People like telling the story of the team who started going one way and ended up somewhere else, successfully, because it conforms to their idea that some kind of attribute must be a discriminating factor (in this case the team is that factor). But I simply don't know if that's a true statement or not. I'd bet that many times it's just a dice roll whether team X can handle as many pivots that are necessary to make it all in one go. That's from my observations of lots of teams, but it's just a guess.
GP hit the nail on the head: good entrepreneurship means knowing when to hang on and when to iterate. And there are no books for that one.
I think that because our brains love order, and because we're always trying to put order on everything we see even if it's not there, we take lots of cognitive leaps to build some kind of mental model on "what it takes to succeed". With the "great teams succeed" model, the brain has a field day because selection bias and circular reasoning is in full-force. Whatever your model is, however, it should include the role of chance or you're likely to be really disappointed. And life is too short to get so distraught over the part that chance takes in getting started.
For example, say that an entrepreneur has this burning idea for a revolutionary new MMORPG, one based around social interaction instead of hack 'n slash. They find out that very few people care, but in the course of building it, they end up building an easy way to share photos online. Should they keep pushing through with their original idea and refuse to fail? Or should they admit that their original idea was kinda useless and go with what people actually want?
This happens over and over again. The podcasting company that realizes their product is lame and invents a way to broadcast 140-character messages by SMS. The server-side JavaScript framework that nobody uses who decides to build a collaborative editing app on top of it. The guys who wanted to put art galleries online, then discover that art galleries don't want to be online and so branch out into general online stores. The HotOrNot clone that compares Harvard students with barnyard animals, then gets busted by the administration and becomes a social network instead.
Problem is that it's not usually possible to tell, a priori, which ideas deserve to fail and which ones will succeed given a bit more effort. Imagine if someone had told Larry Page that his idea to crawl the web and look at the link structure was stupid, and he'd be much better off with a web portal. And that's why entrepreneurship is hard.