So crash course in RPG theory: the basic trichotomy that gets tossed around is the Gamist-Narrativist-Simulationist, which in its extreme cases works like this:
Gamist thought tends to treat RPGs as though they were games like Go or Chess - that is, there's a focus on making the game mechanically balanced, and the point of the game is for the players to win - to beat the DM. Games like old-school D&D and I think to some extent Paranoia tend toward a gamist perspective - and I have to say: that's fine, and that can certainly be fun, but it's not intuitive to me - which is interesting, as I am extremely competetive with regard to board games. RPGs just don't hit those same switches with me, I suppose. Scorpio has a heavily gamist bent, and that has led to some friction in the past. (Jack, near as I can tell, doesn't really have any particular bent, but tends in practice to fit somewhere between narrativist and gamist, probably weighted toward narrativist.)
Narrativist thought views the players as providing the protagonists for a story, the rest of which is written by the GM. Extremely narrativist games, like Polaris or Universalis, dispense with the GM entirely, instead relying on things like ritualistic interplayer negotiation (for Polaris) or an interplayer economy (for Universalis) to determine things about the game world and resolve conflicts. I'm much more in sympathy with the narrativist perspective, obviously, but even though I have no problem viewing my own life, human life in general, and indeed the life of humanity as a whole as a fiction - and in fact probably couldn't view it any other way if I tried - I am not entirely satisfied with narrativist gaming.
Simulationist thought tries to model the game world as accurately and realistically as possible. It tends toward the complex rules systems - Shadowrun, for instance, which can fill several books with mostly rules and tables of vehicle stats, or D&D 3.5, the Dungeon Master's guide for which is very possibly the most stultifyingly boring RPG book I have ever read - and the point is that a realistically modeled world is more immersive - which is a relatively difficult quality to achieve when what you're looking at is the people you're in college with, a bunch of funny-shaped dice, and a Word document with a lot of numbers on it. And here's a weird thing: perhaps because worldbuilding is one of my favorite aspects of fiction, I am at heart a zealous simulationist. I know that byzantinely complex rulesets are no fun to play with. I understand that there is relatively little difference between rolling to throw a grenade, rolling to determine scatter, rolling the target's dodge, and rolling the target's damage soak, and and the GM simply interpreting the results of one die roll, except that the first one is a bitch to do and will make combat unreasonably long. But, dammit, it feels different. And of course there's no way so far as I can tell to have quick, clean combat resolution and accurately modeled conflicts. The rules hack for Shadowrun I and my players are working on is supposed to contain various levels of granularity, to allow you to make that trade-off on a case-by-case basis, but that's a makeshift solution to an apparently insoluble problem.
But that's only a problem because we have a limited amount of processing power, yeah? Given the ability to easily remember a very large number of rules and the ability to perform complex calculations very quickly, simulationist games really come into their own. More than that - we can get more complex than that - truly three-dimensional space combat, computer hacking that involves actually finding holes in the programs and systems the game has constructed, magic duels with ebbing and flowing mana, attacks and counterattacks on your opponent's qi, fights won by wits and swordsmanship rather than by dice rolls (which, just between you and me, has always seemed like an unsatisfactory conflict resolution mechanic - though comparing static numbers is worse). And since we're operating theoretically - the things we could do with unlimited processing power! And so on until, like the map of Borges' cartographers, our simulation is precisely as complex and as intuitive as the universe itself. That, to my mind, is the perfect game.
Turns out, we can't do that. So near as I can figure, the difference between an actual simulationist and an actual gamist or narrativist is that the simulationist probably sighs and looks up at the stars, half-imagining what cannot be, before going and playing a gamist or narrativist game.
Addendum: that perspective on games certainly opens one up to accusations of escapism. That's valid insofar as games are actually used as an escape from the burden of actually doing something with one's life. And this runs into the problem of me being a very bad existentialist. See, I think that a person's life is inherently meaningless - that the meaning of one's life is self-determined. I then want to turn around and tell some people that the meaning they seem to be deriving from their life doesn't count - people who work their way up the company ladder and then retire to a country home and a yacht, people who work in a factory nine-to-five and drink beer with their friends on the weekends in front of the TV - in fact, my (totally unreasonable) instinct is to exclude politicians, lawyers, doctors, and teachers from the category of people who have meaningful lives, leaving only scientists, artists, and people like Joshua Abraham Norton I, Emperor of these United States and Lord Protector of Mexico, whose lives are in themselves such interesting narratives that it counts as production. Of course, this is only my own personal existential choice - these are the lives that I personally would find meaningful.
So, back to the point, which is: I don't think playing games is sufficient for a meaningful life - it certainly is not for me, and I want to say (but can't) that it isn't for anyone. However, I do think that it can be part of one - whether I end up doing the art thing or the life-as-narrative thing (scientist being, unfortunately, not really one of my options), that games can add to it. I learn a lot about storytelling, among other things, from playing these games, and that is valuable. It's just not valuable enough, on its own. (I should reread Nabokov's The Defense, which if I remember correctly is on more or less the same subject, but with chess instead of Shadowrun.)
Further addendum: it's pretty clear that many of my positions are philosophically incoherent. I'm okay with that - in fact, it's somewhat deliberate. The line of reasoning is this: I am never going to be right about everything. If I attempt to construct a coherent philosophical weltanshauung, any error in the initial premises is like to be amplified until I find myself believing unacceptable things once I get down to particulars. Besides, I have a certain amount of faith in human instinct - not in any mystical way, but simply that our ethics, our art, and our drive to know are all products of pretty nonrational impulses, which means we already have pretty okay heuristics for a lot of things. Not, obviously, unimprovable, and I'm certainly not rejecting reason - I can be logically convinced that some of my heuristics are bad and should be changed - but by and large, I think it's pretty reasonable to go with one's initial reactions on a lot of things.