There are three dominant aspects to virtual worlds: protocol, world, and game.
The protocol aspect is important to keep the virtual world open ended. The protocol is the bearer of worlds.
The protocol aspect sets the basis for flow, empowers the user by providing new and diverse ways to express him/herself, and enables the virtual world to act as an extension of consensus reality.
Virtual worlds where the protocol aspect is dominant are a chat room at worst and a sophisticated and integrated set of tools at best.
If we push the protocol aspect out 'downward' from the virtual world, we get "build your own virtual world" software technology.
If all a world has is empty space, then it is not much of a world. To claim that title, it has to have a wholeness, a flow of relations, an energy dynamic within / as itself. The world not only has to have a stand-alone quality, but the ability to relate 'outside' of itself.
With enough complexity, it wouldn't be too wrong to think of it as a living, breathing entity: a dividual individual.
The world is the bearer of games (relations).
A virtual world which emphasizes the world aspect at the expense of the protocol aspect is one which we find it hard to operate in.
A virtual world which emphasizes the world aspect at the expense of the game aspect is one which we find boring.
For a virtual world to be successful, it has to support many games (many ways to play a game).
For the user, the goal is to play the game (develop/manage/balance relations, relationships, presence).
A successful game can uplift the virtual world; give it the massive attention it needs in order to be more persistent, more real and allow it to be more alive. (Which, in turn, would empower other games based in this virtual world.)
A virtual world that emphasizes the game aspect at the expense of the protocol and world aspects may become too abstract to relate to.
(That's why we have the remnants of the real-or-virtual debate: software games progressed from more abstract (game) to more real (world), and it can be hard to keep up with the jump.)
Well, we need all of them.
Each of these three aspects (protocol, world, game) have a stand-alone quality to them. As virtual worlds become more complex, I believe these three aspects will be among the first to fall into separate development categories.