[ExI] Bayesian epistemology

[Jef Allbright]

On 8/6/07, Russell Wallace <russell.wallace at gmail.com> wrote:
> On 8/6/07, Samantha Atkins <sjatkins at mac.com> wrote:
> > If a mathematical method useful in decision making has become an -ism
> > then we are surely lost.
>
> Well it's been floating around as an -ism (or I think Eliezer called
> it something like "Bayesutsu", only partly tongue in cheek) for awhile
> now, so while the conclusion that we're surely lost is a little on the
> pessimistic side even for me :), I figured there was no harm in
> discussing the reasons why, while the mathematical method is great
> stuff as far as it goes, the -ism aspect takes things too far.

Russell, I'm sorry if you felt an implication of harm in my response
to your initiation of this thread.  I've been deep in software
engineering for the last several weeks and it tends to bring out my
analytical side, parsing **everything** literally and logically and
critically.  You and Lee and Gordon wind up being the unhappy
recipients of my "gifts" of criticism, perhaps unaware that if I
didn't respect you I wouldn't have even bothered to engage.

I strongly agree with your point that the Bayesian suffers when it
becomes the Bayesianistic.  Bayes theorem is elegant in its power and
applicability, but it's by no means the right tool for every job. That
would be like applying the elegant but abstract principles of
trigonometry or Newtonian physics to the task of catching a tossed
ball.

- Jef