Don't be silly -- you can't compare pictures taken by different people under different circumstances and draw any conclusions. The DxO Mark is a far more reliable measure than some random pictures from the web taken by random people.
And the side-by-side comparison pics you posted are from a review the author essentially admitted was done with different levels of noise reduction.
I would absolutely love to test both of them myself (and maybe when I get a 7D, I'll rent some Nikons and do just that).
Ideally, I would like to see shots taken at the same time, same conditions with the 70-200 2.8 IS II on the Canon and 70-200 2.8 VRII for the Nikon (or some other set of comperable, high-end lenses). It would eliminate pretty much all outside variables and provide a real world set of visual examples. DxO mark does not take into account resolution or lenses in any way, which (in addition to post production) are HUGE factors in creating an image. The site really only exists for e-peen waiving, IMO. But maybe if you take pictures without a lens, your eye can see 1/10 of a stop difference in dynamic range, and resolution doesn't matter, then DxO Mark would be your place to be. :thumbup:
Unfortinately, there seems to be no such comprehensive mass database of image comparisons, and my personal budget doesnt warrent me to undertake such a project myself. Cameralabs has some good sets though.