If I recall correctly from a conversation in the office a few years back with someone who'd looked into the issue, there are two problems caused by the pixel sensors getting smaller.
The first is that fewer and fewer photons can hit the sensor so that the recorded values are inherently becoming noisier and noisier. Related to this is that each pixel 'bucket' can only hold so much 'charge' and so you lose dynamic range. As soon as some % of pixels become fully exposed, the picture has to be finished.
The manufacturers, of course, don't want to spend silicon on a bigger sensor but appear to be locked in an almost senseless resolution arms race. <shrug>
Of course, take all of the above with a grain of salt.