Regarding that blog post, I'll just link to Repi's response.
Now. Some prioritization lists from the other side of the fence...
Game development and hardware --- if I get to choose:
1) make it playable on most machines out there
2) make it really awesome on machines with badass hardware
There is (or will be) a minspec for the game. The minspec is to be printed on the game's box. Manufacturing of boxes has long lead times (let's say ~1 month).
Game development and drivers -- if I get to choose:
1) make it work properly on the very-latest drivers
2) detect old drivers, and turn off stuff that doesn't work there
3) develop workarounds which makes all the newfangled stuff work on old drivers as well
Option 1 is easy for games that do tested-and-true stuff. BF3 pushes the envelope and therefore requires quite an effort, both from us and from the HW/driver people. You most likely will need the latest drivers (at launch day) if you want all the bells & whistles. One of the guiding principles is that a game should never crash. It may run with features disabled, or refuse to run (and explain why) but it should never crash. Crashes are bad because the user is then unable to play the game AND the user doesn't get any information as to why he/she is unable to play the game.