[adam419]

I've thought about this myself too. I think it has to do with the fact that when you become convinced of something enough to actually work on it, you depart from the exploratory/learning approach and begin what more closely resembles a waterfall development approach.

That's why I think going the route of doing (or at least testing) the things you may believe won't work is better, because you just simply don't know what you don't know.

I also think a lot of successful companies were like this. They were sort of welcomed and blessed by serendipity.