Debate update: Obfuscated arguments problem — LessWrong