Michael is dead-on with his assessment in this essay, but is perhaps a bit too kind to DxO.
Let me "put the cat amongst the pigeons" as much as DxO did with their MF chart, by asserting fairly baldly that their camera-measuring site and numbers are as close to useless nonsense as we need bother going. I'm not saying their data is wrong or that their intentions are anything but good, far from it.
While I understand, and even struggle with myself, the measurbatory instinct which drives DxO Mark, and so much of the camera-chatter online, they are producing a template for discussion which is so fatally divorced from the creative process that it is functionally valueless.
There is one, and only one, meaningful measure of any digital camera's quality, and that is how the finished product looks in its finished form. (Indeed, that form will itself in many instances alter the outcome of the evaluation).
And that's the rub. A properly exposed (to the right) file looks awful out of the camera most of the time. The real road to finished image quality leads through the issue is how that data 'stretches' (my word again) over the tonal range in post. The real finished quality can only be truly assessed in the finished product, which requires the extensive intercession of subjective human choices of creation, which cannot be quantified.
Files from certain cameras are vastly 'deeper' (my word) in the sense that, with equivalent exposures, they offer far greater maleability of the image in post-processing. From recent experience in Antartica, I can say with certainty that files from a Phase back render tonal subtlety and richness at a level unachievable any 35mm dslr I have ever used. That same relationship appears to translate downwards, viz-a-vis even superior point and shoot cameras. This is likely due to bit-depth, but the actual effect of it on finished image quality is only visible after the craft *not science* of post-processing has been applied, subjectively, by the artist in question.
The DxO model is somewhat akin to rating paintings based on a chemical analysis of the paints used. Interesting, but just not that useful in any serious conversation about aesthetics. If we all photographed test targets for a living/hobby, such measures would mean more. If we are trying to use technology to acheive self-expression, the serve little purpose at all.