[extropy-chat] Coin Flip Paradox

[gts]

On Wed, 31 Jan 2007 14:57:06 -0500, Jef Allbright <jef at jefallbright.net>  
wrote:

> it appears that as an instructor, Michael Resnik had constructed
> a strawman with the Principle of Insufficient Reason

Not a strawman at all!

Either the principle is a sound approach to decision-making under  
conditions of ignorance, or else it's not. He demonstrated with a very  
simple and general problem that it's not.

> .... demonstrating that no method, including the PI, is capable of  
> providing
> good priors under conditions of ignorance.

Exactly!!! That is exactly what he demonstrated, at least concerning the  
PI.

So you can throw it out as logical principle. Use it when you absolutely  
must, but don't imagine you're doing something with a firm basis in logic.  
Also you needn't think you're necessarily doing something illogical if you  
set your priors by some other method. It's okay for Bayesians to use  
different priors, as the subjectivists have known all along.

-gts