A more reasonable way of putting it would be, what would be really interesting is a breakdown of the languages used by the last 100 successful internet startups, where "successful" is whatever metric you feel like.
It might be a better way of putting it, but it's not a useful question.
No language can be called "successful" if only 100 internet startups use it. Look at languages that are used in the real world, and don't artificially limit your selection process.
I might be OK with making the criteria: languages used for new programs (rather than languages in use for a useful product). I'm not sure what their criteria is though.
