# Yoav Ravid comments on Can Bayes theorem represent infinite confusion?

• This math is ex­actly why we say a ra­tio­nal agent can never as­sign a perfect 1 or 0 to any prob­a­bil­ity es­ti­mate.

Yes, of course. i just thought i found an amus­ing situ­a­tion think­ing about it.

You’re not con­fused about a given prob­a­bil­ity, you’re con­fused about how prob­a­bil­ity works.

nice way to put it :)

I think i might have framed the ques­tion wrong. it was clear to me that it wouldn’t be ra­tio­nal (so maybe i shouldn’t have used the term “Bayesian agent”). but it did seem that if you put the num­bers this way you get a math­e­mat­i­cal “defi­ni­tion” of “in­finite con­fu­sion”.

• The point goes both ways—fol­low­ing Bayes’ rule means not be­ing able to up­date away from 100%, but the re­verse is likely as well—un­less there ex­ists for ev­ery hy­poth­e­sis, not only ev­i­dence against it, but also ev­i­dence that com­pletely dis­proves it, there isn’t ev­i­dence that if agent B ob­serves, they will as­cribe any­thing 100% or 0% prob­a­bil­ity (if they didn’t start out that way).

So a Bayesian agent can’t be­come in­finitely con­fused un­less they ob­tain in­finite knowl­edge, or have bad pri­ors. (One may simu­late a Bayesian with bad pri­ors.)

• Pat­tern, i mis­com­mu­ni­cated my ques­tion, i didn’t mean to ask about a Bayesian agent in the sense of a ra­tio­nal agent. just what is the math­e­mat­i­cal re­sult from pluck­ing cer­tain num­bers into the equa­tion.

I am well aware now and be­fore the post, that a ra­tio­nal agent won’t have a 100% prior, and won’t find ev­i­dence equal to a 100%, that wasn’t where the ques­tion stemmed from.