To celebrate the 50th anniversary of Physical Review Letters last year, a series of nine Essays by physicists and editors has been published. This series concludes with the following Essay, which looks ahead to the future of scientific publishing.

Jack Sandweiss



Jack Sandweiss

The history of the past 50 years of Physical Review Letters should certainly give one pause before undertaking to write on the future of scientific publishing. Nevertheless, in the spirit of those who in the past have not been deterred by such considerations, I offer my thoughts on the subject.

The most difficult problems that face the future scientific publishing enterprise are those arising from the ever-increasing volume of published scientific research. For example, it is currently impossible for anyone to read all of Physical Review Letters or even to casually browse each issue. If all the APS journals are included, as well as the various other publications of high quality, peer-reviewed science, the impossibility is manifest.

The individual scientific reader is then reduced to reading an ever-smaller segment of published results and to relying on meetings, conferences, selective perusal of the arXiv, and the “grapevine.” That leads to an increased narrowing of the scientific background and depth of the typical scientist and of the scientific community as a whole.

Such a narrowing has several unfortunate effects. The individual scientist will be more likely to miss a development in a related field (or even in his or her own field) that could be important to his or her own research. In the past, a physicist knew enough about other physics subfields that it was possible to change from one subfield to another with relative ease, for example, from particle physics to astrophysics, condensed matter physics to quantum information, and accelerator physics to plasma physics. So such transitions will be less well motivated and less likely to occur. The past is replete with major discoveries that arose, at least in part, from such “transplantations.”

The unity of physics and the ability of physicists to broadly understand most of physics has made physicists particularly useful in a variety of social and policy roles. That breadth physicists have had is gradually (or not so gradually) disappearing. Accordingly, individual physicists will be less useful in social and policy roles.

Of course it is natural and desirable that research continues and that its results are peer reviewed and recorded for use by others now and in the future. Research should not be inhibited from its natural evolution. It is the scientific publishing enterprise that must work to deal with the expansion and evolution of scientific information.

There are promising developments on the horizon that are aimed at the problem. The examples I mention below come from physics because whatever expertise I have lies there. Without doubt other areas of science have analogous cases.

One example is the development of virtual journals, which collect links to articles in a particular field from many different journals. These journals, such as the Virtual Journal of Quantum Information and the Virtual Journal of Biological Physics Research, relieve the subscriber from having to search many different journals for articles in a given field.

Another development is the new APS journal Physics, which is intended to be accessible to all physicists. It publishes accounts of important articles in the APS journals as well as links to those articles.

A more speculative idea is the development of a computer program that will be able to “interview” each physicist to find out what articles he or she would have selected if he or she had indeed been able to read all of the scientific literature. The interview would identify the number of articles each physicist would select and his or her personal priorities with respect to subfields. Such an artificial intelligence (AI) program would then read all the literature and select the relevant articles to provide to the subscriber.

The journals would continue the way they are now. The Editors, just as they do now, would work to maintain and improve the journals’ standards. However, the journals would become the database upon which the AI program operates. I omit the interesting discussion of how the business model of the operation of such a program would work.

Clearly such an AI program is highly ambitious. However, physics does have various features that would make such a development easier than an AI program that would mirror an individual’s interests on a broader scale.

Some AI efforts are already under way. For example, Google (more specifically Google Scholar) covers peer-reviewed papers, theses, abstracts, and other scholarly literature from all broad areas of research—a collection of material that might loosely be termed the “Academic Web.” This collection would be part or related to parts of the AI system I have described. Such an AI program would be very helpful to the research part of the narrowness problem but would probably not do much for the unity of physics. To maintain the breadth of physicists, projects like Physics or various evolutions of that journal are needed.

It seems very likely that the future of scientific publishing will involve major innovation in electronic aids to read the literature and in new journals with broad goals that wrestle with the flood of information.