Well, my suburban Chicago roads are anything but smooth, especially ITTET. I think that the trick to mirror use is to glance at the image, but not try to study it for detail. I glance at my Mirrycle road mirror (sti mounted) and I know exactly what is behind me. A glance tells me how many cars behind me, what color they are, how big/what kind of vehicle they are, how far they are, and about how fast they are going. Perhaps my brain just fills in the rest and gives the illusion of detail. But the only time this doesn't work is crossing railroad tracks, then the image is uselsess. Solution = look at the mirror before I hit the tracks. If I try to identify the driver's face, license plate, sometimes vehicle manufacturer badge, I can't see squat. But the mirror otherwise works GREAT! In fact, I think it works better than looking over my shoulder for all areas except for the area right behind me and up close. So a look over the shoulder is best when contemplating a move in the kill zone. But for objects further away, things that would be in the periphery of an over the shoulder glance, viewing with my more central vision via the mirror works way better for me.

edit: my experience reminds me of the astronomer's dilemma/phenomenon. When looking through a telescope at a maginified image, more detail can be discerned by not looking directly at the object of interest, but slightly off of it (ie: not your peripheral vision, not dead center either). Try to see the detail and it evaporates, try not to look directly at it and the detail pops out.