Again I fully agree with Syncaine in that it is a lot easier to make a successful PvE theme park game than to make a successful PvP sandbox game. The market for PvE theme park games is obviously much bigger, and even a half-decent WoW clone like Rift can get 1 million customers (if not subscribers). Having said that, I think that Syncaine is wrong in attributing the difference to PvP vs. PvE. I believe that it is easier to make ANY theme park game, PvE OR PvP, than to make any sandbox game.
For example World of Tanks is up to 5 million players in a game with absolutely no PvE at all. But the PvP it has is very clearly structured, there is a clear narrative telling players where to go and what to do next. Meanwhile the excellent PvE sandbox game A Tale in the Desert has only a few thousand players.
It appears to me that the large majority of players likes to be told what to do, have a clear path of advancement with every step laid out in front of them. They gladly follow the instructions of "do A, then do B, then do C", but get confused if given the free choice of doing either A, B, or C, especially if told that these options aren't equally good. Even in a theme park PvE MMORPG, when given a choice, players hurry to find a guide or website telling them which option is the best. Anybody making a choice out of his own free will and ending up with the option that is 0.07% worse than the optimum is a clueless n00b. If you are forced to always make the absolutely best choice, you simply can't afford to think for yourself. And you have to avoid games that give you too many choices, aka sandbox games.
Sandbox is too hard for the players, thus too hard to make successful for the developers.