'It seems more plausible to take Wittgenstein as saying that the ability to use the concept game effectively typically does not depend on knowing of, accepting, or using any such definition.'

I thought this was a 'weak' thesis ( it's come up several times already in this thread, by the way) because it doesn't entail the stronger thesis that there is only a family resemblance between games (or whatever), but is entailed by it. If indeed there is a commonality between games  (or whatever), it suggests the possibility that we are actually aware of that commonality and that it directs our linguistic usage, even though we have a good deal of trouble expressing it in a definition. Some evidence for that is that we can tell that an alleged commonality is real because it explains and illumines linguistic usage. It seems to me that we sometimes may have inchoate theories implicit in our concepts. The philosopher seems to be telling us something we already knew, it rings a bell, it explains distinctions we make naturally without knowing quite why. You recall Plato's charming theory that we've all seen the forms, that they still operate linguistically, but we 'forgot' them in the trauma of being born. The job of the philosopher is to help us recollect. Perhaps the forms are contained in our linguistic heritage, inchoate but operative,
and the philosopher helps us say explicitly what has been motivating linguistic usage all along.

I'm away from home and don't have  my copy of the Investigations, so I can't say much about 83, except that I don't see it creating difficulties. As to your examples, a good deal of what I say earlier in this thread pertains to them, even mentions them, as does my short paper. thanks