I'd be really interested if at the end of the cycle they published the success stats vs. their original invitee success stats. I would expect a lower overall success result - I'm sure there are startups that would never work ("our business model is to just directly give our VC money to our users and profit on scale") they would be denied, but it would be curious how the stats diverged.
e.g. comparing overall performance when controlling for gender, race, religion, field, etc