next up previous
Next: About this document Up: THE FUTURE OF SCIENTIFIC Previous: Introduction



Publishing on line permits much smaller delays. However, the time for referees to report is often the major delay with print publication. Refereeing serves four functions.

  1. It conserves the limited resource of a publication for editorial time, printing and mailing and library shelf space. All of these limitations go away if publication is on-line and the author, as is increasingly the custom, does the typesetting.
  2. It serves as a competitive quality filter, so that authors can compete on how many publications in refereed journals the have gotten accepted. Other systems, such as inclusion in specialized lists can serve this function even better. The very best hiring and promotion committees don't just count refereed publications, they actually evaluate the work.
  3. It makes the papers better. Authors often acknowledge contributions of anonymous referees.
  4. It has an archival function. Once published, the paper cannot be revised by the author without a new publication. This helps historians and provides a basis for settling disputes about priority. On the other hand, it leads to unnecessary publication when an old paper could be improved, but a new one has to be written. With electronics we can have the best of both worlds. There can be two copies of a paper--one at the journal which is archival and another at the author's site subject to revision. The journal can link to both.

On-line publication will be entirely viable even if it imitates print publication in its refereeing style, but here is a proposal that might make it better. I call it light refereeing, and it has a distinguished precedent. I was curious how the unknown Einstein, an employee of the Swiss patent office, got four papers into Annalen der Physik the world's leading physics journal, in the year 1905 and wondered how long the refereeing process took. These papers revolutionized physics, but how could the editor know that in advance?

It seems that Einstein was not quite an unknown, having published before in Annalen der Physik. That journal's custom was that the first paper submitted by an author would be carefully examined, and Einstein's first paper had been reviewed by Max Planck. Once the author had been blessed, his papers would be published on receipt, and this is was the case with Einstein's four 1905 papers. Alas, we don't get to see a referee's report on the first paper about the theory of relativity.

Returning to the present, we can imagine the following light refereeing system. An author's first paper is refereed in the standard way. Once an author is blessed his papers are lightly refereed. Namely, they are immediately scheduled for publication after three months, but are sent to a referee who is asked to suggest improvements in style or content. If the referee does not respond, the paper is published as received or as the author has spontaneously revised it. Such a system will be more prompt than present journal publication and may be preferable to the growing custom of using preprint servers. Of course, the editor could decide that a particular paper required more or less refereeing than the standard light refereeing.

Jeff Ullman makes similar points in his Diatribe Against Paper Journals, also published in Computing Research News, May 1996 with the title ``Web will change the role of journals''.

next up previous
Next: About this document Up: THE FUTURE OF SCIENTIFIC Previous: Introduction

John McCarthy
Mon May 20 17:24:22 PDT 1996