Barkley_Rosser comments on Probability is Subjectively Objective - Less Wrong

22 Post author: Eliezer_Yudkowsky 14 July 2008 09:16AM

You are viewing a comment permalink. View the original post to see all comments and the full post content.

Comments (68)

Sort By: Old

You are viewing a single comment's thread.

Comment author: Barkley_Rosser 17 July 2008 01:57:46AM 0 points [-]

Cyan,

Why should there be convergence to some such point when there is no underlying "true" distribution, either subjective or objective? Are you counting on herding by people? It is useful to keep in mind the conditions under which even in classical stats, Bayes' Theorem does not hold, for example when the underlying distribution is not continuous or if it is infinite dimensional. In the former case convergence can be to a cycle of bouncing back and forth between the various disconnected portions of the distribution. This can happen, presumably in a looser purely subjective world, with even a multi-modal distribution.