It seems like there could be more than one way forward. The big successes dominating point to one path: finding those wild outliers.
But many startups exit without making it big and return 2-3X to their investors. If the median startup were doing 3X, you'd be making a great return. I don't hear much about this strategy. It would be interesting to look at the remainder of YC's portfolio with Dropbox and Airbnb removed to see what that distribution looks like.
My guess is that it will still resemble a power law curve.
If we just examine YC exits to date, Heroku and OMGPOP are around $200M each, $400M in total, which is likely over 50% of the rest. If we remove those too, there are a few exits in $30-60M range (Loopt, CloudKick etc.) which are quite likely to be over 50% of the rest. At the lower end of exit price curve, the power law might not hold, as acqui-hires typically have a rather standard price ($1-2M per an engineer)
But many startups exit without making it big and return 2-3X to their investors. If the median startup were doing 3X, you'd be making a great return. I don't hear much about this strategy. It would be interesting to look at the remainder of YC's portfolio with Dropbox and Airbnb removed to see what that distribution looks like.