Bounds of Attention — LessWrong