Of course, because startup success is not a linear distribution. Most startups fail, only a minority succeed, so "randomly choosing one" is a wrong way to asses

Of course, because startup success is not a linear distribution. Most startups fail, only a minority succeed, so "randomly choosing one" is a wrong way to assess the whole market. And that was my point about bankers looking at moonshot projects as "money loosing". They should looks at startup ecosystem as a whole and assess how much return on investment it has long term, and then look at Google's moonshot projects the same way (or more favourably, because Google is not a random startup - they can execute much better).