I really don't get it. As an end user i pay for the best possible picture. So if i get 40 MP resolution with the Canon-Sigma combination and 37 with a Nikon-Sigma combination (assuming Vignetting etc. is comparable) i would expect a higher rating for the Canon-Sigma combination.
How the reolution is achieved is secondary. If Canon uses a radically "overpixeled" Sensor or Nikon uses a better geometry (or whatever) is part of their strategy. In my personal opinion the Dxomark rating is misleading in this way. If i search for the best possible combination i have to go down to place 9?! If i compare it with CPU benchmarks i'am also not interested, how a good performance is achieved (MHz or architecture), in a Single or Multi-core benchmark i'am interested in the absolute performance.
I know this subject is a matter of opinion but i understand when people are confused about it.