I think in many ways those are the same thing. It depends on how wide of a market you want to capture. If you're just solving a single problem, like web forms (

I think in many ways those are the same thing. It depends on how wide of a market you want to capture. If you're just solving a single problem, like web forms (wufoo), then it's easy as long as the problem is real. If you're trying to start a company that changes an industry, like Hulu or Tesla, your predictions need to be quite a bit wider in scope. I'm referring in the article to the latter kinds of companies.