There is a bit of a movement towards player generated content. One undercurrent is the idea of leaving almost everything in the hands of the player. The biggest example of this is Secondlife, but there are others. Eve Online and Dwarf Fortress are NOT examples of this.
It has become clear that there are some holes in the "everything by the player" philosophy, but it is also clear that the "everything by the designer" philosophy hasn't got much juice left in it. So the question is not "is player content a good idea" but "how do we make the most of player content".
Let's talk about foundations.
In Secondlife of old, you were pretty much allowed to build whatever. You used fundamental shapes, colored them, textured them, scripted them... you could build anything from wearable wings to huge castles to automated war robots. However, the barrier was high. Higher than most people think.
Creating content was difficult, sure, but more than that, there was no structure on which to hang it.
This lack of structure results in a lack of feedback. Aside from the commentary of other players, there is little feedback anywhere at any point in the game. Now, player feedback is important. But player feedback as the only feedback is like a ship with no rudder. Yeah, it'll get somewhere, but not anywhere specific.
More accurately, yeah, it'll get somewhere: porn.
When there is no rudder to judge content against, players will judge it based on their own personal opinions. Since they have no particular unmet needs or dangers, those will basically be their luxury opinions, which mostly means porn.
There are other kinds of content in Secondlife. There are motorcycle races and casinos and all of that stuff. But they operate in one of two ways. Either they manufacture a rudder (motorcycle races have no fundamental rewards or penalties) or they tenaciously hang onto one of the game's incidental rudders (money).
In fact, even the pornographic content grips a rudder the player brought with him. All gameplay is judged against some kind of dynamic, some kind of fitness system... a rudder. And if you don't provide one, they'll have to create one.
Players will create them anyway, sure, but even allowing that you may WANT porn to be a driving factor in your game, player-created rudders are inefficient in most cases. It's far more effective to provide a rudder that actually points the player in an interesting direction, helps him accomplish his goals in an interesting way.
An example of this is Eve Online, which is also driven by player-generated content but in a framework of mad capitalism. Many tools are provided to the players to help them in their rush to Make More Money, and this means that the money-related gameplay in Eve Online is orders of magnitude more interesting than the money-related gameplay in Secondlife.
Eve Online is not really "better" than Secondlife. Secondlife has a much broader spectrum of things you can build, so their casinos are all blingy and bloopy and effective at stealing your cash. But in terms of the gameplay of making money, Eve Online goes a lot further with a lot less.
Dwarf Fortress is another example, not even a multiplayer one. Dwarf Fortress is a roguelike where you sort-of control a large number (often 100+) ascii dwarves in their ascii fortress.
While the game doesn't have any explicit goals, it has a framework of implicit goals built right into the gameplay - much more powerfully than SimCity. These kinds of implicit goals are created out of the gameplay rather than out of any scripting, and they not only provide good rudders for players, but they can actually provide more than one rudder per rudder.
Each of these implicit goals isn't just ONE rudder. For example, the implicit goal of surviving against enemies can be perverted into any number of more colorful goals, such as surviving against enemies while having no military dwarves, or surviving against enemies in a haunted region full of the nastiest things around, or even one of those goals combined with another goal, such as no military dwarves and you're not allowed to cut wood.
These recombinant goals aren't unique to Dwarf Fortress, of course. Almost any game can have them, such as seeing who can fling a Warthog highest into the air using grenades, or creating a useful town in Sim City with no pollution at all.
But these player-driven variations on implicit goals are rarely used in multiplayer games, because when you're going up against other players, the basic rules of the game serve to let you compete against the other guy, rather than serving their own ends. There's nothing wrong with that, but it's important to remember.
Anyhow, I'm wandering. The essence of what I'm trying to say is that the gameplay of a MMOG radically changes the kinds of player-generated content you can have (gameplay including how you allow players to collaborate). If you don't have any gameplay worth mentioning, the players will import their own and go skewing (and screwing) off at random.
But providing rudders - and perhaps a clear way to build new rudders off of them - allows player-generated content to serve a purpose. It makes it possible for even a poor player to make a judgement call about whether something is good or bad, cool or boring. Combined with a gentle touch to allow non-productive players to serve some purpose (such as filtering content or choosing evolving content), I see this as the future of most massively multiplayer game designs.
Now, Spore failed at this. I won't pretend to dictate to you, but if you do happen to think about why Spore failed - and how it could have done it better - I'd love to hear it. And, of course, whatever other educated opinions you might have.