of the journals. This raises the question of timing, with the same concern arising obviously for E prints that may eventually be published in refereed journals: electronically distributed preprints may be old news to many in the community by the time they finally appear. However this seems to be less a problem in our field since a paper in mathematics has a half life something on the order of ten years, much greater than is the case, as I understand it, in physics.

Next I want to discuss peer refereed electronic journals, which may be part of the answer to the timeliness question. They provide for relatively quick appearance of papers, relative ease of access, relatively lower cost, or no cost in some cases, to subscribers, and an acceptable level of peer review. We are currently indexing and reviewing seven electronic journals which, by the way, have seven completely different combinations of files, formats, distribution, announcement, etc., etc. It is not a trivial thing to obtain these papers, process them, print out a copy for the reviewer, if that's necessary, archive them, do whatever troubleshooting we have to do. This is obviously the case with many electronic E print servers as well. However, again, the electronic format of these E journals is not important; that's a chimaera for Math Reviews. Since these really are peer reviewed journals in the traditional sense, it doesn't matter whether they are electronic or paper; they still fit the editorial requirements for indexing and review in Math Reviews.

Will MR be involved in tracking electronic papers from E print to archive? For now, the best I have to say is not yet. We continue to believe that there is great value added in identifying, archiving, and reviewing the final, permanent form of a paper in the mathematical literature, and we will continue to expend our efforts and resources to that end. We feel that MR's value to and reputation within the mathematical community rests on its comprehensiveness and its reliability.

What might cause us to change this policy? Now I'm talking for myself and what I say does not represent any policy of Math Reviews or the AMS. The first and most important thing that could cause us to change our point of view, our editorial policy, is a sea change in the mindset of the mathematical community towards acceptance of E prints as part of the permanent literature of mathematics. This is the single most important thing that could cause us to change. I think it rather unlikely. Another thing that could cause us to change our current policies is the influence of other fields covered by Mathematical Reviews, theoretical physics and theoretical computer science being two particular examples. If these fields go to a culture in which E prints are part of their permanent literature, might it not force us to review things that we would not normally review? Perhaps, although I think it just as likely that Math Reviews would say okay, it's not in final form, it won't be reviewed. Finally, what of final form E prints: those papers that are not traditionally published, either out of rebellion against traditional peer review (and I'm sure there's nobody here who represents that point of view) or out of insulation from it. The person who says I already have tenure, I don't need to submit this to a journal, it's up on the E print server, it's fine. Again, MR has no plans to index or archive such e-prints. Could we list them in our current awareness journal? Sure, we could do that. But why if they are searchable in an E print data base? Should we provide links to them if they are cited? Certainly we could, just like we hope eventually to provide hypertext links between reviews and papers, and between citations and reviews, and reviews of those citations. But will this sort of thing influence MR editorial policy? I suspect that it won't. Perhaps new modes of progression from E print to journal may arise to help deal with some of these problems. And I mean journal not necessarily in the traditional sense. Maybe papers will be selected by established journals directly from E print servers and tagged as such by the journals, not by the authors, an imprimatur that we don't currently see. Perhaps MR might then pull and track such E prints for early review processing. But, again, I don't think a review would appear until those papers appeared in final form in the journal. Perhaps new journals of this type might even be administered by professional societies, and I think this is a topic that Michael Peskin will address in his talk later on. Thank you very much.