It argues for validating your assumptions as quick as you can. Which is why mvps, prototypes, and click demos can be a good thing. It doesn't always work though. Digging in and burning through a few million of investor money before your product encounters real users, is the expensive way to do it.