Yep and that's good. The problem with designing code around the FOSS drivers is that, right now, you have to give up about 8 years of advancements in graphics. Even though I hate writing blob-specific code, the alternative is no GLSL, no FBOs, no PBOs - it's just not feasible! (especially so in Wine)
Granted, I'm talking from the perspective of high-end graphics. If all you want to do is write a 2d game or a bzflag clone, I guess it's fine (not great, but iyou can make it work).
But once the memory managers and the new driver interfaces hit the streets there will be little reason to use the blobs if you're a desktop user (well, maybe if you got nvidia ). All I'm saying is that the end goal should be to start developing with the open source drivers in mind as a priority.
Usually someone leaks them on the "dark" places. IRC and Usenet for example. Google doesn't find them, you have to know where to look. But it's nothing to go crazy about. If you find them, the result is simply a disappointment that comes earlier than if you had waited for the final version