My issue with the 10% rate is that the population of startups is not homogenous. Put differently, many startups are essentially doomed to failure at the time of the founding, and others have much better chances.

At least, that's the main thing I got from this:

http://www.paulgraham.com/die.html

If one takes serious measures to belong to that second group of startups which are not (necesarily) doomed to failure, then the really relevant question is: what is the success rate in this group? 25%? A third? A coin flip?

Of course there's also a non-trivial chance of belonging to the first group, but not acknowledging or being aware of it.

So with all of that in mind, it would be really interesting to see an update to the conclusions from http://www.paulgraham.com/notnot.html