the point isn't that good ideas are always crazy, more that most startups, even big successful ones, fall into a pattern of efficient innovation, which tends to create boring, uninspired products. point here is create a structure wherein you're encouraging people to generate unusual ideas. grab the ones that don't suck.
Then still Larry's quote was about the idea to scan & index every book ever written. It would be a very nice addition for the people that haven't read that particular quote.
Where are the idea examples he tried successful in this blog after he launched his crazy startup idea?
If you want to know how companies like Amazon & Google do this kind of experimentation but in a systematic way read Jim Collins.
