A Fervent Defense of Frequentist Statistics — LessWrong