Are you all seriously arguing with Paul Graham about something in his realm of expertise that only someone with a vast experience in multiple startups can form an informed opinion of, or someone that has undertaken a serious research study thereof? If you are not one of those two, just smile, nod, and change your name. If pg suggests something, he thinks it's important, if he thinks it's important enough to write an article on, he's pretty sure he's right and it's important.