You can't get the second life crowd to play your game by making annother second life (a third life, if you will). The second lifers are already playing second life!
Instead, you have to make a game that is unique and strong on its own merits. At that point it makes sense to advertise it in second-life-type games, yes.
Furthermore, the engine is not well suited to a social game. I'm sorry, but its primary features simply don't mesh well with it. There's nothing you can do with this engine for that game that you can't do just as well with a much lighter engine.
Now, I would add that any kind of action game by default absolutely must include vehicle simulation; the world is so large that it would be crippling to lock the player to only the places that they get sent in missions.
Train simulation: well... let's see... you've got a map the size of a planet, and you only want to look at the parts of it that are next to train tracks?
Flight Simulation: Now this makes sense! Particularly, it would be VTOL-type aircraft primarilly, to increase the number of areas the player can land to investigate closely and fully make use of the grass terrain generator.
Surface vehicles: Cars, etc. This is basically the same as a foot-based action game at this point. It's married inseperably to the military simulator aspect.
Fantasy/Midievil game: Not really practical. Either you'd just be copy-pasting the same two dozen castles and muddy villages around the world (even in, say, south america), or you have to hire several hundred artists to come up with a set of midievil mgical cultures with different architecture, clothing, and plots.
Spaceflight games: See Flight Sim.
Virtual Worlds: I'm not entirely sure what this means. Perhaps something like Minecraft? In which case, minecraft probably does a better job. Much the same problem as the social game paradox.
Movies: ...Well, maybe. The biggest problem is you'd have to be able to switch to a "render" mode which makes full-quality frames every frame, with no skipping, all the bloom filters... and you still probably won't look quite as good as properly ray-traced renders. How well would you engine support things like faces and animations?
Strategy: One thing that this engine is ideal for. The largest obstacles I forsee is that a modern air unit might pass through a region in seconds, stressing the rate at which chunks can be loaded, and more troubling, any kind of modern force could have hundreds or thousands of units, which might end up splitting up, meaning that hundreds of separate chunks need to be loaded simultaniously. If those problems can be overcome, perhaps by having a far simpler terrain map for places where only non-player units are located, then this would be a very unique and very impressive game.
Physics Sim: Sounds difficult but not impossible. You'd want to have a very good collision detection system... I imagine building vehicles of all types would be a major part of that game. As I've written elsewhere, a modular, modable garry's mod-type game would rock.
Citybuilding: As we've already seen, building cities is very well suited for this game. A good economic simulation would be necissary, but aside from a few of the traveling units problem that a strategy game would have, this sounds good.
Ship/Submarine: See Surface Vehicles.
Poll
243 (20.5%)