In games where PC death is off the table then either other loss conditions will replace it (loss of wealth, or status, or health, or whatever) or the game will be in win-only mode - and given the definitions we've seen so far, would win-only mode even still count as a 'game'?
I already noted that if you look around you'll quickly find a great many definitions of "game". And, I'll bet that most of them were created before RPGs, or were created by people who were not considering RPGs. Or, they were actively trying to exclude certain kinds of activity from the "game" category.
And, I think moving forward with a concept of "game" that is not generally inclusive of a lot of RPG activity... is going to end up in tribal argument - like trying to tell people that any particular edition of the game was "not D&D."
Aside from all that being rude, annoying, and kind of jerkish... I think that in so doing you
lose huge amounts of wisdom about play. As a result, in terms of having language to discuss what we do.. our language would be incredibly lacking.
And, maybe therein lies a major thought - if we are looking for a discussion of RPG theory, a discussion of "game" in the classic sense of the term... is perhaps missing the literal central bit: the Playing. I was askign bout the goal of play, not the goal of the (sub)game.
So, then, I think you are being remarkably limiting in thinking of play only in terms of win/loss. If we think of the game as structured play (which may or may not have other attributes, but I suspect we can agree on structured play), then winning and losing are really only one of a wide range of considerations.
Upthread, Belerophon listed some 8 different kinds of games from an essay. In only one of the 8 (the Challenge Game) is determining win/loss necessarily a primary goal of play. In many of the others, who wins or loses may be irrelevant to the players, or even contrary to the basic goal of play, insofar as determining win/loss ends the play, which may not be the desired outcome.