Abstract: A Bayesian definition of surprise is as follows: Given a distribution P over models M and some observed data D, the surprise of D is the Kullback-Liebler divergence from the prior to the posterior:

http://goo.gl/cKE6w

A good experiment or a good conversation is one that has a high expectation of surprise, E[S(D, M)]. In this meetup we will practice having good conversations by using the strategy "Ask the question whose answer will surprise you most."

The meetup will begin on Wednesday at 7:30pm. For directions to Zendo, see the mailing list:

http://groups.google.com/group/bayarealesswrong

or call me at:

http://i.imgur.com/Vcafy.png