[falcolas]

> He basically admits his model is no better than a monkey

Well, technically, a monkey with a coin to flip. "Heads this startup will succeed, tails it will fail"

[gerhardi]

Now coin flip wouldn't be anything near 50% accurate, unless there is a world where more than a small fraction of startups survive!

[Hasu]

What? A coin flip would still be 50% accurate. If 10% of startups succeed, and we say that heads is "succeed" and tails is "fail":

5% of startups will be predicted to succeed and will succeed (correct prediction)

5% of startups will be predicted to fail and will succeed (false prediction)

45% of startups will be predicted to succeed and will fail(false prediction)

45% of startups will be predicted to fail and will fail (true prediction)

50% true predictions, 50% false predictions. A coin flip is always 50% accurate.

[cousin_it]

A coinflip is always 50% accurate, no matter how improbable the event you're trying to predict. If I flip a coin to predict whether Cthulhu will rise tomorrow, I have a 50% chance of getting it right.