I don't understand why Airbnb was considered as a crazy idea and I hear literally everyone refer to it as an example of crazy idea. I always thought it was a genius idea from the first time I heard about it. But the way I see things, it doesn't matter what the idea really is as long as there is a market (not too big nor too small) and the team's ability for solid execution of the idea (kick ass product, partnerships and vision).

It used to be "sleep on the floor during conferences on air mattresses, with the host providing airbeds". I personally would have had a hard time extrapolating from "surge capacity for huge numbers of people during events" to "routine spare-room or whole-house rental", which seems a lot more like homaway/vrbo/etc. to me., a good business. The former is more like couchsurfing combined with the Red Cross, and IMO not good as a business.

You probably heard about it post-yc. It was initially a thing for renting out airbeds (airbedandbreakfast.com)

I think even that is a great idea...albeit with a much smaller market capability than what airbnb is now.

That's pretty much what http://www.couchsurfing.org do, but for free. Clearly there is both supply and demand for such a marketplace. Were airbnbs or YC aware of couchsurfing?

It wasn't a crazy idea at the time. couchsurfing.com had been a thing for years. "couchsurfing with a better UI" doesn't sound like a billion dollar idea, but it doesn't sound crazy either.

What everyone else said, PLUS it was a marketplace startup in a crowded space. Marketplace startups are ridiculously hard to curate, Airbnb did some really great hacks (with Craigslist especially) to finally make it work.

But there were many people who thought this kind of thing could be huge, and plus the non-profit Couch Surfing was pretty big at that point if I'm not mistaken.

Market is semi-legal. How do you gain traction? Network effects (that whole traction thing (re: gaming/tuning Craiglist)), liability, etc.

Retrospective analysis biases "initial" perceptions post-facto.

Ok, so what were the predictors of failure?

IMO, if you can talk about these features/predictors, they are true predictors. Otherwise, if people can game the system if they knew these features, I'd think they're more artifacts about your current process than attributes of failing startups (and if it turns out that these features are actually causative of failure, rather than just predictive, everyone can be a better startup too).

The fact that one can conceal a flaw doesn't mean it's not a real flaw.

In your defense, after the first few rounds of YC I remember VCs and bloggers squawking about all YC companies being features and not products, and that you were deliberately mass producing acqui-hires.

To me, it's clear that the quality of the average YC company has gone up.

"Things are always breaking at YC, because our strategy is to find bottlenecks by hitting them."

Can you share the details of what broke and how?

Even if you won't share your strategies for detecting & fixing the bottlenecks, I'd love some background on what they are.

The craziest ideas are usually "frighteningly ambitious" ones, I guess. So, it's easy to understand that these are the ones you, guys, like most.