Anyway Games has some good thoughts up (I especially like that continuum in the title between "testing" and "fun") about what beta tests are really for these days. It seems you can't publish an MMO without having a beta test-- the world is so big and meant to support so many players that you not only need volunteer help for QA, but you have to have enough people to stress the servers, just in case.
But is it OK to invite players into a game that's not actually done yet? Richard Garriott blames TR's slow start on a poor beta reception, and I was one of those players who wasn't impressed with the early beta (although I don't know if the game has actually improved since then). I also have been playing the Pirates beta since a few weeks ago, and after playing it again yesterday afternoon, I was pretty astounded at how far the game has come just within a week or two of development. There is no question that the game I was playing a while ago was definitely unfinished compared to the game as it is now, and even though it's in open beta, there will undoubtedly be improvements before the game goes live.
So what's the purpose of a beta? Is it marketing for the game, or a massively multiplayer quality assurance session? Every developer has to decide for themselves, obviously-- there's no golden point at which the game is good enough to be played, but buggy enough to still be tested by the unwashed volunteers. The best you can ask for, at this point, is a development schedule that gives you a game strong enough to give a great experience while buggy, and a beta testing crowd interested enough to stick around and help you fix the bugs that are left.