From: "RICHARD S. PALAIS" Newsgroups: sci.math.research Subject: AMS preprint database and server Followup-To: poster Date: Mon, 8 Aug 1994 17:29 EDT Organization: University of Illinois at Urbana NNTP-Posting-Host: symcom.math.uiuc.edu Originator: dan@symcom.math.uiuc.edu There have been several messages recently on sci.math.research discussing the new AMS preprint database and server on emath, and I guess this is a good time for me to supply some background and information concerning it. First the background. About a year ago, in accordance with an AMS Council resolution, a new AMS committee was organized, called the Committee on Publication Policies (or cpub for short). By the way, although I chair this committee---and this message is in some sense a report concerning cpub activities---I would like to make it clear that I am speaking for myself, and I have not circulated the text for approval by the committee. As is becoming increasingly clear, scientific publication is on the brink of enormous changes. There is every indication that there will soon be a rapid transition from paper based journals distributed by mail to electronic based journals distributed and made available over the Internet---and perhaps also archived on CD-ROM. One of the major responsibilities of cpub is to help the AMS react appropriately and responsibly to the challenges of these unprecedented technological changes. As an early experiment in electronic publishing, cpub after long discussion, recommended to the Council at the annual meeting last January that the AMS set up a couple of new electronic services. Here is the (slightly amended) resolution that the Council passed in Cincinnatti: The AMS shall establish two electronic service: 1) An electronic Research Announcements Journal, with articles selected for interest and plausibility but not checked for correctness. The standards should be comparable to the current Bulletin Research Announcements. 2) A preprint database, server, and an interface to other such databases. The preprint database is envisioned as an evolving service. Care should be taken that it is as accessible as possible to the electronically disadvantaged, with the Ginsparg physics database as a possible model. Future developments might include a searchable database of abstracts. Details should be overseen by subcommittees of CPUB in cooperation with CEPS, and the EBC in choosing the editorial board for the RA journal. [CEPS is the AMS Committee on Electronic Products and Services---the committee, chaired by John Franks, that among other things has oversight responsibility for emath. EBC is the AMS Editorial Boards Committee.] Let me come back to the electronic RA journal later, and first discuss the preprint database. A subcommittee of cpub, called cpub_h was formed to oversee the startup of both the preprint database and the RA journal. Its members are: Svetlana Katok (chair), Andy Odlyzko, John Franks, Haynes Miller, Fan Chung, and Bus Jaco. Bus, who is the Executive Director of the AMS, gave the startup of the preprint database a very high priority, and it was ready to go in a very short time (much of the actual software implementation was done by Patrick Ion, of the Math Reviews branch of the AMS staff.) It should be realized though that this is just an initial implementation, and there are many features missing. For example, access to this service is currently via WWW (the World Wide Web) and requires something like Mosaic. This means that it is not available via a terminal interface, and certainly not to those with only Bitnet level access to electronic communication---so this does not meet the Council requirement that "Care should be taken that it is as accessible as possible to the electronically disadvantaged". But eventually, their will be a Lynx server on emath that will provide a terminal interface to WWW, and it is also planned to have some sort of mail list interface to the preprint database. John Lee brought up the question of cross-posting articles to several electronic preprint databases, and I would like to comment on that. First, let me make it clear that the AMS has no plan or desire to pre-empt the field and have emath play the role of a unique math preprint database. (That should be clear from the phrase "and an interface to other such databases.") In fact, during the cpub discussions that lead up to the above resolution there was a substantial minority that felt that emath should _only_ play the role of a clearinghouse for information about preprints in other databases. The existing databases though are mostly devoted to various specialties, or are associated with individual math departments, and we eventually decided that it would be best for the AMS to maintain a general preprint database for those who did not have reason to submit their preprint to a more specialized or local one. I certainly see no reason not to make a preprint physically available on more that one ftp server, but I do think it would be a mistake to "register" a preprint in several different databases. Recall that these databases will eventually become integrated (and that is a primary goal of the emath preprint database) and the difficulties that can arise by having several records in a database point to the same object are well-known. I guess my current advice would be to submit your preprint to a specialized preprint server IF that server has an arrangement that makes its database known to emath, and otherwise submit to emath. Finally, regarding the new electronic RA journal, it is scheduled to go into operation in early 1995, at the same time that the Bulletin will stop publishing Research Announcements. Part of the haste in getting the AMS preprint database in operation stems from the fact that the preprint server and electronic RA journal will have the same underlying software base, and cpub_h wanted to have that software working reliably before trying to start up the RA journal. I am copying below the report that cpub_h has put on the agenda for the Council meeting at the MathFest in Minneapolis next week. It concerns the details of the electronic RA journal. Please understand though that at this time it is only a proposal to be discussed. Richard Palais Chair of CPUB ===== REPORT OF THE SUBCOMMITTE ON THE IMPLEMENTATION OF THE ELECTRONIC RA JOURNAL AND THE ELECTRONIC DATABASE. The AMS Electronic Preprint Server has been created and will be practically implemented as soon as the Instructions to Authors, which already exist, are sufficiently simplified as to suit an average, in terms of computer litaracy mathematician. It is a HTTP (WWW) server (i.e. easily accessible from Mosaic) that accepts both abstracts and full preprints. The same or very similar software can and should be used for the Electronic RA Journal. It was, obviously, used for the new Electronic Journal of Combinatorics, which, in our veiw, is presently the most convinient for a lay user of all existing electronic journals. The main charge of our subcommittee is therefore not in the technical but rather in the "intellectual" implementation of the RA Journal in its new electronic form. What kind of publications are RA's? A description which presently appears on the cover of the Bulletin of AMS says: "A Research Announcement should be designed to communicate its content to a broad mathematical audience and should meet high standards for clarity as well as mathematical content", and we see no problems with this description. A RA is a report on an important new result. The author should have a proof, although the area may not be yet completely developed, all corollaries obtained etc. Important reasons to publish RA: to establish priority, to make a claim and continue to devolop the topic. It is only natural to publish a long version of the paper with complete proofs elsewhere. During our extensive discussions before the Cincinnati Meeting Dick Palais said: "I do not mean to suggest that electronic RA's should be any less severely reviewed than the current RA's. I feel that only a small fraction of new mathematics needs or deserves the special status that is conferred by being chosen as an RA before normal publication." John Garrett adds: "RA's materials should be "distinguished" for this purpose either by the importance of the results or (perhaps) by the need for the feedback at this particular stage in their evolution... Speed is also very important: another topic for ongoing monitoring. I also think that in an electronic environment, where plagiarism and idea-theft become, unfortunately, easier than ever, the case for establishing priority will become more urgent. This argues for both the need for the RA's, and the need to ensure that they are providing the service for the right authors with the right ideas at the right time." There are two questions we have to answer. 1) What do we want to achieve with reborn RA's? 2) How would we achieve this? We start with the hypothesis well formulated in Odlyzko's article that electronicjournals will have the same purposes and provide the same services as existing paper scholarly journals, i.e. - provide reliable results; - alert the community to what the most important recent results are; - help in making promotion, tenure and grant decisions. Here are the answers we offer to the Council: 1) The electronic RA journal is a prestigious AMS journal publishing announcements of important mathematical results, a publication in which is taken siriously for evaluation of the authorUs performance for promotion, tenure and grant procedures. In other words, the journal must have a flow of good papers. 2) In order to achieve this it is necessary - to have a representative editorial board; - to minimize the time between a submission and publication; - organize an agressive advertizing compain. Here are our suggestions on the organizational procedure. 1. THE EDITORIAL BOARD. The Editorial Board will consist of approximately 25 people, representing all main areas of mathematics, one of them is a Managing Editor. In order to establish high standards and high visibility, at least the first Editorial board should primarily include the leading active mathematicians in the respective fields. 2. THE PROCEDURE. Each member of the Editirial Board should have an agreed upon sufficiently broad area of expertise (such as algebraic topology, applied analysis, mathematical phisycs, logic etc.) which may (like in CMP) or may not (like in MRL) appear in an official advertisement. 2.1. A paper can be submitted to any Editor. If it belongs to Editor's area of expertise, (s)he may make a recommendation based on either h(er)is own or an outside review. (S)he has a right to reject the paper. 2.1.1. If a recommendation is positive , it is communicated to the whole EB. An abstract and a review are posted (electronically) for a fixed period of time (1 or 2 weeks) durung which any Editor has a right to request the text of an RA and make comments and suggestions. If no consesus is reached, a vote should be taken, and a fixed high proportion (75\% or 80\%) of the Editors should approve. 2.1.2. If a recommendation is negative, the author is advised to withdraw the RA. In case of disagreement the Managing Editor sends the paper to an outside review, then this review is posted with the original negative review, and the previous procedure follows (see 2.1.1). 2.2. If the Editor to whom the paper is send is not considered an expert in the field, (s)he should forward it to the Managing Editor or to the expert editor. 2.3. Papers may also be submitted to the Managing Editor who forwards them to expert editors. Expert editors should be encouraged to consult each other and make joint recommendations, especially in cases when the area of a RA can't be clearly delineated. 3. THE GUIDELINES FOR QUALITY CONTROL. We realize that to review a RA is harder than a regular paper. Although the Editors should not routinely request an additional text with complete proofs, the existence of such a text, even in an imperfect form, could be used in the decision making process. For example, an Editor handling the paper may ask for a preprint and guarantee, if necessary, that he will only use it for a purpose of review and not disseminate it. Other possible sourses of information are seminar talks, opinions of individuals who undersatand the proof etc. The existing journals Comptes Rendues, Doklady, IMRN in Duke Math. Journal and MRL are working along those lines. Traditionally, in the paper journals the quality control is closely related with restriction of volume. Since in the electronic media there is no natural restriction on volume, the quality control of an electronic journal should not be linked with any specific restriction on the volume. Instead, the Editorial Baord should collectively strive to establish reasonable absolute criteria based on significance, originality and technical difficulty of the result. =========