Speaker: Sam M. Austin (Michigan State University)

Panel Discussion: Society and Physics-Wide Strategy

The purpose of this paper is to encourage us to take the long view as we consider the means of publication in the future, as we move beyond the availability of paper journals in a library to the opportunities that could be provided by a unified electronic information resource, a knowledge server. In this talk I won't be bound by considerations of technical feasibility; I assume that the technology will be available to do the things we might think of today. The greater danger is that we will be too conservative; that financial considerations and the fear that our institutional journals will be driven from the field by the electronic bulletin boards will induce us to accept fragmentary solutions and to think too small. We need to determine what we need, assuming that the technology, and with good luck, the institutions, will be there to provide the needed service. I shall argue that what we need is beyond that provided, or likely to be provided by the electronic bulletin boards.

Let's begin by assessing where we are today. We have a set of journals that on the whole serve the physics community well. Some areas of physics have a well developed system of reprints to provide early notice of research; in some areas of physics these preprints are provided electronically by the Los Alamos bulletin boards or by repositories such as those at the Stanford Linear Accelerator Center and at CERN. These systems are remarkably inexpensive. For example in the field I know best, nuclear physics, at a rough estimate Physical Review C publishes a sufficient number of papers to describe all the nuclear physics done in the US. The cost of all library subscriptions to Physical Review C was $1,100,000 in 1993 and the total federal support of nuclear physics research was about $375,000,000. I.e., the cost of APS publication of US nuclear physics would be 1/3% of the cost of doing the research. Although I have not attempted to estimate them, it seems unlikely that the costs for other fields are higher by a large factor. The cost of the various bulletin boards is more difficult to estimate on the same basis, but is small.

It seems then, difficult to believe that the cost of journal subscriptions is a truly major problem. Rather the problem is an outmoded approach to the financing of research publication. The agencies supporting science, the publishers, and the libraries could surely find a way to support publication in a way that reduced the burden on the libraries and that less distorted the relative roles of journals that have page charges and those that do not. They could surely find a way to remove the burden that substantial page charges would place on the individual investigator. It should be one of APS's goals, as a major physics publisher, to work toward such a solution.

I conclude that the difficulties faced by the publishers and the libraries, while real, should not be allowed to dominate our choice of how publishing evolves in the electronic age. It is much more important to take the opportunity that electronic publication will offer to improve greatly the way we communicate our results to our peers and the way we archive them for reference by future generations of physicists.

I propose that our goal should be a knowledge server for physics. It should, ideally, contain (or provide access to) all information related to physics research and professional society activities that a physicist wishes to access. This might include: new scientific papers and an archive of previous papers; supplementary information that makes these papers more useful, such as data tables and computer programs; current contents listings; announcement services; electronic preprints; meeting abstracts; announcements of meetings and other events; and job openings. This data base should be searchable. Finally, so as to provide continuity and certainty over a time frame of decades, this server should be institutionalized--it should be operated, or at least coordinated, by an institution that has as a major responsibility the provision of this service.

How then do we reach this goal? First, we must decide whether we evolve the journals or we evolve the preprint servers? Do the present publishers have the foresight and energy to carry out this evolution on an appropriate time scale? Are those who are operating and developing the preprint servers willing and able to expand and unify their service, provide peer review, provide a uniform information source, and obtain commitments from their institutions for long term archiving and other support functions?

My own view is that it the present publishers, or some of them, are more appropriate for developing such a knowledge server. Their institutions are in place, and many are pushing strongly in the direction of publishing their journals electronically. The electronic bulletin boards will probably also play an expanding role, but perhaps one more appropriate to their lean, creative, and idiosyncratic structure. For example, they might become both more universal and integrated, and more closely connected to the archival electronic journals. They might assume a larger role in publishing conference proceedings--in many cases the information in these proceedings is relatively ephemeral, playing an alerting role, and is published in final form in the refereed journals. Hence the requirement of long term institutional support is less important. If refereeing approaches can be worked out, they might play a role in screening papers for publication in certain archival journals.

It will be difficult to achieve the goal of a knowledge server for physics, and it seems clear that the issues are not technical ones. For example, how does the physics community induce all significant publishers to make their manuscripts available to the same server?. There are only glimmerings of how that might happen, but it seems clear that APS is an appropriate institution to lead the way.