Now that DxOmark has "rated" the Pentax K-5 and Nikon D7000, the nutty interpretations and spirited discussions have begun in earnest. But let me introduce a completely different idea: I don't care what DxO's number is. It's meaningless to me, and probably should be meaningless to you.
No one--not even DxO--has ever given me a seriously considered argument of how their single number summary applies to my photography. I regularly use cameras that have numbers ranging from a low of 47 to a high of 88 on the DxOmark scale. Do I care? Not really. Are my photos with the camera that scores a 47 terrible and my photos with the camera that scores an 88 great? No, it doesn't work that way. Not even close.
DxOmark's numbering scale is 100% arbitrary. You could use the same underlying data to come up with another scale, which would produce "different" results (i.e. cameras would move relative to one another). But worse, some of the underlying data itself is open to debate. The "dynamic range" (DR) test, for example, I believe defines DR as full well to SN Ratio=1, all derived from a single step wedge test. This doesn't reflect how I use a camera and says nothing about the character of the noise itself.
Don't get me wrong. I've long said that every serious photographer should measure his camera, understand how it produces images, and come up with a strategy for optimizing their photography using that camera. What I fail to see is how the DxOmark overall number gives me any useful information or method to do just that. Some (and I emphasize some) of their underlying measurements are indeed useful in understanding how a camera captures data from a scene, but you still need to understand what those tests are measuring and how they might apply to your photography.