NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/comp/ai/4953

Text File | 1993-01-21 | 3.0 KB | 68 lines

Newsgroups: comp.ai
Path: sparky!uunet!spool.mu.edu!agate!ames!ads.com!marcel
From: [e-mail] (Marcel Schoppers)
Subject: Re: New Problems in IJCAI Reviewing (long)
Message-ID: <[e-mail]>
Sender: Marcel Schoppers
Organization: Advanced Decision Systems, Mtn. View, CA [phone]
References: <[e-mail]> <[e-mail]>
Distribution: na
Date: Thu, 21 Jan 1993 19:10:50 GMT
Lines: 55


Once upon a time, when I was still a student, I was talking with people who
were well-known in AI.  Perceiving the consequences of their inability to
devote equal attention to everything they "ought" to be monitoring, I said,

    "The truth value you assign to a grounded, non-indexical
     statement depends on who the speaker is"

and I was ignored.  Coincidentally, about a week later, a visitor to the lab
gave a talk to the same set of people, in which he said,

    "The truth value you assign to a grounded, non-indexical
     statement depends on who the speaker is"

and the audience thought it Very Good.

Seeing this as proof of my statement made a week earlier, I sent a message to
some of those same people in the lab in which I said,

    I said: "The truth value you assign to a grounded, non-indexical
    statement depends on who the speaker is," and was ignored.
    He said: "The truth value you assign to a grounded, non-indexical
    statement depends on who the speaker is," and was believed.
    Therefore I was right, and furthermore, I said it first.

and the answer I got was "Smart-ass."

Of course, in retrospect I see that I simply made a statement and did not
carefully argue my case for a whole hour.  That may have had something to do
with it.  ;-)

But in general, I think it wrong to publish something from a well-known person,
just because s/he has built up credibility, that would be regarded as of
doubtful value when coming from someone else.  A case in point is the paper by
Stuart Russell on page 15 of IJCAI '91, which I bring up because the situation
was quite public, with Mr. Russell as an innocent bystander. I vividly remember
Mr. Russell, at the beginning of his talk, telling the audience that one of
the written reviews had said that the rewiever was suspicious of the paper's
value and was recommending it for acceptance only because of Mr. Russell's
credentials, when Mr. Russell himself was actually in doubt about the paper's
usefulness and was wanting the IJCAI reviewers to decide.  I thought it
shocking, not only that a reviewer had such a bias so unashamedly, but that a
large part of the audience failed to see the bias and regarded the situation
as funny.  Although I have since read the paper several times and do not
disagree with the decision to publish it, I still think that that reviewer's
bias was reprehensible.

    I continue to believe that good reviewing is a learned skill, and
    that reviewers ought to be accountable and punishable for their
    mistakes, as opposed to a scheme of things in which reviewers are
    untrained and untouchable.

Marcel