Thanks to the glory of the Order of the Stick, I have just learned of the existence of the Edition Wars. If you, like me, are one of the few people in the gaming world that doesn't play D&D, then you won't know that apparently there are heated debates raging across the internet about which edition of D&D is better. Personally, I don't care; if I'm going to play a fantasy game, I'd rather play GURPS.
Anyway. In googling the Edition Wars to find out more about it, I encountered an argument between two players. A minor point of the argument was one debator was saying that he feels that in 'modern' gaming (as he calls it), there is too much of an assumption that the players will succeed. He feels that at no point is there consideration that the players will fail. This is a problem primarily because the players, under the belief that the GM will not allow them to be unsuccessful, enter situations that are too risky for them to handle.
This is an interesting viewpoint. I don't know if it says more about me that I've never made that assumption, or if it's just proof that I'm an 'old-school' gamer. But either way, it seems like a silly assumption to make.
There should always be a chance of failure. As a friend of mine once put it, 'I don't like to kill a PC unless the player is doing something really stupid.' Which reminds me of the time (I think I've mentioned it before) that I tried to lure an opponent into casting a spell that would kill him, but was unsuccessful in dodging that spell. The GM tried to bring my character back from the dead, which only irritated me; I'd put my character in harm's way, I should face the consequences of that action.
However, failing at a task shouldn't preclude the overall success of a campaign. A perfect example of this is in the description of Set Piece adventure design from Robin's Laws of Good Game Mastering. In this method, the story is planned as a series of encounters, with a rough idea of how the players can transition from one to the next. It is recommended that the GM consider at least two possible ways to get from one scene to the next. And one of the easiest ways to do that is to prepare one method as the result of success, and the next as the result of failure.
For example, if you were running a spy-thriller campaign designed in the Set-Piece style, you might have a scene in which the players have an intense battle on a boat that was drifting out of control towards a waterfall, and you need a way to transition to the next major scene, in which they fight their way into the enemy's secret hideout. The success option is simple: they beat up the bad guys on the boat, get clear of the dangerous river, and interrogate their prisoners. But what about failure? What if the bad guys win the combat, or the boat goes over the waterfall? Obviously, you don't want to kill the PCs outright (but don't go overboard in trying to keep them alive either). Perhaps you could just leave the PCs unconcious on the river bank while the antagonists escape. Then they can ask around and find some witnesses who saw the NPCs running into a certain valley a few miles away...
The point is, failure doesn't end the story. It just makes it more difficult to get to the next stage. Ease of transition is the reward for success in the last scene, not the guaranteed right of the PCs just for being PCs.
On the flipside, I was once playing in a game in which the successful resolution of the story depended on my character making a Wyrm Lore roll. Imagine the GM's consternation when I botched that roll! The PCs ended up fighting on the wrong side, and the servants of the Wyrm desecrated Central Park. Because of a single failed die roll, the entire campaign ended in failure. This is also not a valid approach to success and failure in a roleplaying game.
But the important thing to remember here is that success should not be a foregone conclusion. Otherwise, what's the point of playing, really?