Re: [UAI] UAI and double blind reviews

From: Thomas Richardson (tsr@stat.washington.edu)
Date: Fri Nov 16 2001 - 14:10:03 PST

  • Next message: Kathryn Blackmond Laskey: "Re: [UAI] UAI and double blind reviews"

    > TR: It might be argued that in small communities double-blind
    > TR: reviewing is more necessary since such communities are likely to
    > TR: be less accustomed to seeing papers written by "outsiders".

    Alexander Dekhtyar replied:

    > However this does not address the problem that in small communities the
    > reviewers would most likely be familiar with the work of the authors
    > "within" the community and less familiar with the work of "outsiders".
    > Thus, double-blind reviewing still does not blur the line between
    > "insiders" and "outsiders".

    Some lack of familiarity with the work of outsiders is inevitable. My
    point is simply that if the reviewing is not blind then reviewers may feel
    tempted to make less effort to understand an unfamiliar paper, instead
    resorting to: "I've never heard of these authors, probably it's not worth
    trying to read this paper thoroughly."

    Concealing the identity of the author may help because the fact that
    material in a paper looks unfamiliar still leaves the possibility that it
    was written by an "insider" who simply has diverse interests - this
    (albeit small) uncertainty can help to keep reviewers honest (and
    open-minded).

    Certainly the current UAI procedure whereby reviewers discuss their
    findings is also very helpful in this regard: if you do a sloppy review
    then you may be called to account.

    In short, I'm more concerned about preventing reviewers from recognizing
    that they don't know outsiders rather than preventing them from
    recognizing that they do know insiders (which is probably not completely
    possible anyway). Consequently I certainly wouldn't require authors to go
    to any enormous lengths to conceal their identity when writing papers.

    Thomas



    This archive was generated by hypermail 2b29 : Fri Nov 16 2001 - 14:17:55 PST