It is true that most of the games which are crap are NOT pygames and conversely so too. But then the relationship with its converse is not symmetric. For example if a game engine has some award winning[1] titles then we can say that the game engine has good potential. On the other hand if the same game engine has games which are crappy then we cannot say that the game engine is crappy.
Hence, if there are some excellent games made in pygame then it gives weight to the statement that pygame has good potential irrespective of the fact that there are also many crappy pygames. And when I stress on commercial I do not want to know whether I can make a good commercial game out of pygame or not(although that can be a corrolary). When I focus on commercial[success], it is because people care enough to part with some of their dough, otherwise lip service is easy to come by. There are many friends of mine who will say that my game rocks, but is it good enough that they will actually part with some of their hard earned money? That is the primary reason why I stress so much on commercial concept of proof.
[1] Some might say that 'award-winning' is no big deal and they do not work for any award or money, which is fine. But that does not add any value to my argument above about commercial games(and/or awards).