For more on the issue of how many uninsured there really are and how long they are uninsured for, go look at The Bloviator’s post. His and my conclusion is that the Census Bureau has mislabeled the 15.3% of the population that is uninsured "for the whole year." That number instead represents a snapshot of those uninsured at any one time. Still a big number, though with many people moving in and out of it, hence a big political effect.
Regarding the political effect, my friends at Harris Interactive have given me a peek at the numbers in their classic question about the US health care system. They have been asking the public about their view of the system forever. When the number of people thinking that the system needs to be completely rebuilt goes above 30%, as it did in 1991 and 1992, its time to pay attention to health care politically. Well as you can see below, its over 30% now!
2003 data from Harris Interactive:
System works pretty well/minor changes are needed: 17%
Some good things/but fundamental changes are needed: 50%
So much wrong/system needs to be completely rebuilt: 31%