home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!think.com!ames!saimiri.primate.wisc.edu!zaphod.mps.ohio-state.edu!cis.ohio-state.edu!bounce-bounce
- From: byland@cis.ohio-state.edu (Tom Bylander)
- Newsgroups: comp.ai
- Subject: Re: New Problems in IJCAI Reviewing (long)
- Date: 21 Jan 1993 10:57:22 -0500
- Organization: The Ohio State University Dept. of Computer and Info. Science
- Lines: 91
- Message-ID: <1jmh52INN1i8@iris.cis.ohio-state.edu>
- References: <C10q84.390@cpsc.ucalgary.ca>
- NNTP-Posting-Host: iris.cis.ohio-state.edu
-
-
-
- I think IJCAI's reviewing system could have been administrated better
- (i.e., let authors know that the reviewing would be anonymous and that
- abstracts should be on a separate page), but I disagree with the
- other arguments against anonymous reviewing.
-
- Several reviewers have noted that it is often easy to guess who the
- author(s) of the paper are anyway. Part of this is because authors
- did not know about the anonymous reviewing. It does not take much
- work to make the guessing a little harder, though admittedly it would
- still be easy much of the time. However, focusing on this problem
- misses a key point. The point of anonymous reviewing is not to make
- authorship impossible to guess in every instance. Many complaints
- about the IJCAI reviewing process (probably about every reviewing
- process) is that there is a significant bias in favor of recognized
- researchers and places, and that these complaints come from people
- outside of this grouping. Certainly, there are many papers whose
- authors cannot be easily determined, and at least these papers would
- be on a level playing field. Reviewers (primarily researchers who are
- "inside") find many aspects of anonymous reviewing irritating, but
- they should remember that "outside" researchers can come up with a
- compelling explanation for this (compelling to those on the "outside"
- at least).
-
- Now I feel I must respond to some of Brian Gaines' points.
-
- > Authors have to modify their papers
- > extensively to remove any evidence of their identities, and these
- > modifications may distort the content.
-
- I don't anyone is requiring that authors remove every evidence of
- their identities. It would be simple enough to refer to your own
- previous work as if it were somebody else (e.g., avoiding phrases like
- "in our previous work \cite{...}"), and to remove identifying
- information from references to unpublished, relatively unknown
- papers. I doubt that this level of anonymization distorts much of
- anything, and this year it would have made some easy-to-guess cases
- much more doubtful.
-
- > The 'meaning' of text is situated in its origins and in its place in the
- > stream of discourse. The same statements do not have the same meaning from
- > different authors. The same text has a very different status as a report of
- > the ongoing research previously reported by one group, than it does as a
- > report of replication of that research activity by an independent group. In
- > the event of replication, the relation between the two groups, the degree of
- > independence, is very relevant.
-
- Give me a break. In general, either the paper is sufficiently
- persuasive or it is not, i.e., it presents sufficient evidence that
- the reviewer can evaluate. I cannot imagine a comment like "accept
- this paper only if it was written by x, y, or z" being taken seriously
- by any fellow reviewers, or "the claim by this paper is true if x
- wrote the paper, but false if y wrote the paper". In those cases in
- which independent authorship is important (i.e., "accept this paper
- only if it wasn't written by x because it claims to be an independent
- evaluation of x's work"), this is something that editors are good for.
-
- > A brief statement of a research result from a researcher with a
- > major track record of excellent, replicated research may be acceptable on the
- > reasonable assumption that she or he knows the state of the art, artifacts,
- > etc. The same statement from a researcher new to the field may require
- > in-depth justification. This is bias, but it is proper bias reflecting the
- > communication process involved, not improper bias reflecting personal
- > prejudice or self-interest.
-
- Argument by authority is a classic fallacy, and, because it is a
- fallacy, depending on such an argument has its own significant fraud
- potential, note Gallo AIDS research case and David Baltimore case. In
- this political climate, we had better be careful that the evidence is
- there whether or not the researcher is recognized, or we will pay for
- it later.
-
- > An improper effect
- > would be one that damaged the progress of science because one group had direct
- > power to prevent the dissemination of knowledge rather than the indirect power
- > of showing it be incorrect. There have been many independent empirical studies
- > to determine whether such improper effects exist in the scientific community.
- > The difficulties of measurement and interpretation are such that no studies
- > are conclusive, but any effects found have been slight.
-
- No doubt the "empirically measurable" effects of anonymous reviewing
- are slight as well, so this is no argument against anonymous
- reviewing, per se. It's only an argument for keeping things the way
- they are because we (on the inside) are comfortable with it.
-
- Tom Bylander
- byland@cis.ohio-state.edu
-
-
-
-