Today's Times has an article about some early uneasiness among Republicans about Democratic intentions regarding the 2010 U.S. Census. At issue in 2000 and today is exactly what methods other than mailed paper forms, visits, phone calls, etc. to employ. Specifically, there is a spectrum of opinion about whether to use statistical methods to alter the final count based on priors about undercounting some subpopulations. Another issue is how to count individuals living abroad, like Mormon missionaries.
A big part of the problem is that to my knowledge, nobody has studied how to use statistical tools to adjust Census counts, and in so doing, figured out how to do it right.