The best comment I've heard about Google Wave so far: "Wave is simultaneously less and more than I'd hoped." Me, I already have plans on the table to set up my own Wave server: my personal "Wave 1.1", as it were. Or even Wave 2.0.
The comments about Wave so far seem to fall into three buckets: 1) it's rather slow (probably due to so many people pounding on it at once right now); 2) it's not meant to substitute for services like Twitter or Facebook, but to aggregate conversations you might have in various places; 3) you need to use it correctly to keep it from turning into a productivity-killer. And 4): it's not just going to be a Google thing, because in theory anyone can set up a Wave server.
But I do think most of the really intriguing and powerful work to be done with Wave isn't going to come from Google.
In a way, it can't. Because Wave is so difficult for many people to wrap their heads around (it's been defined mostly by what it's not, from what I've seen), it needs to have some distance from its own creators. It needs to be used in ways not originally imagined by Google, to be kicked around and re-implemented and re-re-implemented. Random example: a PHP or Python implementation of the protocol, rather than Java, which would make it easier for Joe Domain-owner to set one up on his server.