It pretty actively depresses me that my statistics class never actually discussed how loose the connection between hypothesis testing (what we normally call a statistical test, frequentist) and a genuine updating of hypothesis probabilities (Bayesian) actually is.

Dear frequentists and Bayesians: could you people kiss and make up already? The rest of us are pretty tired of having to run bad numbers to get papers published and of having computationally intractable statistics to run. Please come up with a compromise.