http://hss.caltech.edu/~hanson/calibration.ps
I had mentioned theoretical results which indicated that Bayesians
can't have persistent disagreements. In my research I focused on the
critique that such models require far too much computational abilities
of their agents, and generalized the result to highly computationally
constrained agents. Here is a title and abstract:
"Disagreements are about Computation, Not Information".
Consider two agents who want to be Bayesians with a common prior, but
who can not due to computational limitations. If these agents agree
that they can reason abstractly about the fact that they have these
limitations, then they can only agree to disagree about their estimate
of a random variable if they agree to disagree (to a similar degree)
about both their average biases. Yet average bias can in principle be
computed independently of any agent's private information. Thus
disagreements must be fundamentally about computation, rather than
about the actual state of the world.
Be warned: the paper has a lot of math.
Robin D. Hanson hanson@hss.caltech.edu http://hss.caltech.edu/~hanson/