Interactive fiction writer and designer Emily Short recently considered multiple-choice interactive stories, where -- instead of a completely free-form interactive mode where the game parser tries to figure out what the player asked for -- the game supplies a limited set of pre-determined choices for the player to select from, each leading to a different pre-determined event within a particular narrative that defines the story the player experiences.
Her concern is that this type of game doesn't allow the kind of emergent storytelling possible in "open-world" roleplaying games (RPGs) such as Fable 2 or Fallout 3. On the other hand, she notes that RPGs aren't very good at focusing game events on a particular well-paced and engaging story, and she attributes this to the smaller "granularity" of events in an RPG which are harder to tie together into a coherent dramatic narrative.
I think that's a rather good way of analyzing these two approaches to storytelling in story-driven games. I do look at it in a slightly different way, however. The incompatibility I see between character stats and story-relevant possibilities is that each approach puts dramatic choice in a completely different person's hands.
Emphasizing character stats means that attributes of the character, such as intelligence or lockpicking skills, must to some degree condition or even determine story choices, taking gameplay out of the player's hands. Too much of that and you get a simulation that plays like a movie -- you lose the "interactive" part of interactive fiction.
But emphasizing on-the-spot decision-making by the player can make the game about the player, rather than about a character existing inside the secondary world of the fiction and acting in a dramatically appropriate way for that character in that world. The fiction has to constrain choice in some way or it's impossible to tell a coherent and world-appropriate story.
Two approaches to synthesizing these models of play in interactive fiction might be described as The Middle Way and Some Of Each.
In The Middle Way you'd try to pick some in-between spot between the player and the character -- a medium granularity. That's probably relatively simple to implement technically; I'm just not sure how satisfying it would be as story-based gameplay. The character's nature would sort of matter, and the player's imagination would sort of matter, but neither could be strongly activated. I suspect there are already some games like this....
Alternately it might be possible to develop gameplay where you have both small choices (determined by the character's nature) that add up over time and drama-important choices (actively selected by the player) that form the core of a particular story. This sounds a bit to me like a system in which the player decides "what" to do and the character's nature (as encoded in RPG-style statistics) have an impact on "how" each choice is expressed.
I like the sound of that second approach; it feels to me like it might have a "best of both worlds" quality. But I can see a couple of potential gotchas. One is technical: for each major choice you'd have to code multiple ways it could be expressed based on each one of the relevant qualities of a character's nature. That could wind up being pretty cost-intensive, even if the payoff might be significant.
The other possible problem is how players might feel about such a system that takes some choice away from them. If for example I chose to create a character with a roguish nature, should I be unhappy if, when I make a particular choice at a dramatic opportunity, my character twists my choice in a roguish way with consequences I might not have preferred? Or would that help my choice feel even more satisfying than those common in today's story-based games where my "character" is little more than an empty vehicle in which I-the-player ride?
Still, I think the Some Of Each approach probably holds the most promise for interacive fiction that is both satisfying as drama and enjoyable as gameplay. I'll have to think some more about this.