Less Wrongians are frequentists to a greater extent than most folk who are intuitively Bayesian. The phrase "I must update on" is half code for (p<0.05) and half signalling, since presumably you're "updating" a lot, just like regular humans. When you consciously think "p<.05" do you really believe that the probability given the null hypothesis is less than 1/20, or are you just using a scientific-sounding way of saying "there's pretty good evidence"? Might this just be that people on LessWrong have (I'm assuming) nearly all studied frequentist statistics in the course of their schooling but most probably have not studied Bayesian statistics?