The Premise: A game is most enjoyable when the player's real progress mimics the development of the player environment. A good analogy would be that learning a musical instrument is most rewarding when the music pieces provided as a challenge to play become more and more complex yet still doable. However, artificially modifying the player environment to automatically account for the player's state of progress is counterproductive: this would be analogous to pieces of music automatically being "simplified" to suit a learning player yet losing depth and quality in the process. In RPG games, the player's environment and character tend to progress independant of player development of skill, or irrelevant to it. In action games, the player environment has to be "unlocked" by the player demonstrating a sufficient amount of skill but the player character often does not change. The Alternative: An action/character developmental hybrid in which there is a systemic world that does not arbritrarily react to player skill (No "monsters get weaker in stronger areas if you're weak" a la ES:Oblivion) yet can be progressed through at whatever rate the player sees fit, with the contingency of a player character providing a material counterpart to the player's skill. Essentially, there are now three factors to consider: the difficulty of the area you are in, the prowess of your avatar, and your own play ability. Your avatar does not progress based simply on time investment, however. Activities in the game that enhance player skill increase player avatar ability (A good example of this is the skill system used in Aggressive inline. Using/practicing moves and tricks improves the avatar's ability to execute them, augmenting player skill and providing a maleable form of progress). In a sense, by teaching yourself, you are "teaching" your avatar. In this way, player progress is not completely unhinged (As in a system such as Halo 2's matchingmaking system, where time investment is only neccesary for the system to accurately assess skill in most of the lower levels. A champion player can reach level 50 far more quickly than a player who is above the level 50 skill cap but does not defeat opponents as often, simply due to statistical accuracy requirements), yet progress is not determined in lew of any consideration of actual player activity. In this way, both the player avatar (learning character) and player (learning by character) learn and progress in a world that is defined based on its own properties and does not change based on the player in a suspension of disbelief breaking sort of accomodation.

In action games, there certainly are character 'stats' to keep track of, and they're pretty simple. Your choice of weapons/gadgets, to name the most prominent one. :)

Personally, I don't like my character to have too much stats, as I'm essentially giving away control: not my skills are what count, but some characters skills. This breaks immersion for me, or at least lowers it. When I know I'm pretty accurate with weapons (in games ;)) but my character isn't... then I can't connect with my character anymore in that area. Upping such skills by frequently using them probably lowers this problem, and I think that's a good choice: not simply rewarding time investment but specifically skill usage. However, I believe this strongly depends on what kind of players you're targeting. For example, as a FPS gamer, I don't want my character to change, because I am that character. He changes as I change. Especially the first person view helps a lot to achieve that, and I don't want things that break that immersion.

This would ideally be a sort of RPG, hence the question of the alternate exeprience curve. Adding player-separate character stats into an FPS is tricky due to the general twitch-based nature of those games (Character development separate from player development in most FPS's is mistakenly on a scale of "suck to expert," the problem being that the default player character ability in an FPS is closer to expert to begin with. In a game like counterstrike, your "character" is actually far better at using guns than you would be in real life. Most FPS/RPG type games start you off as a novice, instantly robbing the value of the game. The point is supposed to be that you play as a character that essentially has superhuman skills: any form of advancement should come on top of that, such as reducing recoil etc...but I digress). In an RPG, adding in action elements does not neccesarilly have to degrade the character "development." In fact, adding in elements of gameplay in synergy with the impacts of a skill system only really adds to the strategy. In any case, I'm lucky that I don't have to tackle the issue of putting the RPG into action, just the other way around.

So... a system that relies on player skill, while still giving a character progression and empowering feel, and whilst not 'gimping' the lower skilled avatar, making the player 'give up' early on due to... well, boredom?

So let's see... we have a few goals... to write it better, I hope:

Toon skills shouldn't limit the player's skill.
Progression and accomplishment still need to be in place.
Synergy between player skill and toon ability should never be broken.

Let's look at an FPS.
Most people want their character to respond as fast as possible so that it's their own skill that gets them killed, right? Well... the thing is, there's a bit of leeway even in this. Look at the difference in speed between UT2k3/4 and say... Half-Life(CS) (hell, look at the first UT). The speed difference is staggering. So, I'd say as long as speed isn't annoying and it's almost always fair, people would enjoy it, no? There's also the reaction time difference between a keyboard and mouse compared to a console game (Halo). If it's taken into account, apparently you can still find a sweet spot for FPS fans. ~.^

The problem comes in when you try and pump in RPG aspects into that situation. If stats start speeding up a player by any significant amount, the lower people will almost always die in combat, and that's no fun. Hitpoints, armor, and weaponry (think itemization) are arguable, as in almost every FPS, (save Halo, that comes to mind) there is a hitpoint/armor disparity. Weapons... there's always a score of choices. ^.~
So stats come down to, if they make a n00b a gimp, and a pro... a pro, people will find it very annoying, no?

How about a four way tag-team leveling scheme?
1. Personal player skill.
2. Avatar stats. -lvl dependent
3. Weaponry and itemization. -lvl dependent
4. Avatar skills (usable skills and 'learning') -lvl dependent

1. Personal skill is obvious, so I'm not going to say much about it. ^.~
The interaction of the rest goes as follows:
2. Avatar stats determine how heavy of gear you can carry, how big of a gun you can reliably fire, the recoil of such, and the speed at which you can move around. (with weight taken into the equation)
3. Weapons and itemization are the player's to decide. However, their avatar stats will determine how well they can use the big guns and armors. What your avatar knows also comes into play... As in, if your avatar isn't smart, good luck using that super tech rifle!
4. Avatar skills and knowledge will develop as the character goes up the levels, etc. Skills should run the gamut from basic, every day things to specialized utility coming along with that special knowledge.

It should be obviously set up so that as your character progresses and gets better equipment, given they stick to a similar progress in effective weaponry, they will react almost identically 'speed' wise. This way enables characters to start off being just as fast (on average) as the big boys, but not as effective. Obviously a problem, but if you come across a high level, you can always run... This also enables a player, as their character develops, and they start to pick up on what they like and want to do, they can start to build their character towards what they want to do. In the end, a player that is rediculously fast can build a rediculously fast character, but they'd have to have semi-weaker weapons and equipment, 'balancing' it out with slower players. This can obviously be tweaked to any extreme, making the a truly 'fast' avatar easy to kill, or make it come truly down to player skill. (the other dude would have to be good at hitting that moving target)
Usable skills and education come along as the player develops and finds out what they like, as well. These are very specific, so hard to discuss in the big picture... but the general idea should be they assist the person's play style, giving them extra options in the field as they really learn how they like to play. Knowledge also opening new bits of specialized weaponry in the case the player really wants to go that route.

So overall, it turns out avatars develop as the player finds out what they like, and build towards it. Player skill is preserved so a lower level, skilled player still is capable of standing up to a high level, lesser skilled player.

*cough*Wolfenstien: Enemy Territory*cough*