//=time() ?>
For US census block data, I think assigning the populations to the polygon boundaries makes a lot of sense (vs. a choropleth), since these people usually live on streets.