From: Crosby_M (Crosby_M@bls.gov)
Date: Thu May 08 1997 - 15:17:42 MDT
Robin Hanson wrote:
<My model applies to *any* situation where any two agents disagree
about some factual claim, as long as they understand what that claim
means. They do not need to share any understanding of other
context.>
I was going to ask, does "understanding what the claim means" imply an
agreement on the prior probabilities involved in the problem
statement, or just an agreement on what the terms of the debate are?
(I thought, very simplistically I'm sure, that the essence of a
Bayesian analysis was that you had these prior probabilities
describing the likelihood of a particular outcome in some problem
domain, *that everyone had to agree upon*, and THEN you try to come up
with a new likelihood estimate, or decision between several options,
given some additional evidence or arguments.)
BUT, after rereading your abstract, it seems that you're really
describing how to measure the average bias *between* competing
evidence or arguments, that is, quantifying the level of disagreement,
rather than trying to establish the likelihood of either one of the
individual arguments being correct, or of making a choice between the
two arguments.
Again, this is all based just on your abstract and my VERY inadequate
understanding of the subject, so thanks for your helpful response.
Mark Crosby
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 14:44:25 MST