Automating reasoning about the future at Ought — LessWrong