(sent to analytic re: whether misplaying in a "language game" is a matter of breaching an implied customary rule for communication. Here's the quick answer: the idea is too anthropologic and needs something ideational)
... I am so happy you brought this up. Because this is exactly what bothers me about people who contemplate Wittgenstein. I think your position is both pragmatic and sensible. I think, however, it is fundamentally mistaken. It turns Wittgenstein's central idea into an ordinary and non-threatening conception. It makes him no longer a giant. It's the version of Wittgenstein we might find by good encyclopedia writers. (I say these things as observations, not as jabs. In truth I am trying to pay a compliment to what I see to be good, orderly thought).
But here is the larger trouble.
The game cannot simply be composed of implicit rules of social custom. That turns the idea of grammar into etiquette. For Wittgenstein, one does not trip themselves on etiquette; they trip themselves on conditions of assertability. The unit of analysis is the proposition. Better yet, it is the anthropology of the proposition and what brains do to that anthropology when it is their turn to play. Do me a favor. Write down the implied rules of communicating food orders. Then go to a restaurant and break one. Tell me if what results is anything relevant to either philosophy's mission or Russell's set of problems. "Language game" does NOT mean norms that accompany communication. That's the kind of stuff they have in communications departments.
Now let's get to Moore.
If communicative customs are what mattered, Moore would have been well within the rights to his "food order." It was indeed well within social custom to say he knew he had a hand and to proceed with that as exhibit 1 in the ritual of "proof" during philosophy's dark period. But what makes Moore's assertion problematic is that he deploys unsophisticated grammar. He played the idea of removing-doubt ("knowledge") in a situation where one does not meaningfully deploy this brain task (to something self-evident). In so doing he either says something nonsensical or pointless. He also confused the "folk sense" of knowledge, which meant "being aware," with what we might call its more relevant sense. So the point is that Moore had incorrectly derived the grammar of knowledge and then proceeded to make a logic out of a matter that never truly required any such thing.