Those founders would have learn a lot for next businesses in the future. The problem is that they did it with someone else's money.

It is a good idea letting people to fail early with the minimal amount of money possible.

That was exactly the idea of YCombinator. Experimenting before scaling could be cheap.