I don't understand why 'distortion' (things being rendered ± 1 pixel over on some screens vs. on others) is a problem unless your GUI frameworks and/or apps count on pixel-exact layouts for some UI elements. But why would they? Isn't the entire web built on a reflowable format that works pretty well? Shouldn't those tiny 1-pixel differences be like the easiest possible variation for a GUI system's layout engine to cope with?

Do we have lots of scalable UI elements that expect to line up with raster images a certain way on most operating systems?