The following Report was presented by Benjamin Bederson, Editor-in-Chief, The American Physical Society, to the APS Council Meeting 6 November 1994.

REPORT TO COUNCIL ON E-PRINT ARCHIVE WORKSHOP, LOS ALAMOS, OCT. 14-15, 1994
B. Bederson

APS organized and sponsored a meeting at Los Alamos on Oct. 14-15, 1994 intended mainly to explore the rapidly burgeoning use of the Internet for the presentation of early versions of articles, prior to submission to archival journals, and prior to refereeing. APS considers this phenomenon to be directly relevant to its primary mission of advancing and diffusing the knowledge of physics, and at the same time it is naturally interested, and even concerned, over the impact this new mode of dissemination will have on its own research journals. Accordingly we took the initiative in bringing together a variety of individuals representing the main groups involved in the generation, distribution and use of these so-called "eprint archives". These included developers of eprint archives, librarians, editors, APS and AIP representatives, representatives from several non-US physics societies, representatives from several non-physics societies that have been actively engaged in similar projects, as well as a representative from a government granting agency and a legal expert on copyright and intellectual property issues.

The organizing committee consisted of the following:

B. Bederson (APS - Chair)
• J. Clem (Iowa State U.)
• P. Ginsparg (LANL)
• P. Kreitz (SLAC library)
• R. A. Kelly (APS)

The workshop was organized entirely on the Internet; I don't think that a single paper communication reached anyone. About 80 people attended. It started with a dinner and a keynote address by Paul Ginsparg on Oct. 14, went through the full day on Oct. 15, and ended with a dinner that evening. Apart from the fact that the weather was bad and the meeting room was completely without heat (Harry Lustig wore a scarf and had to borrow a parka from Michael Keller to keep marginally warm), I believe that the meeting was quite successful, and accomplished its primary purpose, which was to explore the present and future ramifications of the eprint phenomenon, and to help us learn how to respond to it, in terms of both preserving our present very important (and successful) journal operation, and in moving to meet the challenges we are facing. No plan of action evolved, as how could it? In fact, from my own viewpoint the principal outcome of the meeting was to define not a course of action but a series of questions, which I will delineate later, and some feeling for how the answers to these questions will evolve with time.

Bob Kelly and his staff at Journal Information Systems organized a section on APS' WWW server which presented contributed papers, general discussion, and updated practical and program information to attendees. I reproduce some of this material which is of more than passing interest, including the original meeting announcement, in Attachment 1. The meeting was taped in its entirety; disposition of this record has yet to be determined.

The format of the meeting consisted of a number of panels, with several discussants presenting short talks, generously interwoven with audience comments. Each panel had a discussion leader, who was primarily responsible for the selection of the panel members. The Program is included as Attachment 2. In what follows I give some personal impressions, not intended to be definitive or to represent "minutes". I will present only my ideas of the highlights; not all talks will be mentioned.

In Paul Ginsparg's keynote address he outlined his views of the future of "eprint archives"--that is, of the posting of papers on line, available to all, basically free, as soon as they are submitted. Eventually some sort of selection, other than the obvious elimination of nutty papers, will be included, as well as all sorts of enhancements previously unimagined, such as past and future linkages to other articles, ad hoc comments by readers, etc. Already there is an enormous volume of papers in high energy physics, primarily theoretical, that appears on the LANL hep database, ever growing. The printed journals as they currently exist have a dismal future, Ginsparg believes. He hopes that APS has sufficient foresight and vision to acknowledge this situation and to make an effort to adjust its mission of "advancing and diffusing" the knowledge of physics, taking into account the electronic revolution that is taking place before our eyes.

Harry Lustig commented that in fact our journals are growing at a rate greater than 8% per year, and in Physical Review D the growth in theoretical high energy physics papers (the principal subject of eprint archiving) was far greater than 8% per year--a phenomenon that does not seem to suffer from the eprint challenge. How does Ginsparg account for this? His answer, not entirely satisfactory in my opinion, was that submission to journals was essentially free, the additional effort to submit to journals being insignificantly greater than that required to post on the Internet. Probably the reasons for our growth are more complex than this.

In introducing the first panel, I again raised the question of growth. Despite the looming clouds on the horizon, a (maybe the) principal problem of our journals relates to their success, as represented by the unprecedented growth in submissions. APS and its members are and should be proud of the preeminence that its journals have achieved in the scientific community. Growth causes considerable concern because of its inevitable impact on library budgets, the sheer volume and mass of our journals, and the difficulty of the individual scientist in coping with the flood of information. These are, in fact, among the principal reasons for our interest in moving towards electronic publishing, of which eprint is one aspect. Growth is of such concern to us that a special Task Force on Journal Growth has been established, with Eugen Merzbacher as Chair, to study the problem in an attempt to understand it, and to recommend means for controlling it. Maurice Rice, one of our panelists, is a member of this Task Force.

At the beginning of the first working session, CURRENT THINKING, Bob Kelly made a statement that everyone could agree on, "one size fits all" cannot apply to our future endeavors. Needs of individual subfields can be quite disparate. Addressing such needs could require quite disparate approaches.

Pat Kreitz, from SLAC, discussed and demonstrated the SPIRES hep database. This database currently contains bibliographic summaries of more than 280,000 (!) particle physics papers, including preprints, journal articles, technical reports, theses, etc, some fully formatted. Most of the SPIRES preprints come from Ginsparg's LANL database. Kreitz was abetted by Annette Holtkamp of DESY, which cooperates with SLAC on the hep data base. An important question was asked of Kreitz and Holtkamp: why is hep theory so dominant in eprint? They opined that this is partly attributable to the fact that the hep community has many decades of tradition in the distribution of paper preprints behind it, so that going online seems like a natural continuation of this long tradition. It will take a while before other physics communities acquire the habit.

Bob Hanisch discussed a similar workshop held by the American Astronomical Society several weeks ago. In fact, he said that had our two societies known of each other's plans we might have held one, rather than two, meetings. AAS is very well connected electronically. All authors submit abstracts electronically, for example. Astrophysics Letters is going online soon (funded by NSF). AAS considers peer review to be crucial--no matter what the future holds, peer review will survive. He expects that the time interval between submission and publication will be no more than 10 days, not including the peer review process.

R. Youngen, of the American Mathematical Society, described the debate going on within AMS right now: should electronic preprints be encouraged or discouraged? The balance appears to be emerging on the positive side, and accordingly AMS is planning to set up a preprint server. Their copyright policy seems to be that it remains with the author. They will withdraw articles from the server after publication, with a pointer remaining to indicated same. Their biggest problem, not surprisingly, in the tracking of preprint to published paper, is how to maintain journal revenue.

W. Taylor, MIT, a user, is concerned that online access to unvetted articles might prematurely endanger printed journals. He took a cautionary tone, basically urging care in moving into the electronic future. He did offer an image of the future journal fully integrated into a hypertext network. He presented two scenarios--pessimistic and optimistic. For the former, journals would be weakened, diluted by volunteer-run online eprints, for the latter, journals would acquire a far more flexible format than they currently possess; acceptance of articles for publication will be greatly accelerated. There will inevitably be a redefining of the research journal's mission, to concentrate mostly on good editing and providing maximum information.

A. Cohen, Boston U, spoke about the journal of the future--a complete database, organized with sufficient flexibility to satisfy the changing needs of the physics community and of the physicists themselves. He felt that there should be some sort of ranking system for eprints.

Ginsparg gave some interesting demos of the LANL eprint archive.

In supplemental notes after the meeting, Kelly made the following statement:

``My goal was to set the stage for the rest of the day ... APS will take from the meeting ideas like:

1. The landscape of scientific publishing is changing rapidly.
2. We can use technology to not only accelerate the process but to improve it and add value.
3. A 'one size fits all' paper based policy is obsolete. We can craft a process that fits the different needs of many users.
4. Author tools and education are necessary.
5. There is a place for eprints in the current and probably future APS publishing policy.
6. The definition of future processes should be explored through small experiments.
7. We should establish a collaborative effort with universities, libraries, and other societies; representatives from all of these were present at the workshop.''

Maria Lebron, APS Associate Publisher, chaired the panel on INTELLECTUAL PROPERTIES ISSUES. This is an area where there are more questions than there are answers.

Paul Berman, of the law firm Covington and Burling, is a copyright and patent specialist. Covington and Burling is APS's principal legal advisor. Berman gave a very thoughtful and comprehensive review of these issues, as they relate to eprint archives. His views as to what constitute publication were fairly strict; he is very confident that posting something on an electronic bulletin board for free or on any readily accessible basis most definitely constitutes publication, legally speaking. This is in principle different than a paper preprint distribution which usually is limited to a relatively small (not more than a few hundred) circulation. Electronic posting is without control, so that the potential audience for an eprint can be as large as you want. The courts would surely interpret this as publication (although this has mostly not yet been tested in the courts). Operators of eprint archives are publishers! It is very important to know who owns the copyright--we are in a new era with no definite answers. The situation is very confusing.

B. Hagen, IEEE, stated categorically that it is counterproductive to destroy what has gone before. It is important to maintain continuity with the past. There will inevitably be a transition to electronic journals, but this transition needs to occur peacefully and smoothly. Peer review does not have to be reinvented--we already have it. He is strongly in favor of having the copyright retained by the publisher. IEEE has a rigid copyright policy.

Ann Okerson, Association of Research Libraries, urges societies to seek partnership with the research libraries. ARL is developing an eprint strategy--original works are assumed to be protected, presumably in the name of the author (and perhaps the author's institution). At present eprints do not constitute a major problem, although this of course may change. No journals have been cancelled because of eprints, to her knowledge. It is very likely that we will see a change in the basic concept of copyright very soon.

Some lively discussion between panelists and the audience ensued; it can be fairly stated that there was not complete agreement among participants concerning copyright ownership and in the unrestricted dissemination of scientific information. Conference proceedings, most agreed, could readily be put directly online thereby bypassing the tedious and generally very expensive production of printed versions. Scientists want to see their work read!

Berman reiterated his belief that journals represent the most rational medium for the control of copyright. Dispersing this would not be in the best interest of the scientist, and could lead to great confusion in the scientific and publishing communities.

Peggy Judd, AIP, ran the panel on TRACKING OF PAPERS--how should a creative work migrate from its first posting on the networks to its final authenticated version in an archival, trusted journal?

R. Austin, Princeton, discussed eprint and general electronic access from the viewpoint of a "small science" user. He was looking for added value in eprint. He opined that it is crucial that experimentalists trust posted results; unfiltered postings could be dangerous. Preprints when posted should not be archived, and should be removed upon publication. He would love to see good and understandable review articles generously peppered with hypertext, and fantasized about a "living book" that improves with age as it is refined by users. He also made a passionate appeal to the ultratechnically endowed literati: make things simple and workable for the individual physicist whose time is so limited that any surplus should properly be spent with graduate students or taking data, not trying to decipher Unix.

John Light, U of Chicago, is the Editor of Journal of Chemical Physics, a highly regarded and successful AIP journal. He has instituted the so-called JCP Express, which offers an online service, without charge to those already subscribing to JCP, presenting papers accepted for publication, i.e., having undergone peer review, but have not yet been published. The posting is voluntary, and authors must submit their papers in suitable latexed form. Results are thus far disappointing; only perhaps 5% of JCP papers appear on JCP Express.

In a remark from the floor, Peskin was concerned about changing versions of the same paper. Journals need to be very careful about labeling versions; it is the responsibility of the author to keep the paper up to date.

I chaired the panel on PEER REVIEW. I started the session by addressing a "peer review" issue mentioned by Ginsparg in his keynote talk. People are always asking about peer review for eprints. Surely uncontrolled postings can sometimes cause real problems. Ginsparg described how he himself addresses this concern for his own papers: he is careful to distribute them to knowledgable colleagues in advance of posting. This is, in effect, peer review (though of course non-anonymous and less objective than journal peer review). I noted that while not immediately obvious, this is related to an interesting phenomenon in our journals, and in other well-regarded non-letter archival journals. Why are our rejection rates so low? Apart from PRD15 (which includes general relativity and cosmology, and for which the rejection rate is about 40%) our journals have rejection rates in the neighborhood of 20%. This is attributable in large measure, I believe, to the quantitative and disciplined nature of our field. Most respectable practitioners are self-critical, and are pretty careful about what they submit to our journals. This has been, and will continue to be, an important factor in maintaining the quality of physics articles, eprint or otherwise, even without formal peer review.

The session started with presentations by two senior APS editors, Jack Sandweiss (PRL) and Lowell Brown (PRD). Sandweiss first briefly described current review procedures, and offered a spirited endorsement of peer review. Imperfect as it is, it is at the root of our publication policy. Peer review serves several purposes: it certifies reliability, it encourages careful presentation, it improves papers, it serves to promote careers (however, see below). Brown followed with a presentation of statistics from PRD15, intended to illustrate the added value that peer review gives to the published article. Coincidentally, Peskin then volunteered that he had prepared similar data attempting to compare articles which had appeared as preprints with the same articles that ended up in PRD1. He demonstrated that in his sample study there appeared to be little difference in the quality of the majority of preprints as compared to the final, refereed versions. These talks were followed by two from non-editors, Geoffrey West and Maurice Rice. These were not as supportive of peer review as were those of the editors. West questioned whether the perceived advantages of peer review, as noted above, really applied, apart perhaps from the "career" factor. West also made some other interesting if provocative comments: "no one is reading anything any more", computer literacy on the near side of the generation gap is fantastic, with the converse also true; it's too bad that we haven't heard much from fields outside hep. We also seem to be ignoring industrial science, although it is true that networking is not as active in that segment. West believes that some sort of grading is necessary. Eprint will continue to drive specialization even further, narrowing people's expertise even beyond present narrowness. PRL articles are difficult if not impossible to understand, he states. He believes that articles in PRL are too short (it is important to observe that in eprint, length is not the tyrannical constraint that it is with paper. We don't know how this will play in the future, but I'm guessing that articles will become longer, rather than shorter.)

Despite his generally downbeat tone, he ended with a true laurel for Physical Review Letters. In the entire world of physics, PRL is the only journal that is truly ecumenical. It covers all fields of physics. It is still possible to leaf through an issue of PRL and see (if not comprehend) articles in virtually every field. This is a precious asset that should be preserved at all cost.

Rice claims to be a peer review agnostic, taking a neutral position on this issue. As a HTC physicist, real validation comes from repeating experimental results. Peer review of course cannot validate experiment. He does not think that peer review improves papers significantly. He thinks that public forums, the number of citations, the number of accesses, may be a better way of evaluating papers than peer review.

The session on SOCIETY AND PHYSICS WIDE STRATEGY, chaired by M. Turner, considered general issues; what should APS's role be in nurturing eprint, what should its role be in fostering cooperation among physics societies outside the US. Peskin offered his vision of a journal of the future, or rather, of a strategy for publication, since eprint, archival journals, and eprint archives, all would play important roles.

Turner believes strongly in the survivability of the traditional journal. The role of eprint initiatives is to experiment, from which journals can learn. Sam Austin advocated the need for commentaries for eprint articles. What is most needed is a unified source, everything of a given field available in one place.

Peskin stated that the principal value added by journals is refereeing. He noted the information explosion (I almost said deplored, but how can you deplore the inevitable?).

Alan Singleton, of the Institute of Physics Publishing (UK), urged cooperative efforts not solely related to the generation of uniform standards. Eberhard Hilf, (Oldenberg), represented the German Physical Society (DPG) and made a similar appeal.

In the last talk Harry Lustig, taking off his scarf and parka, made an attempt to bring some of the more visionary participants of the workshop down to Earth, by presenting a discussion of the operation of APS, of the many functions it performs in the service of physics, physicists, and science in general, and how our journals represent the financial foundation and source of funds for most of its activities, in addition to their scientific value. It would be folly to surrender our journals' financial strengths without offering equivalent replacement. However, he believes that the growth of eprint archives will affect our refereed journals and that a link could and should be established to maximize the viability and effectiveness of both.

I left the meeting feeling, as did many of the attendees, that we had heard a lot, and that what we had heard needed digesting. I did not feel that my generally optimistically cautious views about eprint and journals had been changed much. I thought there was surprisingly little bashing of the overall performance of our journals, while at the same time there was considerable urging for APS to think boldly about the future, and not assume that its current position as the overall leader of physics publishing in the world will be secure forever. There was virtually universal agreement that peer review, which lies at the heart of the modern scientific research journal, must and will survive in the electronic future, although its form may change.

To conclude the workshop (which went on so long that we had to forgo a general discussion session and a formal "wrap up"), I stated that I would not present a series of conclusions, but rather I would offer a series of questions, raised in one form or another during the workshop, whose answers would become apparent, perhaps, with time. [Items in brackets indicate my attempts to envisage some answers]

• Does the traditional peer review journal, paper or electronic, have a future?
  [yes, at least in the "short" term]
• Will the eprint revolution spread throughout physics?
  [definitely]
• Will peer review survive in some form?
  [yes]
• Will copyrights and general intellectual property rights survive in their present form?
  [I don't know, but it seems doubtful]
• Will technology continue to explode at its current rate for a while?
  [for sure]
• Will hyperlinks connect the full physics literature?
  [depends upon time scale. Yes, eventually]
• Will libraries survive in their present form?
  [probably not, but they will surely survive]
• Will library, society and other providers of scientific information join in new forms of consortia?
  [probably]
• And specifically, what should APS do now to help position itself for the electronic future, apart from its commitment to online publishing?
  [to be continued!]

I would like to thank the many individuals who contributed so much to the workshop's success. First, I thank Kathleen DeGregorio, of APS Journal Information Systems, for her dedicated and skilled efforts in organizing the logistics, and Jessica Binder, Administrator of the high energy theory group at Los Alamos, for her parallel efforts on site. I thank the organizing committee for all its help in developing the program, and especially Paul Ginsparg for his willingness to cooperate with APS, despite his expressed reservations regarding APS vision. I thank Bob Kelly, Director of APS JIS, for his indispensable and enthusiastic efforts to bring off the meeting. Finally, I thank APS itself, which contributed generously via the Treasurer and Executive Officer, for its financial and moral support.