Alex Taborrok wrote:
> Returning to a more Armchairish point note that there has been an
> awful lot of discussion about the "will of the people." But we know
> from Arrow's theorem that the very notion of "the" will of the people is
> incoherent.
I agree that the notion of "the" will of the people is incoherent,
but does Arrow's theorem really say this? Doesn't it just say that under
certain assumptions a non-dictatorial procedure does not exist? If you
relax the assumptions (such as IIA), then many aggregation procedures
exist, including the Borda count. Of course, all procedures are flawed
in the sense that they don't reflect "the" will of the people and
can be manipulated, but this seems to be a general conclusion that comes
from studying these particular procedures, not just from Arrow's theorem.
And in any case, are these kind of aggregation problems really
what people have in mind in the current discussion about "the will of
the people"? I get the impression that they are referring to
correct revelation of preferences, not aggregation. Arrow's theorem
assumes all preferences are revealed with 100% accuracy. This is not
probably the case in Florida.
Interestingly, the popular vote is also so close that Gore's
national margin is probably well within that caused by voter mistake and
human error nationwide (and also in other close states). I wonder if
Gore's supporters who are arguing that a revote should take place in
Florida would also argue that revoting be done elsewhere?
Alex Robson
