A party of adventurers was seeking a certain secret: a name, lost in the mists of time. Their journey led to a vast and ancient library. They searched among the scrolls night and day, but found no vestige of the name they sought. Had their journey been for nothing?
Then one of the brethren, the keepers of the books, told them of a man who lived on a nearby mountain – a man impossibly old, and sure to have the answer to their question...
...And of course they find the old man, have to prove themselves worthy through trials, he imparts the Name to them, and they go on with their quest (with some healthy new XP).
The PCs rolled a skill test in Research, the roll failed, an NPC sends them to a more important NPC, from whom they learn what they’re looking for. Their failed research roll did not prevent them from finding the Name, just gave them delayed gratification.
If they hadn’t failed the die roll, they wouldn’t have been sent to the “old man on the mountain.” In fact, for game purposes, the old man on the mountain would never have existed. Thus the real game effect of their “failed” research roll was to remove then Name from the library (they couldn't find it there, so effectively it didn’t exist there) and to relocate it inside an NPC’s substory. Is getting it that way harder? Sure, but it’s also pretty interesting.
This raises the question of whether any skill roll ever actually fails.
It also brings up one of my core gaming theories: I don’t let players retry a skill test that another player has already failed. If the scholar fails the research roll, that’s it. If the rogue fails her lockpicking roll, that’s it – they have to come up with a different strategy.
Why do I do this? Because, frankly, having to look for another strategy is more interesting than searching the same library location over and over. Even if I only allowed one research roll a day, odds are they’d eventually succeed – and at that point nothing new would have been added to the story except some wasted time (which is a good plot device in some, but not most, campaigns).
Now, this creates a very interesting game effect. It means that rolling lockpicking doesn’t determine whether the door opens right now or doesn’t open right now – rather, it determines whether lockpicking can be used in this situation. A failure simply means that the ol’ lockpicks won’t get the characters out of this particular mess; they’ll still have a chance to get out, but it’ll be delayed gratification, which of course means they’ll have more opportunities to get nabbed along the way.
Treating rolls this way forces the players to keep thinking of new solutions – this keeps the mind moving, and the story as well. It also keeps the GM spry.
Of course, it has some significant implications. Now if one player hunts for food and fails, the party is going to have to make new dinner plans. If they roll Strategy and fail, they’re all surrounded. If they roll Research and it fails, the GM will have to come up with a side quest, as you saw above.
But all this generates good plot. As in the story above, the GM look at that failed Research rolls and interprets that now an NPC is needed – one who possesses the information. And since the GM already knows that information, it’s a piece of cake to whip up the appropriate NPC on the spot.
So that’s my theory: it’s the player’s job to constantly find “bridges,” from one part of the plot to another. Rolling a skill such as Research determines whether Research is a “plot bridge” your characters can use – success means it is, failure means it isn’t.
So characters who have high ranks in Research are going to be more likely to be able to use it as a bridge, but that route still has a chance of failing even for them. Failing a roll doesn’t mean that the character “should have been more skilful,” it means something far more interesting: it means that, in effect, the dice have just relocated the characters’ destiny.