Decision Theories: A Semi-Formal Analysis, Part I — LessWrong