MMORPG environments is something I've happened to spend a lot of time thinking about. The way I see it, the primary shortcoming of the current wave of MMORPGs is that they place a premium on two things, Grind and PVP. Each are emphasized I think because of the perception that they are the things that generate money. Force people to Grind in order to be competitive, and they will spend time, and thus money, doing so. The irony, I think, is that the reason most people actually stick with an MMORPG is because of the social environment, not the grinding or PVP. Those things are incidental to most people. The shortcoming of most MMORPGS thus is that they actually force people to grind simply to participate in the social aspects of the game, an aspect which in theory should require no leveling of any kind (if I want to be a manipulative behind the scenes rumourmonger, why should I be forced to level?)
The way I see it, the only MMORPG that ever embraced true open endedness was Ultima Online, and then only in its earliest days. Ironically, the thing that made the game a success, and that gave it its appeal, was the fact that it was open ended and completely anarchic. All organization was done by groups of players, using only their creativity. Unfortunately, because the game had consequences, for poor play, a lot of people bitched until the game was made mediocre and sterile, and consequently, devoid of excitment.
What made ultima online great was X things. First, it used a skill based system, which limited the end game, and made all players at any level capable of contributing. In the game you capped out, but that was a choice of game design that could have been better executed, such that there would be no true end game, but that advancement would become exponentially difficult.
Second, it was freeform in player interaction, so any player could go anywhere and interact with anyone in any way they chose (early on anyway). This meant a lot of idiots doing stupid things, but it also meant really rich highly developed player based communities that were made with almost no designer side support, and which had their own storylines that evolved in real time according to player actions, with heroes and villians, elected officials, player based economies, and all sorts of other rich roleplaying developments. And all of this happened because it was left up to the players to decide how to play the game.
I emphasize this point because this was the most radical and wonderful thing about UO, and was the thing which was lost within two years because the designers were afraid of the consequences of open ended play. But in the end, their efforts to curb the supposed "downsides" of open ended play, really just limited all the players, and drove away the very people that were intially attracted, and very loyal to the game they had started.
What I learned from that game, and what I have yet to see any game duplicate, is that if you give the players the tools to organize the world, you dont need to do it yourself. And in the end, the world they create will be more dynamic, more interesting and more rewarding than anything a limited group of game designers could ever conceive of. Therefore, if one wants to set out to make a game which is both unique, and which caters to a roleplaying mindset, the way to do it is to create an environment, and then to give the players as many tools as possible to shape that environment in their image. Sterile directed grinding leveler games are yesterdays news. I think if you want to make a game that is truly groudbreaking, you need to be willing to do something different. At first you may have chaos, but given enough time, out of chaos will come order.