Logarithms and Total Utilitarianism — LessWrong