One thing that the article misses is that there is a way to tell the difference: do you find your product useful? Clearly there are major limitations to this h

One thing that the article misses is that there is a way to tell the difference: do you find your product useful? Clearly there are major limitations to this heuristic; you might be strange or exceptional in some way that makes the inference fail, and a sample size of one (or even two, three, four etc.) will not give you the same certainty as a large userbase. It is, however, better than nothing. Note that this is not an "exercise at the whiteboard"—you have to actually build the thing. Given this caveat, their claim holds, but there's a clearly identifiable step (actually, there are several, but this distinction is not necessary for the argument at hand) between scribbling on a whiteboard and releasing a product for public use.