Pokemon is a game I wish I could get more out of. There is no game which has more explicit and elegantly integrated stakes.
For those not familiar, Pokemon is a weird idea that has grown huge in its success. The idea is that kids in the setting go out and capture stange, kind of magical-through-a-sci-fi-lens creatures. They then train those creatures and have them fight one another for badges and prizes. There might be other stakes riding on a fight (We're trying to save Pa's farm!) but the stakes are always explicit, and the path to them (victory) is always apparent.
This is awesome.
The only way it could be more awesome is if the creatures themselves were also the stakes. That is to say, if the prizes _were_ the creatures, and by deciding what you were fighting with, you're deciding what you're willing to risk. If you have stable of three critters, Alphasaur, Betadon and Crusher, you decide how much you value each one before you put it on the line for a conflict. If Crusher is your go-to critter, maybe you don't want to use him in a rsiky situation. Of course, if you do use him, you increase your odds of victory, but that's what makes it a tough decision. You're getting a huge amount of narrative power out of the mechanics without ever stepping out of play to establish stakes. That's _hot_.
Now, there are some mechanical hurdles this sort of system creates. Most notably, once someone starts winning, they tend to keep winning, because they accumulate resources. Similarly, if you lose your best critter, you're more likely to lose subsequent matches. This is a real, but not insurmountable problem. Having robust ways to refill your stable, and structures that limit advantage - like a fixed stable size - can go a long way to allowing a winner and advantage, but keeping it from being overwhelming.
The rub of all this is that this suggests the shape I want for conflict in a political or intrigue-centric game (Polimon?). Instead of creatures, you have connections, and the simple decision of what favors to call in and connections to use determines you capability _and_ what you're risking to accomplish it. It's harder to translate this thinking to more direct conflicts without getting very abstract, though if you are willing to _get_ abstract, the applications are obvious. Still for games with a strongly active setting - games where NPCs have agendas and activities - like Amber or any game where gods walk the earth, to say nothing of spy games and political thrillers, this could go a long way.
OBAMARAMA! I CHOOSE YOU!