That caught by eye too, but they address it:
When we further take into account that not all startups reach the point of getting venture capital funding (although Sam’s track record at Google suggests he is relatively apt), the risk-adjusted returns of a startup stake may be in the same ballpark as remaining a salaried employee, despite much higher unadjusted expected returns.
Well, in any efficient market model, the risk-adjusted returns would have to be the same. And if you don't have an efficient market model, how are you measuring risk-aversion?
See http://80000hours.org/blog/12-salary-or-startup-how-do-gooders-can-gain-more-from-risky-careers.
The expected value of risky careers like startups if often much higher than less risky careers. However, this is more than offset by peoples' risk-aversiveness due to diminishing marginal utility. But... if you're an effective altruist, the money you make doesn't have diminishing marginal value. So... it seems that risky careers like startups are a good choice, if you're trying to maximize your positive impact.
What do you guys think?