The following is a system meant to help DMs manage sitations where the players must accomplish a grand, overarching goal through the creative use of numerous different abilities. This is meant to be a step above traditional skill challenges in scope; standard challenges have encounter-size goals like "disarm the complex, multi-stage trap" or "escape from the city guard". Meanwhile, grand challenges can take a session or more to accomplish: "Discredit the mayor in the eyes of the town populace", "Negotiate a peace between the Orcs and the Dwarves", and "Improve the morale of an army's troops on the eve of an epic battle" are all viable examples. In the end, what really distinguishes Grand Challenges is that they adjudicate and assign value to the player's strategies in a non-abstract manner.
Grand challenges work by breaking up player's actions into different Approaches. On the first round, the player introduces their first Approach, by coming up with a tactic and then carrying it out.
If the players want to discredit Mayor Argus, they might try researching his past to dig up any dirty laundry, or spreading the word that old Argus is too "chicken" to take the fight to the bandits who have been plaguing the local merchant caravans.
Next, the players have to actually carry that approach out in-game; the DM decides whether this calls for a single skill check, a skill challenge, or something more. At the same time, the DM secretly rates the approach based on its merits. Then they compare this rating against how well the player actually carried the approach out, and use this to determine that approach's Success Value.
The DM knows that Mayor Argus has several shameful secrets, so he'll give the Approach of snooping into the mayor's past a high rating. This means that even a passable job of investigation will yield fruitful results, and an excellent job will net that approach a very high Success Value indeed. Meanwhile, the DM decides that the local populace has bad relations with traveling merchants and aren't normally targeted by bandits, so criticisms of the mayor on this basis could backfire. He'll give that approach a low rating, meaning that an excellent job of rumor-mongering will only yield a small amount of Success Value while a poor job could actually result in *negative* Success Value by inadvertantly increasing the mayor's popular support.
Each round a player or party (The GM can run thisngs either way) can introduce a new approach or continue to pursue an existing one. Both actions will see diminshing returns; each new approach's merit rating is penalized further, as is each repeated use of a given approach. A Grand Challenge is beaten when the total success value of all approaches reaches a given number; this means that player victory depends on them coming up with good approaches, then milking those approaches for positive Success Value by doing a good job of carrying them out (that and recognizing when to stop, lest an attempt results in negative Success Value and costs them valuable progress).
I hope this all makes sense. A subsequent post should offer an example of this system in play, so with any luck that'll make things clearer.