All that said, I believe we're entering the long-tail phase. Linux has stable, use-able desktop distributions, dozens of them. We've got that down-pat. As far as I'm concerned, there are few areas that really need a lot of work. What comes next is honing and refining what we've already achieved to finally fulfil the needs of those users who are holding out. Often you see around here people claiming to dual-boot for the benefit of just one or two applications or peripherals. That's the last thing Linux has left to conquer.
But on the sidelines of the main push to make Linux as good as it can be, you do see some pretty out-there innovations. The Metisse Window system is one of my favourites, particularly, I think the facade system could really revolutionise the way we think about the user interface. Have a look at the sort of thing it can do (the animated gif stops after the first play, so hit ctrl-f5 to see it again if you miss it):
And then of course there is the Google Wave stuff, which I hope really does shake things up. It's something I can see completely changing everything, so long as the developers get it right. I hope they do.
All in all, I think innovation is happening, just that it's mostly happening on the fringes of the major projects. This isn't a bad thing, as eventually the things that are really cool will find themselves filtering down to the main show. And then every now and then, you get something incredible like Wave or Wikipedia coming down on everybody like a tonne of bricks. I don't think they'd be quite as awesome if we had that sort of insane innovation every day.