Another possibility is that YC could maintain a higher percentage of success, at a lower individual return. Because YC attracts some of the best young talent around, is very selective in funding, and seriously optimizes their fundees chances of success, it's not unreasonable that YC could maintain a better batting average than the VCs.
Another possibility is that because YC tends to form definite relationships with the founders they fund, they can count on the law of large numbers within, not just between, teams. For instance, YC lost money on Kiko, but is also invested in justin.tv--if justin.tv makes it, the YC[kiko folk] relationship could still end in the black.
N.B. This is all complete speculation; any more insight from pg would be appreciated.
Higher percentage of success seems like a plausible outcome of the YC model. Everyone involved is clearly committed to something more than the 15-20K. Perhaps that's the advantage of recruiting hackers instead of b-school graduates...it's easier to endow hackers with good business connections than it is to endow b-school types with smartness!?
Another possibility is that because YC tends to form definite relationships with the founders they fund, they can count on the law of large numbers within, not just between, teams. For instance, YC lost money on Kiko, but is also invested in justin.tv--if justin.tv makes it, the YC[kiko folk] relationship could still end in the black.
N.B. This is all complete speculation; any more insight from pg would be appreciated.