The way we managed to fund fewer startups was to train ourselves to pick better. So essentially we said no to startups we would have mistakenly said yes to before.

Unless we can improve our ability to pick winners at the same rate the number of startups increases, which is unlikely, we'll have to grow.

Fortunately we can now. The reason we tightened up our filters in w2013 was to buy ourselves time to figure out how to shard YC. We've now done that, and we can probably grow another 2 or 3x before hitting the next bottleneck-- which will probably be the format of Demo Day.