RE: Doom3 benchmarks, I wouldn't trust any benchmarks until Doom3 gets released. Even then, they'll probably patch it into oblivion with miscellaneous video card tweaks.
As to the question of fairness, is it fair for ATI to use one-and-only-one mode that is faster but uglier than nVidia's best visual mode? I don't think ATI was trying to specifically beat nVidia this way, but now that we're here and there's no true apples-to-apples comparison possible, why should nVidia use their slowest (yet prettiest) mode in benchmarks where visual quality takes a backdoor to raw speed? Now that I know what this is about, I say this is not a big deal, nVidia is 100% correct, ATI is neither wrong nor right, and benchmarks are useless unless you can disable the shading crap entirely and still get a useful score.