Tension, thrill, anticipation, dice and narrative control are many times hard to reconcile. Sometimes dice give too high an odd for failure. This makes it hard to keep control of the story. It may simply be too much risk to take a certain action. Would you roll on a 5% chance of failure when failure means a city gets leveled? On 1% odds? Even a 1 on a percentile die may be too much. On the other hand dropping the odds to zero takes the whole thrill out of the game. What is the point in rolling then?
Even when working with everyday tasks like killing a goblin, it might be interesting to have a absurdly unexpected outcome. A moment when the goblin rolls incredibly well or you roll incredibly bad or maybe the goblin's wife throws a pan at you and scores a one time perfect hit to your head. Who knows, the point is to introduce something that surprises the players, but not to introduce it so often that it seems the dice and not you are in control of the game.
Looking at this from the angle of dice probability distribution I see two key elements that we need:
- Strong central values
- Long tails with many very improbable odds
This way your character should be performing consistently well if the task at hand falls within the character's expertise or consistently bad if it does not. The dice should allow very few failures when the character is good and very few successes when the character is bad at something.
Lets take a look at fudge dice with a plus and with a minus modifier. The odds should look something like this (with good rolls falling above 0):
Good characters should have good odds of succeeding most of the time with a few odd events going wrong. On the other hand bad characters should be expected to fail unless they get lucky. The problem here comes around when good characters are too good or bad characters are just crippled as hell. Since the probability curve is caped at -4 and 4 there is no way to roll below 0 if the character enjoys a +5 modifier. Yes, we can consider a roll outcome of 1 to be a very mild good outcome, but there really is no room for failure. On the other hand a character that suffers a -5 to roll will never get above 0. We might consider the -1 outcome to be a very mild failure, but the truth is the character does not succeed. We fall into what John Wick mentioned in this article as "Yes, but roll dice". A means by which the GM drops the blame of saying "No" onto the dice. "I'm not saying no. (Just making it impossible for you to roll 0 or better)."
Now let us consider the following distribution curves. They have strong central values, but also very long tails. Characters will generally fall within those values, but they're not totally bound within them. There are small odds of amazing success or outstanding failure. These curves can be achieved with exploding dice and you can clearly imagine the excitement on the table as the values begin to pile up into and amazing success! That OMG moment when you see the dice totally in your favor and the GM's face when the high penalties get overwhelmed by your outstanding roll.
So back to defusing the nuke. A task so risky that failure means leveling a whole town. How do you roll and how much is "acceptable failure"? Is the dice mechanism you're using allowing too high an odds for failure? Can it be reduced? And if so to which value? Is this value for failure too low now? And if so what is the thrill of rolling?
This brings me to the topic I addressed in a previous post about the value of the unexpected and how to recreate this in games using dice. With normal bounded die rolls you simply can't. We tend to see dice as random elements in the game because we don't know what value will come up, but we do know. We know a 3d6 will fall between 3 and 18. There is no magic in it and certainly no great unexpected value. As John Wick mentions, bounded dice allow for "Yes, but roll dice" to be a cover-up for simply saying "No". You can't unexpectedly beat the GM with a 3d6 because the GM knows what the limit is and can adjust accordingly. You can't unexpectedly beat the min-max player with a 3d6 because the player has tuned the character in such a way that there is no way such a bounded die roll can defeat the character.
On the other hand unbounded die rolls allow you to break away from such a limitation. They introduce the concept of the highly unexpected into the game. There is always a clear and present danger that something can go against "one's plans". Even the most successful character has a small chance of failing and the most unskilled character a chance of succeeding. More important than these extreme situations is the fact that such improbable rolls can happen in everyday rolls. Every once in a while, maybe once per session, you can roll an outstanding roll when fighting our buddy the goblin and you get to narrate something awesome in what would have otherwise been an off the mill encounter. So maybe the goblin's wife does score an amazing hit with the frying pan that knocks you out and turns what should have been an otherwise trivial retrieval of the key into a journey into the tunnels below the goblin's house. And this to me is what great stories are made out of, incredible turn of events that shock everyone on the table, GM included.