It should also be pointed out that this "generational" model that MS sticks with for DirectX is a very good marketing tool. It certainly doesn't favour the developer, but it does make graphics hardware and games more easy to sell as something as simple as a new version of a software library can suddenly become a feature that can be used as a bullet point on a box and be given a nice flashy logo...
OpenGL for example, more or less keeps up with Direct3D (at least on features not necessarily on performance of implementation) with perhaps a 6-12 month lag, but because of their more standard numerical naming scheme (i.e. OpenGL 2.0, 2.1, 3.0 etc.) People pay less attention and assume the differences between the versions are minimal.
In reality the same can be said of DX10 to DX11, the added features for DX11 will be minimal, with tessellation support simply adding easier ways to do something that a good dev using DX10 could already do, lets not forget that while DX11 seems like the new magical step, DX9 is still being upgraded, they just retard its features to make DX10 and DX11 disparate products.
Being honest, I don't think its possible to get a really good insight into the whole thing unless you work with the libraries in question. If anything, from my point of view, this very stepped upgrade model of DX is a bit annoying. I can write OpenGL code and typically my code that worked with OpenGL 2.0 will still work, I just have access to more new functions with OpenGL 3.0+. The same can't be said of DX code, quite often they simply change the name of an interface within the code from "Direct3DX to Direct3DX+1" and i'd be *very* surprised if the underlying code in the new libraries has altered... but I have to go through and make changes to all of my code-base to make it work in exactly the same way with the Direct3DX+1 library...
I just wish OpenGL could get more of a foothold. But as long as it isn't commercially viable, it will remain a research tool and under-developed... and as long as it is a research tool, it is unlikely to be commercially viable...
Here's hoping Steam on the Mac has an effect! Imagine games written under Windows that can be ported to any platform without having to re-design the game engine itself... Carmack had the right idea from the start, it's just a shame the might of big industry has stunted what should have been the ultimate 3D library solution.