Thursday, October 05, 2006

Embargoed Science

NPR's On the Media interviewed an old college friend of mine, science writer Vincent Kiernan, about his new book Embargoed Science.

"Embargoed Science" refers to articles that some journals, such as Science or Nature, send to science writers in advance of publication under strict rules not to write about the article until publication date. The embargo supposedly gives writers a chance to do interviews and background research before they have to write the stories. Kiernan argues that embargoes misdirect science journalists.

It sets up a system by which journalists are given powerful incentives to cover a certain kind of story about science – the latest "eureka moment," and those eureka moments come several times each week. And so the journalists are directed toward those stories and away from critical, skeptical coverage of science as an enterprise, with all its fraud, with all its successes, with all the ethical issues…Lots of very, very good science gets published in many, many journals. There are literally thousands of journals, and journalists monitor 30, 40, if they're lucky – the ones with embargoes. The ones that are not embargoed, that also publish very, very good science, like a bunch of geophysical journals, they are largely ignored by journalists.
We can learn a lesson here, though perhaps not the lesson Kiernan wants us to learn. If we want more publicity for our best results, we should embargo those results, sending the articles and supplementary explanatory material to reporters in advance. Though this would mean not publicly announcing those results before they appear in journal, something that might not fly very well in the computer science community.


  1. In the THEORY community the notion of
    not annoncing results is nuts- we email
    stuff, we have conferences, etc.
    In fact, its hard to really pin down
    an exact official day something is
    HOWEVER, YES, we should somehow get
    into the media. It would help if we
    had more exciting results that have a
    snappy punchline. This may be easier
    in Algorithms and Crypto then in
    Complexity, alas.

    bill gasarch

  2. I read this differently: Kiernan description just explains how the "science column in the media" works: science writers follow the easy path and write about "preselected" articles, rather than finding the news themselves. In particular, they don't bother browsing JACM to find news-worthy material.

    The issue of "official announcement" might be a problem for an article in a journal like JACM, but this is surely not a problem for articles in CS conferences (theory or non-theory) since I am sure they can report something as "new discovery" even if it has been "public" (e.g. as a technical report) for a couple of months.