Let me try to explain with an analogy. Imagine that you’ve been tasked with landscaping some concrete walkways through a park. You’re given carte blanche to design the park, so you’ve got some choices to make. You can have many different paths from the park’s single entrance to different destinations and exits within the park like the swimming pool, tennis courts, or barbecue pits, or you can have one single path that spirals around the entire park so that the visitors can see everything the park has to offer. There’s a big constraint though - you’re only given a finite amount of concrete to pour.

You could, theoretically, use your concrete to pour a lot of paths that spread out from the entrance. You could have them all branch from the main path to different points of interest, or you could pave a single path that reaches all of the points of interest as well. The path to reach these points of interest generally results in park visitors only visiting a handful of the park’s attractions before leaving, even though your team spent a lot of time building and paving the rest of it. Or you could try the single path design and have every visitor experience (most) of what the park has to offer… but then they’d complain that there’s only one path and that there’s really no reason to ever come back. 

Essentially, we developers only have so many resources to work with in our development schedule - we only have so many developers, so much time, and so much of a budget. For each branching outcome we have to acknowledge, that’s two (or more) possible paths that both need paving. If they converge back to a single path, the player can feel like the choices “don’t matter”. If they diverge, then we’re shortening the total length of the game by the amount of the acknowledgement of divergent outcomes. The amount of “concrete” we need to dedicate to acknowledging each choice you make can be a lot or a little… but the more we acknowledge the choice, the more concrete it takes, and that support means that there’s less concrete to pour for the overall length (and/or quality) of the path.

It becomes a delicate balancing act, especially when you’re talking about the possibility of a save game import. Whenever we release a game sequel, you have to realize that there’s a very statistically significant number of players that will pick the game up for the very first time. This means that they’ll never see any of the additional content that we spend development budget on acknowledging from a save game import. Then, you have to consider the numbers of people who actually do run their save game imports. How many possible outcomes need to be acknowledged? And if so, what needs to be sacrificed for that acknowledgement? Remember - the critical path and narrative of the game must take precedence over side content. More than anything else, a AAA game should be a self-contained experience that is both playable and enjoyable for a first-time player. Anything else is icing.