Several scientific journals have come down firmly against authors putting drafts of their papers on the Internet before publication. At the BMJ we're not so sure.

In the following article I've set out our current thinking on the topic, but we would like to know what you think. Our ultimate policy will be guided by your responses.

We will add the comments we receive to the end of the piece. Include your name and email address in the body of your text if you want it to appear.

Please tell as many people with a stake in the debate as you can: the better the input we receive, the better the outcome.

Tony Delamothe
bmj@bmj.com

Why we should allow authors to post preprints on the Internet

  • Philosophical preamble
  • What are scientific journals for?
  • What might the brave new world of scientific publishing look like?
  • Downsides?
  • Summary
  • Feedback
  • Replies (last updated 20/10/97)

    • Philosophical preamble

    When confronted by any innovation that could change our practice we can respond in several ways. We can, more or less vigorously:

    1. Defend the status quo
    2. Bow to the inevitable
    3. Try to exploit the innovation for all its worth

    Believing that the possibilities of the Internet for scientific publishing are so fascinating and so profound, I have always plumped for the third option.

    What are journals for?

    When trying to work out how we can harness the benefits of the Internet it's worth recalling what scientific journals are for, not what they currently look like. Among other things, they are the means by which authors communicate the findings of original, important, and methodologically rigorous research to other interested parties in a timely fashion .

    Currently, we rely on our peer review process to select papers of sufficient originality, importance, and methodological rigour, and the journal commits itself to publishing them as quickly as possible.

    Because so much intellectual effort has been expended on regularising these processes our easiest response to the challenge of new ways of doing things is to argue for the maintenance of the "rules" that we have so painstakingly devised. But that is to take a very "journal centred" rather than a "customer centred" view, and that way lies oblivion. Electronic developments are rapidly changing the world of our customers (both authors and readers). We might want to stick with the hard copy paradigm, but the market has a habit of determining the eventual outcome.

    What might the brave new world of scientific publishing look like?

    Researchers might begin a study by posting their protocol on a web site for review by their peers, possibly followed by a call for collaborators and for assistance in recruiting research subjects. After the research is completed, early drafts of papers would be posted for comments and criticisms, which could then be taken into account in further drafts. At some point the paper would be transferred to a journal's web site (if the editors thought it had a chance of eventual publication). It might be made available on limited access (to specialist referees and statisticians) or on open access (for anyone to make comments). At some point the raw data from the study would also be posted on the Internet.

    After further revisions the electronic version of the paper would be given the journal's imprimatur and made available simultaneously in hard copy and electronic form. The time elapsing between submission and the journal's offer to publish and between ultimate acceptance and publication could dwindle from the current months to days (or even hours).

    Downsides?

    This more open, collaborative model cuts across the notion thata paper's content is a secret shared between the authors and the journal until the moment of publication. There are potential downsides, but none of the obvious ones seem fatal to me:

    1. Before publication, ideas could be stolen by others If (dated) protocols are posted on the Internet, along with early drafts of papers, priority disputes should be easier to resolve. A faster peer review process could reduce the time available for rivals to act on what they have stolen.
    2. The public could be seriously misled by the results of half baked papers. The public is forever prey to scare stories not based on published papers. We need to encourage the press and the public even more strongly to withhold their belief from scientific stories that have not appeared in reputable journals.
    3. Journals might become redundant if authors could widely disseminate their findings without the need for us. Analogous to the preprint is the conference presentation or abstract, which are already picked up by the press. But because journalists have got their fingers burnt so often with authors unjustifiably hyping their findings they seem to give them less credence.
    4. We lose our "public interest" defence for maintaining embargoes (ie doctors should have received the relevant BMJ before patients hear of important new findings) As for points 2 and 3 we need to encourage the belief that the only research findings that matter are those that have received the journal's seal of approval.

    Summary

    The arrival of the world wide web gives us a good opportunity for us to decide exactly how we add value to the dissemination of scientific information. Perhaps we're not adding enough: adding even more seems like the best way of countering the "threats" posed by preprints. Success is more likely if we work with authors rather than against them.

    Authors already circulate hard copies of drafts of their papers to colleagues for their comments. Using the Internet for this purpose is an obviously more efficient way of doing this. As the intended outcome is better papers we would be cutting off our nose to spite our face by trying to prevent them.

    And lastly, if we adopt electronic peer review, authors will be struck by the inconsistency of our desire to post their unpublished papers on the net when we have been conducting a campaign against their doing the same.