I don't think learning a problem is harder than you thought and then still solving it is a detriment. If anything this community should be the ones that understand that the full scope of a problem often becomes apparent once you're 80% into it.

How many startups are "guaranteed to revolutionize the world in 2 years", take 4 years to get a minimum viable product, and are gone in 6?

You solve the problem, you get my credit. Fully self driving cars though I think should have been more obvious that it wasn't going to happen anytime soon.. not just to Musk but to everyone working on it. It'll get there eventually too, but it's still early days.