Publishing in journals still important

By Franco P. Preparata and John E. Savage

The rapidly developing public interest in the World Wide Web is causing many in the research community to propose that the Web become an integral part of the publication process. Although technological breakthroughs often profoundly alter established modes of operation, it is important to choose carefully the options they offer without sacrificing the good features of the old modes.

Disclosure versus publication: In the dissemination of research results, we should distinguish between two major objectives: disclosure and publication.

Disclosure is used to make results immediately available to a research community. Traditionally this function has been served by author-published research reports and, to some extent, by conference proceedings for which the publication delay is within reasonable limits. For this function the World Wide Web is ideally suited. However, for it to be a success, the number of Web sites holding professional materials should be limited, and good Web search engines need to be employed.

Anyone who has spent more than a few minutes using one of the general Web search engines is aware of the enormous amount of highly superfluous material that can be generated. However, experience with area-specific database search programs, such as Glimpse, demonstrates they can be enormously useful. That's both good and bad news because it's easy for papers not indexed in well-known databases to get lost--like the proverbial needle in the haystack.

Unfortunately, the disclosure process described above does not offer the close scrutiny provided by the journal-reviewing process. Journal publication in computer science is typically handicapped by a very long reviewing process. However, the outrageous publication delays we experience today are not intrinsic to journal publication. In other fields, notably physics and biology, journal reviewing is done in a matter of months. In computer science, papers languish for years. Given the long delays in journal publishing it is no wonder some senior members in our field do not encourage it.

Reviewing practices: Conferences, by the very nature of their reviewing process, cannot provide the rigorous standards enforced by the best journals. Even the most prestigious and highly competitive conferences have popularized a style that at best falls short of standards and at worst hampers future research. Limited space has licensed withdrawal of details (proofs omitted). Unfortunately, this has become the practice even when the length of the paper is well below the allowed maximum. When a conference paper in this format is not expanded for journal publication, a potentially fruitful area of research is stifled.

Disclosure is necessarily generating clutter, and timeliness may fully justify this. It is the function of publication to redress this shortcoming by letting a vast peer community pass judgment on disclosed research. The World Wide Web could supplement traditional media in facilitating access. However, it could be dangerous to replace the traditional editor/referee pool with individuals fully in charge of selectivity. This would remove the feature of anonymity that is crucial to the maintenance of high standards.

Ownership of copyrights: The cost of journals continues to grow at an alarming rate. As a result, research libraries are dropping less popular journals to make room in their budgets for more popular ones. Journals are increasing their charges because their costs are growing. In the process, authors are losing opportunities to have their work read. This may be nothing more than a natural competitive process--one that will work itself out eventually. After all, unlimited growth in any area cannot continue indefinitely without depleting the available resources. Electronic publishing will change the cost structure of research publication and give authors and the reading public more opportunities. It is an exercise that needs to be played out.

A side effect of the high cost of publication is that many publishers are insisting that copyright laws be strictly enforced. Multimillion-dollar awards have been made to publishers by the courts. As a consequence, some universities have become extraordinarily sensitive to this issue.

For example, Brown is warning faculty members that they will be subject to sanctions if they do not follow the law exactly. When the law is explained, many of us find it not only ambiguous but very distracting. The good news is that the Association for Computing Machinery recently changed its policy to allow classroom use of its copyrighted material.

This enlightened policy should be emulated by other publishers--a position taken by the CRA Board at its December 1995 meeting--because it serves them and our research community.

Their copyright statements will be visible to student readers, drawing them to the publishers. Meanwhile the research material, which is supposed to serve the community, becomes more readily accessible.

John E. Savage is a professor of computer science at Brown University.

Franco P. Preparata does research in computational geometry and parallel computation and is the Wang Professor of Computer Science at Brown.