The closest comfortable viewing distance for a print is about equal to its diagonal, I believe. Held normally at that distance it subtends a little under 50 degrees horizontally at the eye. Someone with good vision has an angular resolution of about 1 arc minute (any details separated by less than that are indistinguishable). Allowing one pixel per arc minute and the usual 3:2 width to height ratio, that makes a bit less than 6 MP in total. So for normal purposes (excluding certain specialised applications) what is the point of 15, 18, 20 MP? I'm not saying that there isn't any benefit, but I wonder whether more could be gained for the most common applications by applying the technological improvements that have gone into these dense sensors to something of lower resolution, and getting better sensitivity, dynamic range and colour (and smaller files). I guess that this has been discussed plenty of times before, but I'm curious as to where the tradeoffs are, especially as some of you have experience with this range of resolutions.
Another thing that occurs to me is that diffraction limiting kicks in at wider apertures with denser sensors. The diffraction limited linear resolution (smaller is better) is proportional to the focal length and inversely proportional to the aperture diameter (i.e. it is proportional to the F-number, which is another application of that useful quantity). So in order to take full advantage of a denser sensor you need a wider relative aperture. This would mean that applications that require a large depth of field, and therefore a small aperture, like landscape and high magnification (macro) benefit the least from very high sensor resolution, even if essentially perfect optics were available. I'm not saying that higher resolution makes things worse in this respect in the final output, but beyond a certain point it can't make things any better, and there might be other drawbacks. Even with a 10 MP APS-C sensor diffraction starts to have an effect at not a lot narrower than F8.