Interesting talk on Bayesians and frequentists — LessWrong