If you guys think this is ugly now, it's only going to get worse after Cayman launches.
Barts already has very good tesselation at lower factors, although still not quite as good as Nvidia. Barts, like Cypress before it, still falls apart at higher tesselation factors though, and this is where Nvidia shines.
With Cayman, instead of Nvidia being faster at tesselation in general, what will probably happen is that Cayman will actually be able to beat Fermi in tesselation, at least at lower factos, and perhaps significantly so. Meanwhile, assuming Cayman shares the same drop off as Barts and Cayman, it's safe to assume that Fermi will still rule the roost (by a staggering amount) when it comes to high-to-insane tesselation factors.
Obviously, both companies are going to start pushing for the scenarios that make their cards look good and their competitor's cards bad. Nvidia obviously has more clout with developers, but at the same time I think that insane tesselation factors are an absolute waste of performance for something that isn't even discernable to the human eye without a wireframe view, so I can see a lot of developers going the other way.