The most obvious benefits of such a system have already been stated: the player gets a consistent avatar and, hopefully, story from game to game. This is a great feature to have when one of your core tenets is to offer the player choice in how the story unfolds or how his or her character might develop; the level of reactivity in such a system does a great job in helping the player form an emotional bond both to the character they make and their own version of the story, as well as the game itself. For players like myself especially, I know that it can be rewarding to build up a consistent character both within a single game and between different games; I have a stock "me" character which I use in almost every RPG, and only change it up for the sake of repeat playthroughs.
Additionally, save importing allows designers to flesh out ideas which may have been hinted at in an original game but could not be fully fleshed out due to budgetary constraints, lack of time, and so forth. More optimistically, this also means that, if a developer knows a sequel is a guarantee, they can focus resources on creating a lasting, consistent story arc where players will be able to see the consequences of their actions well beyond the scope of a single game. For a writer who may not be able to reconcile all of his or her ideas in the same relative time frame or other context, it can be liberating to play with where events occur, i.e. if a player decision leads to a large battle, the drama in such a scene might overshadow existing story events, so it would make more sense to postpone it for when the game needs said drama.
The Witcher 2 takes "branching story" to a whole other level.
There are some pretty obvious downsides to all of this, of course, the most obvious being that this requires a lot of work for designers and writers. Game choice and consequence typically branches off like a tree: one event leads to another, and then that event leads to another still, and so forth. This can look pretty benign at first until one actually tries to graph this out as a flow chart or similar and the number of conditions, special clauses, potential plot inconsistencies, and sheer number of different story paths and nodes become apparent. The Witcher 2's story tree was recently revealed and suffice is to say, it's kind of a doozy. Trying to reconcile such a branching plot, including all the little decisions the player may or may not have made with the plot and world of a new game is a daunting task. It's easy to just go by the player's ending alone, but the player may not feel like his or her smaller decisions had a lasting impact.
Of course, on the writing side of things this can get pretty complicated. Writers simply cannot think in terms of one game anymore, otherwise the significance of importing a save file is lost. The entire purpose of such an act is to allow the player to experience a multi-faceted, branching narrative with emotionally compelling choice and consequence, and not thinking about how certain events can carry over runs the risk of trivialising the majority of the game's story. Not all decisions need to have a huge impact on the game, of course, but if the truly significant player choices are distilled down into only a few possible outcomes, with dozens of other minor decisions being totally superficial, it suggests the story is unfocused. BioWare ran into this problem recently with Mass Effect 2, in which they ultimately paid lip service to the player's past exploits without them actually having a substantial impact on the game world or story. Commander Shepard helped his or her share of people in Mass Effect; is there really anything to be gained in having each and every one of them come back and say "by the way, thanks for the help in the last game"?
BioWare's recent Witch Hunt DLC shipped with an embarrassing bug:
the add-on did not recognise the player's Dragon Age ending. Whoops!
There are also a whole lot of technical considerations to make, in addition to the design and writing ones. Save files need to be kept consistent, or at least interpretable between games, and redundant code may be necessary if major changes to a game's file system or storage are made. This also poses the issue of having to locate those save files, since especially on open platforms like the PC, not all save files may be easy to find or read. Worse still, this opens up the possibility of a whole slew of extra bugs and scripting issues that need to be tested. Dragon Age: Origins - Awakening ran into just such a problem, where if the player had a certain DLC add-on, the abilities gained in that DLC didn't work properly. These sorts of issues go beyond the technical in their impact - if the player has been led to believe his or her decisions matter, only to find a bug gets in the way and prevents his or her decisions from being recognised, it significantly damages the player's relationship to the game.
Based on all of this, I have to say that the conclusion I'm leaning towards is that importing save files just isn't worth the amount of time and effort necessary to reconcile all the decisions the player makes. On paper it sounds great, and marketing can certainly run with it, but given how it creates a lot of additional headaches for a development team, and how so many parts of a sequel can become bogged down in the task of living up to its predecessor, I'd much rather take a sequel without a save import feature if it meant that I got a richer story, fewer bugs and an overall more meaningful and reactive experience. Of course, this is one of those situations where I have to admit that I do appreciate the feature, at least if it's pulled off well, but frankly, I have yet to see any examples where it has been anything but shallow, its potential unrealised. For all that BioWare's fans appreciate the experience of importing their character between games, I'm at least one who could stand to do without.
[Image credit 1]
[Image credit 2]