I don't know Vera, but I think that Simon is a brilliant guy, and that
his work with Newell, for all of my substantive disagreements with it,
was a model of intellectual integrity. Nonetheless, I found debating
with him in print a frustrating experience. His argument was structured
around two oppositions: mentalism versus behaviorism and objectivity
versus relativism. It seemed to me that Vera and Simon interpreted
all divergences from mentalism as manifestations of behaviorism, and
all divergences from their conception of scientific objectivity as
manifestations of relativism. In fact, the work they reviewed embraces
neither mentalism nor behaviorism but seeks alternatives to both and
regards the two of them as similiar in many ways, and my own commentary
in the journal was intended to analyze AI work in an empirically serious
fashion that nonetheless does not correspond to simple notions of
objectivity. My own perception, unfortunately, was that Vera and Simon
drastically misconstrued virtually everything that they wrote about,
forcing it into the pigeonholes of behaviorism and relativism through
selective quotation, distortion, and flat unsupported assertion.
Of course I do not think for a moment that they did these things
consciously or deliberately, or that they acted in the slightest bad
faith. I believe that they honestly did perceive the work they were
discussing in the ways that they reported, and that these forms of
perception are very much the norm in the field in which they work. AI
people have had a very hard time getting beyond these simple dichotomies
-- mentalism versus behaviorism, objectivity versus relativism -- and I
think that this limitation has contributed to the field's ongoing decline.
Unfortunately this decline has also been accompanied by a lot of reaction,
such as the bizarre attacks on the field's critics and dissidents (by
others, not Simon) in the AI Magazine. This trend parallels the reaction
against social analysis of science by people like Gross and Levitt
(see their book "Higher Superstition", which is structured by similar
dichotomies and is consequently fair in its opinions when those
dichotomies actually characterize the work they describe, and wildly
unfair in its opinions when they do not). The only lesson to be learned,
so far as I can see, is to maintain intellectual standards and develop
simple, clear ways of explaining the need to get beyond the dichotomies.
Phil Agre, UCSD