Re: Bayesian epistemology versus Geddesian epistemology

From: Eliezer S. Yudkowsky (
Date: Sat May 28 2005 - 20:59:56 MDT

Paul Fidika wrote:
> Oh god no! Hopefully no one on this list wastes their time
> doing such a thing! A word of advice: don't waste your time
> with the "philosophy of science". (1) Bayes already presents a
> solution to the "problem" of induction. (2) Bayes presents a
> quantitative "Occam's Razor", obviating the need for determining
> which hypothesis is simpler.

Er, (1) isn't quite true. To get induction out of Bayes, you need Solomonoff
induction, Minimum-Description-Length, or some other way of assigning prior
probabilities that assigns lower probabilities to more complex hypotheses.

> The applied Bayesian-statistician daily makes use of these
> solutions, while the philosophers will continue to argue about
> these solved problems for decades to come.
> When you have the calculus of science, of what use is the
> philosophy of it?

The calculus of intelligence isn't finished. Don't judge the usefulness of
philosophy by philosophers. Some are good, some are bad, and I'll be damned
if I see how Jaynes could have written Probability Theory: The Logic of
Science without a touch of philosophy here and there. The moral is: know ye
the math, use ye the math, and complain ye not about the math until thou
master it fully.

Eliezer S. Yudkowsky                
Research Fellow, Singularity Institute for Artificial Intelligence

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:51 MDT