Electronic media are revolutionizing the way in which information is handled. The last couple of years have witnessed a surge in public discussion of the way in which electronic publication impacts print publication, and vice versa. Issues include unrestricted and timely access to information, peer review, protection of information (for example, that afforded by copyright) and journal proliferation and cost.

The relationship between electronic print (e-print) servers and traditional print journals provides a pivot around which these can be explored. Its consideration is also timely, with Harold Varmus, director of the National Institutes of Health (NIH), considering NIH support for—and contributing to the development of—an e-print server for the life sciences1,2. And the prediction by one of its lead proponents, Patrick Brown (of Stanford University), that the commercial scientific publishing enterprise, "as it exists right now, is going to collapse in a couple of years", consequent to e-print publication.

The most successful e-print server, in terms of the extent to which it is used by and serves an academic community, is the Los Alamos National Laboratory e-print archive (http://xxx.lanl.gov/), established by Paul Ginsparg in 1991 (ref. 3). The archive hosts articles concerning advances in physics, mathematics, non-linear science and computer science; anyone can upload an article onto the site, although automatic downloading from the site is prohibited. Subsequent to feedback or additional research, revised versions ("replacements") of the article may be uploaded; previous versions are retained. Some replacements are replicas of those accepted for publication by journals, and are accompanied by a comment such as "matches version published in Astronomy and Astrophysics". Indication of acceptance by a journal (not to mention provision of the accepted version) may add value to the online article and addresses a weakness of the e-print server—its indiscriminate nature.

The act of filtration by peer review is a central activity that distinguishes peer-reviewed journals from e-print servers (as we currently know them) and one that publishers routinely cite when defending their insistence on obtaining copyright permission from authors4,5. Central to this defense is the 'sweat-of-the-brow' effort—not inconsiderable—that editors invest in ensuring that data are rigorous, appropriate arguments and caveats are stated, and information is presented in a clear, accessible manner. Some believe that copyright and peer review are inseparable; it would appear that the connection has been made because copyright is perceived to protect the financial interests of publishers and thus the payroll of their employees. It can, however, be argued that this is not necessarily the case. Publishers of scientific journals make their profit on 'first publication' rights; income obtained from reprinting original articles is negligible. It has been proposed that a publisher's 'honeymoon' with accepted articles (during which the publisher would have exclusive right of reproduction enabled by a licensing agreement) would protect her/his financial interests and yet permit post-honeymoon reproduction on web sites other than those of her/his journals6. Of note is the practice of journals under the aegis of the American Physical Society (APS). These retain copyright but permit e-posting of accepted articles on the condition that such articles are not prepared and/or formatted by the APS or its vendors (ftp://aps.org/pub/jrnls/copy_trnsfr.asc) and that notice of copyright ownership provided.

Money isn't everything, of course, and the publication of multiple 'versions' of what is essentially the same 'story' raises concerns regarding utility and branding. Reproduction of articles appearing in the Nature journals is prohibited by copyright law7 (although copyright of articles by NIH employees is retained by the US government), but there is nothing to stop authors 'e-posting' revised versions of their manuscript as it progresses through the review process prior to acceptance. The acceptance of a manuscript that has matured on the web by a journal that retains copyright underscores a fundamental difference in print and electronic publication. Print publication imposes a discontinuity in the telling of a scientific 'story'. Typically, researchers gather sufficient data to 'make' a publication, publish it, and start working on the next episode—which may be published in another journal, at another time. To obviate this static effect, authors may wish to present a continuous story online. Most journals' copyright agreements preclude publication of accepted papers, which makes for a significant 'hole' in the story. On the other hand, the addition of unreviewed data to an accepted paper has implications that should concern everyone. This, coupled with the less-than-alluring prospect of e-novels (as opposed to stories) being posted on the internet, suggests that grafting an element of discontinuity into the design of e-print presentations would enhance utility.

Details of the life sciences e-print server conceived by Varmus, Brown and David Lipman (the latter, of the National Center for Biotechnology Information) have yet to be revealed. A distinguishing feature of this server, however, is a form of peer-review, whose details are currently under deliberation. Clearly, there are practical and political hurdles to be overcome in implementing an e-print server, not least those that may be erected by parties who perceive it to be counter to their own interests. Such parties may include the scientific society and the commercial publisher, whose revenues are dependent on subscription to their journals. This concept is familiar to Richard Johnson and his colleagues at the Scholarly Publishing and Academic Resources Coalition (SPARC), which seeks to inject competition into the arena of scientific publishing—and hence "push back prices [that have risen] astronomically" by supporting the initiation of reasonably-priced, electronic journals. The rise and rise of the subscription rates for some journals is truly extraordinary; an annual subscription to Gene, for example, will set a library's budget back by $6,487. While SPARC has concentrated its efforts on assisting the launch of competitively-priced electronic journals, its call this week for proposals that "offer a promising strategic alternative to inefficiencies in the current, traditional scholarly communication process" (it will award a total sum of $500,000) signals a broadening in its strategy to facilitate scientific communication.

SPARC, the author, e-print server and more traditional publisher have a common goal: the rapid dissemination of valid, scientific information. While the future range and identity of publishing venues is hard to predict, there is no doubt that those able to rapidly adapt to the needs of the community will receive its fealty. In a changing climate, a rapid mutation rate benefits the long-term survival of species—although the underwhelming response to e-print servers currently available to biologists reminds one that most mutations are deleterious and eliminated from the gene pool.

Readers of Nature Genetics will have views about the attractions and drawbacks of e-print servers. Why, for example, have biologists been apparently reluctant to follow a path opened up by others, years ago? We invite your comments*.Footnote 1