Hmm..
The other day, I was talking to an acquaintance who is a video game developer for Midway, and we touched on the unrealistic expectations gamers nowadays have about games, and what games can deliver, given time and money invested.
(This is going to be very general, but should get the point across)
Take for example, transferring choices you make in game 1 to game 2. If you have one single A|B choice (for|against) in game 1, you'd need to develop 2 different paths for that one choice in game 2. If you have one single A|B|C choice (for|neutral|against) in game 1, you'd have to develop 3 different paths for that one single choice in game 2.
So, if you have a game with, say, 5 A B or C linear, non-related choices in game 1, that is 15 paths that need to be developed in game 2 that would affect the new content.
But what happens when choices are not linear (we don't like linear, right?), and an earlier choice affects a latter choice, and so forth?
So in game 1 you have a single choice A|B|C. If choice1 is A, then choice2 gives you options D|E|F. If choice1 is B, then you choice2 gives you G|H|I. If choice1 is C, then choice2 gives you options J|K|L. So that's 9 paths in 2 choices.
Now, what happens if you have choice3, and choice3 is based on both choice1 and choice2? The paths grow exponentially. It is one thing to develop that many paths for one game, translating all those paths to the next game in a meaningful manner to the new content would be a nightmare, not because a tree of choices cannot be drawn, but because the tree of choices would need to be rendered, animated, voiced, and make some sense in the context of the new game and its content.
Think about how many decisions are made at the prologue of TW2 alone, and how they affect the game you are playing even before getting to Flotsam: the Crinfrid Reavers, the Lavalettes... it is just not possible to port all of that in any substantial, meaningful way to the next game without waiting a decade for it and paying hundreds of dlls.