What is publication?BMJ 1999; 318 doi: http://dx.doi.org/10.1136/bmj.318.7177.142 (Published 16 January 1999) Cite this as: BMJ 1999;318:142
Just as in the modern world there is more than one way of being dead, so there is more than one way of being published. Publication is not a dichotomous event: rather it is a continuum. And the academic community should accept this, not resist it. This was one of the main conclusions of a recent workshop in Paris, organised by Unesco, the American Association for the Advancement of Science, and the International Council of Scientific Unions, on developing practices and standards for electronic publishing in science.
Traditionally a scientific paper has been deemed to be published once it appears in a paper journal. Publication might even be defined to the moment by the lifting of an embargo: the BMJ, for example, lifts its embargo at 00.01am (London time) on Fridays. But even in the old, predigital world publication was not precise. Authors often circulated drafts of their papers to colleagues, presented their findings at meetings, and published abstracts months or even years before publishing their papers in peer reviewed journals. The invisible college thus often knew of important research results long before their appearance in journals. Medical editors have agreed that these processes are an important part of science and not “proper publication.” They have thus been happy to publish the final versions of these papers in their journals.1 The appearance of the mass media at many of these conferences (and the issuing of press releases by conference organisers) complicated the issue but did not destroy the agreement.
The concordat of the ancien régime is now, however, being destroyed by the appearance of “eprints,” preliminary versions of papers that are posted on websites and so made available to everybody.2 Such a posting is publication, say bodies like the New England Journal of Medicine3 and the American Chemical Society: they thus won't consider later versions of such eprints for publication in their journals. Nonsense, say the physicists and the astronomers, who both have long traditions of circulating preprints widely: that, they say, is publication with a small “p” and shouldn't stand in the way of publication with a big “P” in our peer reviewed journals.
Cynics say the line of both groups is driven by money and not by the need to communicate science effectively and efficiently. The New England Journal of Medicine and the chemists are worried that the wide availability of eprints will destroy their subscriptions. The publishers of journals of physics and astronomy, in contrast, have had no choice but to give in as eprint servers have appeared and flourished. The physics and astronomy journals do not, however, seem to have suffered: they remain strong and prestigious. The explanation does seem to lie in the size of the “p” of publication: researchers view publication on eprint servers as “weak publication,” while that in peer reviewed journals is “strong publication.” Strong publication is associated with prestige, credibility, reliability, wide availability, news coverage, and permanence. In other words, the want to publish both on eprint servers and in peer reviewed journals. It's not either/or but both.
Medicine doesn't yet have an eprint server, although it probably will soon.4 Those in favour of eprints argue that posting an eprint is not qualitatively different from presenting a paper at a conference: it simply means that more people can encounter and comment on the paper—so increasing the chances that it can be improved before definitive publication. Those against eprints in medicine have two main worries: firstly, that doctors, already stuffed like French geese with information, may collapse under the weight of poor quality (even plain wrong) studies; and, secondly, that the media will seize on silly papers and cause unnecessary scares. Eprint enthusiasts, of whom I'm one, respond by saying that the problem of information overload is so severe that it will not be made appreciably worse by eprint servers and that the media already have many opportunities—through conferences and off the record conversations—of picking up scare stories. Indeed, the appearance of a full account of a study on an eprint server would be much preferable to what happens now, when doctors often have no access to the full study.
Although medicine doesn't yet have eprints, it does have systematic reviews that are constantly updated. These have no definitive or final published version—only an original and a latest version. Along with other journals, the BMJ has agreed that it will publish versions of such systematic reviews despite their already being available in the Cochrane Library.5 The Paris meeting generally agreed that publication in various versions is desirable for the communication of science so long as readers are told exactly what they are reading. Is this an eprint that has never been peer reviewed? If it has been peer reviewed, what was the process? If it is a systematic review, which version is it? As well as letting readers know exactly what they are reading the scientific community must also work out ways of indicating the version of articles in citations.
Communication in science and medicine will not be well served by standing in the way of publication in many versions, and the BMJ is willing to consider for publication eprints that have been posted on websites so long as their status as eprints is clear. In the meantime, authors, editors, and publishers have more work to do to make the status of articles entirely clear. This is the age of transparency rather than paternalism.