Yonatan Zunger’s Blog

[Cross-posted to novadem]

At this site you can find maps of the United States at single-county resolution showing the votes in 2000 and 2004. The results are fascinating and well worth a look, especially if you take both images and flip between them a few times.

1. The polarization of the country is much less evident than on the electoral-vote-by-state maps, and the “enormous sea of red with islands of blue” effect is revealed to be entirely an artifact of the quantization implicit in states. In fact, state borders are almost invisible on this map, which is a rather strong indicator that states are meaningless political boundaries nowadays. The Eastern half of the country is more or less solidly purple, with some clear blue paths (going up the Mississippi all the way to the Canadian border, in various parts of the Northeast, and so on) and some clear red lakes (in the western panhandle of Florida, for instance), but very little in the way of strong polarization. The West looks different: it has large regions of uniform red and regions of uniform blue. While there is some purple in this region, one gets the sense that that’s an artifact of county quantization; a similar map by street address would probably show much cleaner lines of clustering.

   With regards to this clustering effect: If there were an attractive potential between people of like political orientation, one would naturally see clustering and filament formation, unless there were either enough “friction” (or inertia) that kept people from moving or so much motion (“thermal motion”) that it swamped the attractive force. The West looks like a map of what I would expect to see if the clustering were the dominant force, i.e. people tended to move to be with like-minded people; the East is much more purple, and looks like it could be one of the other two limits. Of course, this explanation is post hoc, ergo propter hoc; people’s clustering may not be caused by politics so much as the other way round. (Consider for instance the Democratic cluster going up the Mississippi and eastwards through the Deep South; what you’re seeing there is racial differentiation) But the presence of clear filament structures on one side of the country and uniform structure on the other is clearly indicative that some causal variable – not neccessarily politics, but correlated with politics – does have this attractive property, and the two sides of the country are effectively in different thermodynamic phases with regards to it. Figuring out what this underlying property is is probably going to be crucial in planning a good strategy; taking into account its different phases in the East and West is going to be very important in finding different approaches to those two areas.

2. The geographic distribution is of much interest, as well. The appearance of a great area of red in the Midwest and West is somewhat misleading, since population densities are radically different. One partial indicator is that (especially in the western half of the country) county sizes are chosen to have more or less constant population; very large counties (uniform areas without boundary lines on the map) have fewer people. (And this is a big effect – the total population of Wyoming is about the same as that of Denver) The least populous counties are fairly uniformly red, and the cities are mostly blue. This map has some striking correlations to the map of the world by its lighting as visible from space – also shown at the same web site. The cities are dominantly Blue, the countryside Red.

   This may have a reasonable explanation – cities are where people are forced to live side-by-side with very different people, and that almost neccessitates developing a mindset of tolerating very different types. In the countryside, especially in very rural areas, mobility is low; people have to get used to dealing with the same people, over and over for decades at a stretch, and therefore uniformity of thought and custom is at a premium. Another factor well worth taking into account.

3. Another interesting thing is to look at the time evolution, flipping rapidly between the two images. The most stunning thing is how little change there is between the two maps; you have to pay attention to see which is which, and I can’t find a single county which had a significant change – certainly not a polarization change. However, the 2004 map contains more primary colors than the 2000 map; if you look carefully, you’ll see that the red areas got redder and the blue areas bluer. Not surprising, perhaps, but it’s unexpected to see this so visibly confirmed. Note also that these two maps were taken on opposite sides of Sept. 11; note how little of an effect even that had on American politics.