A very different way to design a story is to simply think of decision points that might make good threshold markers in the story.
As a GM, you might have a specific introduction scene in mind, but you allow the players to follow a variety of paths to get to the next checkpoint. You don't need to know what those potential choices are, you simply know that they will lead to a new decision that will either escalate the story or bring a new element into play.
I'm basically calling this method, the "Path of Clouds" or the "Path of Smoke". The nodes are amorphous blobs that become clarified as they are reached through the story. Each choice doesn't specifically lead to a new point; in fact, there are no specific choices. In each case the players could choose to do anything and it won't disrupt the overall plan, it simply leads to a new part of the story. It takes a decent amount of GM skill to play this sort of game, a lot of ad libbing and an aversion to writing things down in advance (a few pointer notes a usually good though, as is a solid setting understood by all participants). It's basically a way of ensuring a story flourishes in a sandbox environment.
This is basically the way I've been designing my roleplaying games over the past couple of years. Not just modules or scenarios, but the games themselves.
FUBAR works basically like this, on a subtle level. The players know that there will be a number of antagonists equal to the number of players plus two. There will be a number of locations, objectives and conspiracies equal to the number of players. No matter what the players do, the story will work it's way through the antagonists in an order that the GM/Oracle can't predict, it will take them to places, provide them with objectives and have them face off against shadowy conspiracies, all on the path to vengeance. They know that they'll reach the main antagonist eventually, no matter what their actual actions might consist of. At the same time, the GM/Oracle is working their way through a pile of secrets, throwing twists and difficulties at the players regardless of what they might be doing.
As the GM/Oracle's resources start to dwindle, the game naturally proceeds to its climax/end game.
The Eighth Sea works of the notion of a 5 act structure. It doesn't impose any specific story restraints on the characters, it lets them run wild; but the players are explicitly aware that as they accumulate successes, the story progresses. Once enough successes are accumulated, the story moves from introduction to impetus, to complication, and so on. When I'm running the game to a time limit such as a typical 3 hour convention session, I ensure that the first half hour explains the general rules, then each of the subsequent half hours plays out a segment of the story. If players proceed to quickly, I'll run the players through an extra complication cycle, or I'll ramp up with an extra degree of climax. But this actually leads to the next story structure...
Illusionism is a loaded term; it is with no irony that I refer to the above diagram as the "Path of Smoke and Mirrors". Illusionism can be a GMs best friend, or it can be a players worst enemy.
At the simplest level, the "Path of Smoke and Mirrors" could be substituted for the simple path. Each gives a starting point, then forces the players to a new part of the story regardless of the decisions made by the players or characters. But there is something more to it than that.
The path of Smoke and Mirrors ensures a story develops, and a canny GM could devise a series of generalised plot points along a story path (take for example the archetypal "Hero's Journey"). The exact places visited could be left open, but the moments of drama are known in advance. As long as the players continue moving forward, they'll reach the next part of the story...but if they deviate too far from the path, the GM nudges them back in the right direction.
Looking back on it, the GM isn't the only one that nudges the characters back onto the story path, game rules can do it as well. Humanity in White Wolf's "Vampire" or Adept Press' "Sorceror" can be used as a prompt to get the story back on track, MacGuffins could be inserted into the early stages of the game with narrative and/or mechanical effects. Rule mechanisms and macguffins don't have to be used all the time to keep the story moving, in fact they work best if used sparingly.
Going back to my Eighth Sea example, each round of turns taken by each of the players normally constitutes an act. A session can go slowly or quickly depending on the number of players, their willingness to ad-lib and their ability to follow their own agendas. The GM/Captain is constantly inserting mirrors to adjust the pacing of the game (through the use of a metagame currency called "Pieces of Eight"), but then again, the players also have the capacity to add mirrors of their own (because they too have a limited supply of "Pieces of Eight"). It becomes quite strategic as each of the players tries to direct the storyline toward an ending that will be most beneficial to their character, while the Captain is trying to reach the conclusion of the story that will best serve the ship as a whole. They all work against mechanisms that are continually twisting and turning the game as it weaves through time and space. Good sessions are all kinds of over-the-top gonzo awesome, bad sessions are abysmal. (My intended rewrite hopes to maximise the potential for good while minimising the potential for train wrecks).
I'm sure there are more types of story-game structures, including the truly freeform. But that's enough for the moment.