Personally I think that the "separate games with common avatar and spanning community" is a better way to go.
Side note - A couple other players also tried that:
Blue Mars had better technology and 3d graphics, and ability to make some pretty impressive games, but it never really caught on, mostly because the virtual world itself never took off and retained the players.
Raph Koster's Metaplace tried that approach on the web, building games into an avatar community, and players also didn't really warm to it.
Any new game will require you to set up a user account, download or install something, learn the controls, and so forth. A game inside of Second Life would decrease each of these problems if not remove them entirely. That's the potential.
Yes, it has been tried multiple times (in addition to the ones you mentioned, I founded a startup called Syntensity in that field), and always failed.
I think the problem is gaining initial momentum. Second Life had that, no one else did. But Second Life squandered it.
Both StarCrafts and WarCraft 3 had excellent custom game/scenario engines. Three leading game genres, MOBA, Tower Defense and Tug of War, got started as scenarios for these games.
There is a lot of demand for a place to play games, especially if the tools are right.
You described Club Penguin: http://mashable.com/2011/12/13/club-penguin-disney/
I'm not sure the common avatar is a prequisite, though. Steam doesn't have one, but Valve tries to build its social aspects into the games as deep as possible.
> Great technology doesn’t mean a great user experience.
Of course, but IMO the real missed opportunity was that the one virtual world with real traction - Second Life - just made most kinds of games impossible because of technological limitations. The client-server model they chose made it impossible to play responsive games in Second Life.
Imagine if it were possible to enter a SL region and jump right into a platform game or an FPS. Yes, virtual worlds are not games, but they could have incorporated and enabled games.
Instead, Second Life focused on showing how much it was "not a game", and never even tried to address the technological limitations that prevented games from running on it.
I agree with the post on the reasons virtual worlds failed. But it could easily have been otherwise - allowing games would have grown the entire virtual worlds market.