What kind of system would you envision and would you even want to see your favorite games continue to evolve?
Basically, what you're asking for is a way to improve your return on investment (making add-ons/expansions/etc is more economical than making a new game) while providing some benefit to the consumer. Let us call this "revenue enhancement".
Personally, I'm not a big fan of this kind of thing. There are many different kinds of games, and precious few are capable of functioning under the GC2-model of game development.
The GC2-model is functional for the kind of rule-based games that GC2 is. But it isn't even good for all rule-based games. A StarCraft-like RTS game shouldn't be having a lot of updates; if it does, then that suggests that it was horribly balanced to begin with. That kind of game you release, tweak it a bit as adjustments for subtle balance errors, release an expansion to correct gross balance errors, and it either succeeds at being digital Chess or fails. For that kind of game, there simply is no mechanism for revenue enhancement.
The GC2-model eventually has its breaking point for games that are of the proper type. At some point, the ship is as good as the developers can make it. Beyond there, adding more features is just, well, more
, not better
. After a while, the core-root gameplay (the stuff that would only change with a proper sequel) cries out in pain at the weight of all the features built on top of it, and the game is weakened as a result.
For more content-centric games (FPSs, RPGs, action, etc), the only fair kinds of revenue enhancing models you have are episodes.
For content-heavy games, much of the primary cost of development is in assets. Levels, character models, AI, etc. Episodic games work best as revenue enhancement by reusing these assets whenever possible.
The "grossest" version of this is the NWN model. Terrain is built out of mesh-tiles. So if each new episode adds a few new mesh-tiles and bits of window-dressing that can be scattered about a level, it can turn old terrain into something that feels new. However, this requires good level designers and well-built terrain tiles. And if you fail at it, then your new episodes look like rehashed versions of old episodes. This one has the advantage of being pretty fast to produce new levels, so your episodes can be released fairly frequently.
The next version is where you use certain pieces of terrain. Imagine an RPG where you have a large city as your base of operations (with many adventures and possible adventures everywhere), but about half of each episode is spent exploring the countryside and far distant places. In that way, you start with a city-based game and then you expand to unknown territory. It focuses each episode's asset development time on the new territory, which provides the sense of newness to each episode. At the same time, it allows the episodes to be longer and more user-pleasing by reusing terrain (the city). Basically, it's like how most TV programs work; you build a few sets and reuse them forever, while sporadically throwing in a few new sets here or there for newness sake.
This method is advantageous in that the world isn't as tile-based as the NWN model, but it suffers from possible user burnout on the large city. However, the tile-based model would probably wear thin in an action setting, while this model would be much better suited to such things (imagine GTA with mission packs that expand the city, etc). It's how MMO's work to varying degrees.
The last version we'll call the Half-Life 2 "Episodic" model, which isn't. Valve is basically making a new game with the same characters each time. They share characters and weapons, maybe some terrain textures, but lots of asset work needs to be done with each episode. And of course, there's storyline work, dialog recording, etc. It's greatness from a user-value perspective depends entirely on its cost vs gameplay value. The key here is whether the episode's gameplay is worth the money. HL2's episodes don't even bother to try to be economical in their asset use, but a more reasonable developer (one who isn't getting fat off of Steam and thus can't afford to throw money at anything) would be able to reuse more assets.
Ultimately, most game types can provide some mechanism for revenue enhancement. The key being to provide sufficient value to the consumer to make them want to pay for "less" content. Or to put it a less self-serving way, to use less money to provide the user with more enjoyment.
As for user-directed value added stuff, I can't really bring myself to care. I stopped playing with mods after ThreeWave CTF and TF in Quake 1, and I haven't missed anything of value. Most people are not good game designers and do not know how to properly provide a balanced, well-paced experience.
Personally, I don't think games should be extended in this way. Not in the general sense. I prefer that if a game is going to last 10 years, that it be because it is near-perfect like Starcraft, not because there are people poking at the corpse of a good idea ad infinitum. Good books end. Good movies end. Good games end too.