If I had to guess, "being the kind of person who's friends with a developer in tech" is a very good predictor for "someone capable of being successful in tech". If that was the case, most friends-of-techies that did bootcamps would succeed, leading to a general impression by techies that the programs work - even if they fail horribly for 90% of the people that take them.

The other thing to consider is that if 90% of bootcamp grads get an offer (and accept that offer) after their first interview but the remaining 10% are unhireable and proceed to interview with 100 companies each before giving up, you'd have to interview 10 bootcamp grads to find one worth hiring. (Assuming random distribution, that is - if you're the company that bootcamp grads apply to after failing the first interview, you might never get a good applicant) The applicant pool might not match the graduate pool, in other words.