There are some interesting facts that you can check out about the 1DsIII and the 5D. They have been mentioned on here in numerous threads and there has been no discussion I have seen about why the data are where they are.
I am speakng about the fact that if you hand hold the 1DsIII in normal shutter mode at 1/125 of a second and shoot a scene and then shoot that same scene with the 5D, you will not see any difference in resolution. I even put both cameras on my Manfroto 3251, which is a substantial tripod, and in normal shutter operation mode, I get the same result. The two cameras resolve at the same level. There are differences in color and even some dynamic range differences but resolution comes out the same.
If I put the 1DsIII on a tripod and shoot in mirror lock up mode, then I can see the difference in resolution. However, I can hand hold the 5D down to 1/60 of a second with a 50mm F1.4 Canon lens and at that shutter speed, hand holding with my perhaps shaky grip, the 5D out resolves the 1DsIII.
Of course what is happening here is camera vibration/camera movement. Moreover the two cameras are not shaped the same and therefore I will not be able to hold each camera the same way but I don't think that is what is going on.
If the proposal I am going to make has been already discussed, please direct me to the reference. It has been on my mind recently and I have not seen anything written about it.
I think that when the pixel size gets to a specific threshold size, camera movement during exposure(vibration from the shutter is a specialized form of camera movement) results in a dynamic displacement of the image that is hitting a specific pixel. With digital that pixel cannot record line acuity, that only happens when an adjacent pixel has a different color, darkness etc. One specific pixel reports one and only one set of information and if that pixel is moving, say across a line edge during exosure, it will take less movement to mess up the signal from one pixel if that pixel is smaller. It seems logical to me that smaller sensors are more succeptable to blurr from camera movement than sensors with larger pixels.
I just re-read what I wrote and although I didn't state it eloquently, I think the fundamental though has been captured and I think it is worth discussion. To put it another way, camera containing the smaller pixels will require less overall camera vibration to mess up the areal image than a camera that has larger pixels.