The Hard Problem of Magic — LessWrong