This is a core mechanic in search of a purpose.  In the spirit of "having specific questions" -- for what would you use the following?

This grew out of an idea I had for a percentile based game, in which any failed roll immediately improved that skill by 1%.  I figure it's a tiny enough increment that you wouldn't get massive jumps during one session, it makes XP redundant, and it means there ALWAYS a payoff for attempting to do something.  Fair enough, right?  Plus, you're probably going to plateau eventually -- succeeding a lot and therefore not improving.

Then I was thinking about gas-tank games -- games where you have a limited pool of energy or mana or charges or whatever makes your funky powers go.  What if you have a gas-tank set up like that, with a dice pool mechanic where you count successes. 

In fact, let's do a concrete example.  It's a superhero game where your powers are fueled by frustration and emotion.  So if I want to blow you up with my mind, and I have Blow Things Up With My Mind rated at 5, I roll 5d6.  We'll assume that I have to spend "Power Points" to activate the power.  How many points?  I don't know.  I'll swipe the framework from "Aberrant" for the purposes of discussion and say two points because it's a mid-range power.

For every die that comes up 4-6, I do damage.
For every die that comes up 1-3, I gain a Power Point because I'm frustrated that my shot didn't go perfectly.

What does this do to the shape of play?  What behaviors does it encourage?  Is there an optimal point of balance?  I suppose the way to make it far more strategic is to allow the player to slide the success barrier around -- "I'm trying a really hard maneuver, so only 6s succeed while 1-5 generates frustration!"  This would lead, ironically enough, to a total disconnect between player and character, where the player would be hoping the character fails (in order to fuel up).  Is that desirable?  Is it fun to play a frustrated character, in a power-based game where power results from frustration?

It becomes really interesting if you combine it with the percentile experience concept that birthed it.  Now, if you roll under, you get your success.  If you roll over, your skill increases and you get a point of fuel to use.  But at some point, you're going to hit a statistical stasis point where you're succeeding so often that your gas tank never gets filled.  What kind of game does THAT model?

-G.

Hi Greg,

In your example, on average you'll end up with more power than you started with after each attack, assuming the initial attack cost two and you get one point back with half of your dice.  That's just a resource balancing issue, though, and maybe actually a feature. 

More interesting to me is the frustration piece of the equation.  What effect would it have in play?  If it's just color ("You fail!  You're frustrated!") that's a missed opportunity. It'd be far cooler if frustration (or whatever negative byproduct) has its own mechanical or narrative impact.  As a thought experiment, consider what happens when "frustration" is renamed "collateral damage".  Then you miss, you buff up, and you crush a city bus full of crippled children.  Then you are in the realm of tough choices. 

I'm playing with something like this for a wuxia game right now... d6's from traits rolled like so:

1 a point of Joss
2-3 Yin move
4-5  Yang move
6 a point of Chi

Yin is generally defensive, yang offensive.  Joss is used to manipulate circumstances, Chi to power kung-fu.  The 'power up' aspect of Chi is partially there to explain why characters in the films and shows don't just spam their best attack immediately and often.  Joss explains why everyone always seem to turn out to be the son of your dead father's sworn brother disguised as your lover's maid. 

So every die you roll means something potentially interesting, and it gives several axis for special abilities to wonk the dice around.

-B

I recommend a little modification:

Roll d% and succeed by rolling under your skill.

1. If you succeed, roll again. If your second roll is under your skill, the skill goes down 1%.
2. If you fail, roll again. If your second roll is over your skill, the skill goes up 1%.

This trends skills towards 50%. It also discourages players from using their high skills unless they really need them and encourages players to use their low skills all the time. Bizarre consequences ensue.

I think you're on to a keen idea.  There does need to be a balance to keep this mechanic from spiraling out of control itself-- as Jason pointed out with your example.
Ensuring that every failure stings would help to curb test-whoring (intentionally failing meaningless rolls to power-up), and thinking about it would bring about (in a supers game) an almost Spiderman feel.  Spiderman fails at everything but beating up the bad guys: his marriages (plural) are always a shambles, he's a depressed loner in a run-down studio appartment who can't keep a job.  He fails at everything that matters in his life because he NEEDS to beat [Villain of the Week], or people will die.  He's sucking up those Diff 6 "Be there for my wife" and "remain gainfully employed" rolls so he has the Power Points to take down the baddie who will kill all the innocents.  When he does decide to pursue the things that matter to him, his powers fail him (and those he needs to protect.)

I was discussing this with a comics-literate friend last night and he said "This sounds like good genre emulation - there's always the super-guy who fails, and fails, and fails, until he just explodes into utter success."

It'd be interesting if important things were so inherently difficult that the only way to approach them was through a series of failures that buffed you up. 

Jason wrote:
I was discussing this with a comics-literate friend last night and he said "This sounds like good genre emulation - there's always the super-guy who fails, and fails, and fails, until he just explodes into utter success."

...in Spaaace! tends to have this kind of success equilibrium.  Current failures pay for later successes (and vice versa). 

This was actually somewhat inspired by Dresden Codak Casts Magic Missile.  The comments describe a game where you need Charisma to cast some spells, and... um, I can't find it now, but like logic or cogitation or something else to cast the others.  So it's a balance between opposites.  Sometimes you want one, sometimes you want the other. 

This may be a logical outgrowth of the ORE's use of two success axes.  But those are very concrete: "Excellence" vs. "Opportunity" mostly.  What would be other axes you could use?  "Passion" vs. "Reason"?  That would kind of tie to the Dresden Codak idea, a little, somewhat.  You try to make an intellectual attack and if you're too emotional it fails, but you're juiced up for a screaming hissy fit.  Or you try to make a passionate emotional appeal, but you just get drained and depressed and in a strange calmness you fall back on reason. 

Hm.

-G.

Graham wrote:
Cards are rather an elegant way to handle this kind of mechanic, actually. Each player has a hand of cards and, at the start of each conflict, they get dealt another card. If they can't win with the cards they've got, they don't play, and they lose the contest; but they keep the cards for the next contest. So if they lose lots of conflicts, they stack up cards to use in the next conflict.

Or, of course, you can use dice in the same way: if your die loses, you keep it, and add it to the next roll.

Graham

I am making a game system a bit like your % thing.

When you fail, you get a mark on that skill, when the number of marks you have in a skill equals the % of the skill, remove all marks and add 3% to that skill.

I like the dice idea, but it is a bit overpowered. How about for every PP you spend, roll one die. That way you can either blast their freakin' heads off, but stay low on points, or whiff completely, but not go down.