YC isn't just about getting the best talent. Some YC founders are clearly very good, others are unimpressive. I've seen the whole spectrum.<p>YC could pick its crowd at random (not to imply that it does; it doesn't) and it would still work. A few bozos would get in and be likely to fail, but it wouldn't be remarkable considering the statistical noise already in the startup game.<p>Venture capital is a feudalistic reputation economy, and YC is a case of Paul Graham monetizing his reputation. Just being YC establishes that a startup has real potential and puts it ahead of the teeming masses. Paul Graham wouldn't have to do anything to make YC a winning trade for him and the startups.<p>Before anyone says I'm being unduly cynical, let me fill out a few of my points. First, Paul Graham earned much of that reputation. He's a good writer, he understands Lisp at a deep level, and his technical chops are strong even by engineer standards and put him ahead of 99% of financiers. Second, I'm sure that YC's picks are better than its rejects. Third, I'm sure he worked really hard on the YC program. My point is only that he doesn't have to. He could pick startups at random, do absolutely nothing, and it'd still work, financially speaking.<p>Can anyone replicate this? Probably not. Paul Graham deserves credit for carrying the banner. In 2002-4, the mainstream business world had written startups off and people believed there'd never be another 1990s-style tech boom (which was mostly froth, but had a lot of genuine substance). Paul Graham pointed people back to the truths that the dot-bomb-shocked business mainstream was prepared to discard, like (a) that technology really matters, (b) small businesses can be extremely efficient, and (c) smart people require a different management style that most corporations don't get. He deserves a lot of credit for keeping the banner in the air in the gloomy early and mid-2000s. That right move, on his part, is a major reason <i>why</i> he has a monetizeable reputation. Is that feat replicable? Probably not.