In Bioshock 1 the story starts with you arriving at a strange location, and receiving a request for help by radio, which gets you involved in the story. Later it turns out *SPOILER WARNING* that the guy who asked you for help was actually mind-controlling you with the key-phrase "would you kindly", and your character didn't have any choice whether to follow his request or not. I found that a brilliant plot element, because for once there was an explanation for something we actually experience in many games with a story: We don't get many choices which would really influence the story. Most of the time the main story elements are telling you what to do, and your choices are either to follow the story, or to stop playing. In Bioshock you only get to make a single decision (save or harvest the little sisters), and that decision doesn't change anything in the gameplay, it only affects the game over cutscene. Other games, like Dragon Age: Origins, give you only the decision to make in which order to do the various chapters of the story, and allow you some minor decisions which only affect the chapter you are in, and that in a minor way. Why don't games allow us more decisions that actually affect how the story progresses?
Imagine an interactive story where at the start you are asked to make a first decision, which controls in which of two completely separate and independent branches of the story you end up. Then, a while later, each of the branches would have another decision, again splitting up the story in different branches. It is easy to see that quickly you would get a lot of different possible stories, for example with 10 binary decisions you'd get 1,024 different stories. If all of these were completely different from each other, the game would have immense replay value, but every time you play it through you only see 0.1% of the game. Thus from the point of view of the developers of the game, such a system would be extremely inefficient. You'd need to program a thousand times more code than a linear game, and players certainly wouldn't be willing to pay a thousand times more.
Thus if the ideal decision tree looks like a tree with many branches, the real decision tree of existing games looks like a tree that has been severely pruned by an overactive gardener with a chainsaw. For example a story told by a quest series in a MMORPG only gives you at each step the decision to either accept or reject the next quest. If you reject any of the steps, the story simply stops, and you get no further reward. That being the obviously worse option, people just click accept, often without even reading the quest text. As the story is completely linear and doesn't allow any decisions by the players, the players rationally decide that they don't even need to know the story, but just need the short version of "kill 10 foozles" displayed in their quest tracker.
So if programming a story with lots of decision trees is ineffective, how can we possible arrive at games with interactive story-telling? The answer can be found in the stories that player actually tell about virtual worlds, for example on blogs. Nobody blogs about having helped NPC Farmer Brown with his problem of wolves eating his chickens by killing 10 wolves. Instead players blog about their interaction with other players, good or bad, from generous help from strangers to horrible pickup groups with ninja-looters. The more player interaction a game allows, the more interesting the stories that are told about it, which is why sandbox PvP games produce the most interesting (albeit often least pleasant) stories.
The current design of PvE in MMORPGs is completely static. 5 minutes after killing the wolves for Farmer Brown, the wolves are back, and Farmer Brown gives the same quest to the next player. Thus players have no impact on the virtual world, and their "decision" of whether to help Farmer Brown against the wolves is completely irrelevant and changes nothing. To allow interactive storytelling to happen, decisions need to have consequences. But of course we can't make a game where the decision of the first player to kill the wolves has the consequence that there are no more wolves for the other players to kill, because then there would be no game to play for the other players. The solution therefore must be some sort of dynamic situation with several possible states, for example a state in which farmland is small and beleaguered, and another state in which wildlife has been beaten back and farmland is large. When the wildlife side is strong, Farmer Brown would ask players to kill wolves, but if farmland is strong some druid NPC in the forest would give players quests to plant magical fast-growing trees on the farmland to grow the wildlife again. There could be many dynamic equilibria in such a game, and the decisions of players to help one side or another would actually have an effect on the virtual world, and thus be more memorable. Interactive stories would basically tell themselves, without each possibility having had to be individually programmed.
I do believe that MMORPGs as a genre are moving towards more dynamic PvE worlds, which will allow PvE stories to happen. Warhammer Online made a first step in that direction with public quests, and Guild Wars 2 announced to push that concept further, towards a virtual world which changes, and where the current state of the virtual world determines what adventures there are to do for the players.
The other half of the development is more possibilities for positive player interactions in PvE games. Giving players the possibility to positively affect each others gameplay in a PvE game, instead of the negative interaction already possible in PvP games, makes stories happen. Earlier games actually had some good systems, like the vassal and liege system of Asheron's Call, but those have not received sufficient credit and attention. If done well, systems which reward veteran players for helping new players not only create better player-to-player interaction and stories, but also vastly improve the new player experience of older games, and thus improve new player retention.
If these things happen, MMORPGs developing dynamic worlds influenced by the players, and improving player-to-player interactions, we might arrive at a point where those "massively multiplayer" games really deserve that label and don't play like massively parallel singleplayer games. And interaction between players will make interesting stories appear, without the need for mind-control.