Tuesday, November 16, 2004

Politics as Life

In the early days of computers the mathematician John Conway invented a little game which was easy to program and attracted a wide following. The game, called "Life", was an early example of what's now called "emergent complexity", i.e., the rules are simple but the results are complicated.

The game works like this. There is a board with checkers on it and there are rounds ("generations"). Each round you put a checker on an empty square or remove a checker that's already there, depending on the positioning of the checkers already on the board near that square. The point is that from one generation to the next the patterns of checkers propagate, creating new patterns, very much as life propagates itself. A free version to play around with can be downloaded from here. It's fascinating to set up a certain pattern and then let it run for many generations to see what it does over the course of time. Many patterns die off and the board becomes empty of life. Some patterns become completely static. A few really interesting patterns keep reproducing, mutating over the course of time into an endless variety of surprising forms.

The game of Life is based on the premise that each checker exerts a kind of "force field" on the squares around it, so that its very presence will determine the presence or absence of a checker during the next generation on those nearby squares.

Now take a look at the county-by-county results of the latest Presidential election, and click on the tab to compare to the results of the 2000 election. It's apparent that blue counties tend to clump together, like checkers in the game of Life. Thinking of each election as a new generation, we see that the blue areas are propagating themselves (as are the red areas).

I'm not a political scientist and I possess neither the means nor the time to do the experiment, but I'd like to pose the question: is it the case that simple location is a far better predictor of ones vote for President than any other possible indicator, such as family, gender, income, or all the other criteria social scientists are incessantly using to slice us and dice us?

I posit that one can probably determine rules for politics, similar to the rules of Conway's game of Life, so that the presence in the last election of overwhelming red voters in all the adjacent counties makes it very likely that a given county will turn out red in the next election. I posit that the votes of the people around us exert a sort of "force field" on our thinking which strongly, perhaps more strongly than anything else, determines how we vote next time.

It's well known that people aren't rational. Much as we all like to think we're making rational choices for President, that's probably not really true. But it's interesting to investigate exactly what the factors are that are most influential in producing results. Could it be that mere geographic location is highest on the list? It is at least plausible. Human beings want to get along with the people around them. We're hard-wired for this. It causes lots of difficulties to be a Bush supporter in a predominantly Kerry area or a Kerry supporter in a Bush area. It tends to make one doubt ones own belief system. There is a huge tendency to want to vote with the people one knows.

I further posit that, if one could create connections in "another dimension" on the map (such as, for example, a line from Broward County, FL to Manhattan, NY)--necessary because all the people "known" to some residents in some parts of the country are actually residents of some other part of the country--one would be able to almost completely predict how any given county would flip, based solely on describing the way things went in surrounding counties, "surrounding" being taken in this broader sense.

Here's a further question. If this is so, doesn't the Electoral College make a tremendous amount of sense?


Post a Comment

<< Home