As far as we're concerned, missing a good startup is by definition an error in our process. We may never get our error rate down to zero, but we can try. And it's reasonable to assume we have a lot of room to, considering how early we are.

Having funded 465 startups over 7 years may not sound early, and I suppose it isn't in a relative sense, but it is in an absolute one. Society in general and we in particular are still learning how to predict which startups will succeed.