The modern process of scholarly communication relies intensively on refereed scholarly journals. It is worth recalling Steven Harnad's observation about the difference between scholarly writing and commercial writing. In commercial writing, the interests of the author and publisher are coincident; both wish numerous purchases of the published work, and both wish to protect against unauthorized copying of the work. In scholarly publication, the interests of the author and the publisher diverge, since the publisher retains similar restrictive interests, whereas the author would normally seek the widest possible distribution of the written material.

Most scholarly communication takes place in one of a few forums: books and other monographs, conference proceedings, and most commonly, articles in refereed journals. The process of journal publication serves multiple functions for the academic community. All observers can see the functions of distribution of information and its eventual (dispersed) archiving in libraries. These are essential functions for scholarly communication, but they are far from the only valuable functions. Archival activities have no value unless there is also a functional indexing and retrieval system. Journal and library indexing (and storage) systems provide this important function. Another essential function of this system is embedded in the word "refereed" -- the process of certifying quality. This process serves readers by selecting out the subset of all written documents in the field those deemed most likely to interest the reader, or to add to the relevant body of knowledge of a discipline. This process in turn helps to solidify the current meaning of what constitutes a specific academic discipline -- defining the paradigm, to use Thomas Kuhn's phrase. Identifying articles that further this purpose also serves another essential purpose for the academic community, by providing a ready system to rank the quality of scholars. Those scholars who most often write articles deemed "important" (i.e., publishable in the journals held in highest esteem in the field) receive personal recognition that translates into promotion and tenure in our universities and colleges, grant awards from governments and foundations, and success in more commercial endeavors such as serving as an expert witness in legal forums. The quality certification system becomes more, not less important as the costs of production and dissemination of knowledge fall, since the amount of material that a scholar must consider increases as those production and dissemination costs fall. The essential issue is the economics of attention, not the economics of distribution of information.

The process of expanding a field's size, e.g., in terms of numbers of participants, breadth, etc., leads systematically to an increasing specialization. Famed economist Adam Smith perceived this phenomenon clearly over 200 years ago when he declared that "Division of labor is limited by he extent of the market." In modern scholastic terms, what this means is that as a field grows larger, it tends to break into subdisciplines, each of which will seek an appropriate forum for communicating interests among those who share this particular sub-disciplinary interest. As the sub-discipline becomes more specialized, it becomes of more intense interest to a small set of people and of less interest to the wider audience of the broader discipline. The incentives for creating a sub-discipline paradigm increase, and as the costs of publication and travel fall, it becomes more likely that new learned societies will spring up to meet the intellectual demands of this new scholarly group. They will meet, form a journal, choose and editor, publish, and meet annually to discuss their mutual interests. They define, if you will, a new paradigm of learning that will, if sufficiently large, create its own intellectual standards, quality control, and self-defined group of leaders. If they formalize this association, we will see the emergence of a new learned society, or at least (as a minimum) a new journal.

II. The Financial and Intellectual Dilemma

When this process matures, the faculty in our universities and colleges who produce these works will naturally seek to have the journal included in the subscription set for the libraries where they work. Since those faculty will also wish to continue access (and continuous subscriptions) to the wider-area and hence more popular journals in their major discipline, the annual acquisitions costs of libraries will confront continual upward pressure. These pressures are in turn magnified by per-item cost increases that have, in past years, vastly exceeded background inflation levels, particularly (and infamously) in commercial journals published in The Netherlands. Naturally, college and university administrators seek to find some way to control these cost increases, since they directly compete with other desirable uses of resources. Thus, one question of interest is whether there exist any alternatives to the current system of journal publication that will provide relief from these increasing costs, both in terms of the current level, and perhaps more importantly, to curtail the rate of cost growth itself.

It is worth recalling again here Harnad's distinction between commercial and scholarly writing: Scholars wish the highest possible distribution of their work, even if it produces no incremental revenue, whereas publishers and writers of commercial material seek to maximize profits from publication, necessarily restricting access, through positive prices and enforcement of legal restrictions against unauthorized copying.

Any eventual alternative to the paper journal as we now know them must provide all of the functions performed by paper journals as described earlier -- quality certification, distribution, archiving, and indexing -- with equal or better capability, or at lower cost, or both. If any potential alternative fails in providing any of these functions, it will serve as an addition to the current system, providing better capability in some areas, worse in others, and -- at least for the foreseeable future -- scholars (and their universities and colleges) will continue to participate in the paper journal system.

Because of the limitations of earlier technology, each and every key function in the process of scholarly communication is carried out by the same process -- journal publication. Indeed, few scholars ever stop to consider that the various functions could be conducted separately from each other. The emergence recently of computer archives to store and provide access to preprint manuscripts in physics (Ginsparg citations) has called attention to the prospect of breaking off some of the journal functions -- distribution and indexing and (with proper attention) archiving. Indeed, the capabilities of computers to undertake text searches usefully augments the normal library processes of indexing and authority control in a way quite useful to scholars. However, even in the physics community that now relies almost exclusively on the file-server system for access to recent and new manuscripts, the process of submission to standard peer reviewed journals, refereeing by those journals, and publication of the paper copy of the journal article persists. In other words, the file-server system has added onto the paper system, providing enhanced convenience (greatly enhanced in some ways), but as an additional system, not a replacement system.

If the electronic publishing system is ever to serve as a replacement for the current paper system, it will be necessary either to add the "missing" important functions that the current system provides or to decouple the various functions, allowing the remaining missing functions to be carried out independently of publication. This idea of "decoupling" forms the basis of the proposal described here -- complete separation of the credentialing process from the remaining processes of publication. This separation is now technically possible in a way that it was never before during the paper publication process. If a way can be found to carry out and support financially the credentialing (refereeing) process independent of the journal publication process, then the decoupling can and should occur.

In the current system, the costs of the publication process are borne by the consumers of the material (individual and library journal subscriptions) and, in some fields, authors (through page charges). Some journals use a submission fees as well, although these fees seldom amount to enough to support the refereeing process. Many of these journals pay referees a modest amount, e.g., $25 to $50 per manuscript refereed, and at least partly justify the submission fees on that basis. One possible way to finance independent reviewing is to expand this system of submission fees in a way that fully supports the refereeing process.

III. A Potential Solution (and problems contained therein)

At the Caltech conference, participants discussed a process to decouple the certification process from the dissemination process with key collaboration with various learned societies. The central idea would have the learned societies expand their role to undertake a certification process for articles, independently of whether they are submitted for, or are eventually published in the standard paper journal system. Under such a system, scholars could submit articles for review (with an agreed-upon submission fee), and the normal refereeing process of the learned society would determine whether the article qualified for their "seal of approval," which, if received, could be affixed to any electronic version of the article as retrieved by others. With such a certification, if appropriately "honored" in processes that rely upon such certifications, including tenure and promotion in colleges and universities and grant applications from governments and foundations, the necessity to carry on with paper publication (which serves only the certification and editing processes in addition to the distribution, indexing and archiving that the computer file-server system can serve) could diminish or vanish at least in some settings. But until such refereeing, and in some settings, editorial functions are provided, the paper journal system will persist in parallel with whatever electronic system emerges.

Caltech conference participants explored some issues associated with such a system with a clear understanding that it might well not work in all disciplines, and perhaps that it might not work for any, but that we might learn some important issues associated with these ideas by exploring in more detail the issues associated with the proposed system. What follows are some issues that arose during the discussion and preliminary responses to deal with these potential problems.

-- Without the underlying costs of publication serving as a damper to the volume of papers published, would not we experience something equivalent to "grade inflation" where certification lost its current meaning?

This is of course a possible outcome, but not necessary, and perhaps not likely. Learned societies currently face this dilemma, with relatively low marginal costs of additional articles published even in paper journals. The most prestigious locations of publication retain very high standards, rejecting articles eagerly published by others. They will retain their prestigious position in the hierarchy of paradigm definition if and only if they retain such standards, so it will be in their self-interest to do so.

It is important, of course, to find some way to induce the organizations that currently form the highest standard of excellence in a field to participate in this endeavor, because without that level of quality certification involved, the process will fail to attract the highest quality articles, and in turn, will fail to receive the highest recognition in the field.

-- Will not a submission fee drive authors to (for example) commercial journals that make their money through subscriptions, and need not charge submission fees?

Perhaps, but it will be in the economic interests of academic institutions to support at least some submissions by paying submission fees as a routine cost of business, particularly with the ability to focus those funds towards the certification processes of cooperative learned societies.

-- Humanities faculty will never be able to afford such fees.

Colleges and universities, as noted before, can assist in this process. The proper mechanisms for distributing such funds will probably have a flat entitlement to every faculty member, plus some resources focused through the department (and perhaps the school) to support the work of particularly productive scholars.

-- The submission fee will discourage the publication of some work now reaching print, particularly if the process (as envisioned) evolves into a world with fewer paper print outlets.

This concern confuses the processes of publication and certification, which have heretofore been linked, but can now be decoupled. The process of distributing manuscripts will be greatly enhanced by using a computer-based system. Distribution of scholarly ideas will be enhanced. What will possibly diminish is the volume of certified articles

While few would probably acknowledge the thought in person, it would be fair to report that many participants of the conference (and possibly many readers of this report) privately feel that there is too much published material, much of it with little or no sustaining value, manuscripts published because of the pressures of the academic promotion and tenure process. The demise of lower-tier journals, if this is true, would be a welcome change in the academic landscape. Of course, one must ask why such journals continue to flourish in a free market economy such as that of the United States. The answer potentially lies in the resource allocation process of universities and colleges, wherein the journal subscription costs channel through a library staff who receive professional prestige and acclaim by supervising ever-larger collections. Indeed, the prestigious Association of Research Libraries heavily relies on collection size and additions to collections as a measure of a library's value and standing.

Articles brought through this process will not be "counted" in tenure review processes in colleges and universities, or at least the fear of such an outcome will inhibit young faculty from submitting their work through this process alone, bypassing the paper journal system.

This process is collectively under the control of our colleges and universities, but any actions taken to resolve the issue must be widely, not narrowly accepted. It does little good for a professor to achieve tenure at University X on the basis of e-publications if no other university or college accepts those publications in the promotion process, because the faculty member's mobility options would otherwise be severely limited. Achieving widespread support for a system of e-publication as a substitute for paper publication will require the shift of the work of "major" figures in a field to this mode of publication and certification.

Whether this can be accomplished or not remains yet to be seen. It is clear, however, that without the support of the most prestigious learned societies in a field, the transition is unlikely to occur, and with such support, it is more likely. The editors involved in the refereeing and certifying process help define what "counts" in a field, and by so doing, they help define who counts. Thus, the key will be to get the top people in a field involved from the very beginning both in the certification process and the publication of their work through the new e-publication and certification process, bypassing the traditional forms of paper publication. The task of the scholarly community, if it wishes to find a viable alternative to the paper publication system, is to encourage and sustain a substantial shift in the desired modes of publication of the leaders of various fields of scholarly work. When that occurs, the transition will take place.

-- Without paper journals, the role of learned societies will diminish, and they will whither away.

Not if they are resourceful and imaginative. For example, one can readily see an electronic intellectual forum springing up around a learned society's "file server" and certification process whereby people can attach "post-it" electronic comments to electronic equivalents of manuscripts, commenting upon them, and upon each others' comments. Learned societies can maintain their value by controlling access to such forums (e.g., by limiting the ability to read such commentary to those who have paid an annual membership fee, and even by creating a more select group of "fellows of the society" who are allowed to post comments electronically.) Properly guided, the world of electronic publishing offers the opportunity for learned society to expand its role in defining and refining the intellectual paradigm of its members, and can do so in a way that keeps the society's apparatus functional and self-sustaining.

A Modified Alternative

One study of scholarly communication (Dow, 1997) suggests the following as a potentially important alternative: Dow's research of scholarly journal editors' views about electronic journals identifies a hybrid role for electronic and paper media that uses the strengths of both. He proposes a system whereby the electronic medium (similar to as described previously) serve as the initial "publication" forum for scholarly societies, but that, in addition, they use a secondary refereeing process several years after electronic publication (and certification) to garner a subset of all such articles for ultimate paper publication and archival, using (to invent a title to characterize the idea) Archives of the American Society for X as the final publication medium. This "archives" documents would be even more selective, allowing the paper publication of the works that proved most important to the field in the period following their electronic publication. The role of the learned society would increase. The current lack of confidence in the "ephemeral" electronic storage medium would disappeaer, as least for the "most important" works (which, of course, most authors would believe they would reach).

IV. Subversion?

Steven Harnad has described "a subversive proposal" wherein electronic forums for the collection and dissemination of scholarly articles arises in competition with the current paper journal system. His own view is that the existence of such a distribution forum alone suffices to subvert the strong hold that paper publication holds on the colleges and universities around the world. The thrust of this conference's work suggests that while such computer based systems of collection, archiving, and distribution of scholarly articles is necessary, but not sufficient, to subvert the paper publication process. In addition, the development of an independent, decoupled process of certification will be necessary.

It is in the economic and academic interests of scholars and their employers (universities and colleges and others) to achieve a transition away from paper publication as soon as possible in as wide a set of scholarly fields as possible. The economic interests are obvious. Under current mechanisms, colleges and universities pay the scholars to carry out their work, then (in some fields) pay for parts of the publication process (through page charges), then pay large annual fees to publishers for library subscriptions so they can carry out the archiving function so important to academic work. Then, when we wish to use the work of our scholars in our teaching -- indeed, even when our own faculty wish to use the work they have personally created -- we must pay again through copyright fees to reproduce the work for our students.

The costs of acquisition and archiving of scholarly work in journals will continue to get worse, not better. More journals enter existence every year. The per-journal costs climb at rates far higher than inflation, and represent one of the most rapidly increasing demands on university resources. Ultimately, one must also count the costs of the buildings to house the journal collections and the staff to manage them as incremental costs of additional publication.

Electronic publication of scholarly journals solves many of these problems. It produces a one-time reduction in the costs of dissemination and archiving, since (by casual estimates) about half the cost of paper journals is associated with the physical costs of reproduction and shipping. Perhaps more important, the time path of costs will change in electronic media, since the costs of electronic storage have (and are expected to) fall annually, vs. the rapidly rising costs of paper journals. And ultimately it alters the nature of capital and human resource demands on our universities.

Perhaps more importantly, the scholarly value of shifting to electronic publication is great. Dissemination of scholarly work can occur faster, and with worldwide distribution, at very little marginal cost per user. The ability to search electronic data bases for relevant work far, far surpasses even the most lucid forms of cataloging and indexing that currently limit scholarly work in the paper forum. This almost certainly is the prominent reason why many scholars prefer "browsing" in libraries to scholarship that relies on limited indexing and cataloging that paper publication support.) Electronic forums provide exciting possibilities for alternative forms of intellectual commentary, including the electronic "forum" mentioned earlier, that can markedly alter the nature of intellectual discourse within a discipline. Such forums will flourish and grow much better in an environment where the entire medium of communication of the scholarly community is electronic, not just paper, plus perhaps annual face to face "conventions" where scholars meet and talk about their work.

We have in our power as the leaders of higher education in the United States, and hence, the world, to guide and support a transition to electronic publishing in a way that will supplant and replace paper publication, or we can sit by and let the process continue on its own current drift, a process that will very likely lead to the parallel development of electronic distribution systems in concert with the continuation of the paper system of refereeing and distribution. We have nothing to lose except our budget over-runs.