Okay, I have to chime in on this one. (Though, of course, I could just be deluded by what my bosses at the Census office are telling me, and they in turn could be deluded by what their bosses are telling them, and on and on…)
This spring, I and many others, with the help of hand-held computers, went door-to-door to locate residential units. (We also had a list of addresses which we used to confirm whether a place was a residence, some kind of group residence, i.e, group home, nursing home, prison, etc., or a non-residence.) We did not ask who lived at a particular place, nor how many people lived there. All we did was use the computer to put a dot on a map so that next year the head count is done more quickly and efficiently.
Right now, (at least in Montana and northern Wyoming) we aren’t using the computers, we’re using paper maps with those previously-gathered mapspots on them in order to correctly identify what type of group housing different places are. I believe that next year when they do the count, different forms get sent out, depending on the type of place. Again, we’re still not asking WHO lives at a place, nor exactly how many DO live there, just how many COULD. So the whole idea of using this information to target a specific person’s house is ridiculous. The whole idea of the Census is (as Acci said) to find out where people live, and how many people there are living here, so that we can determine the number and location of representatives, not just at the Federal level, but also locally, as well as funding for various programs. Using the computers makes it easier and faster to locate residences, etc.—they even include caves, under bridges, anywhere people live, because they want to count EVERYONE.
I’m not sure how much of that makes sense. It’s just too darn early, and I wouldn’t even be awake yet if I hadn’t had to be on a conference call earlier. 
