Open problems in human rationality: guesses — LessWrong