Monday, March 23, 2009

Blathering About Gerrymandering

While frequently decried as an abuse of the political system leading to the disenfranchisement of voters and poor representation, gerrymandering is the solution to the problems it creates. In the current practice, districts are divided based on demographics, typically grouping majorities of one with minorities of another to maximize one party's "safe" votes. This leads to some truly bizarre arrangements, such as the districts surrounding Houston, TX where relatively liberal Houston is cut into very thin pies and bundled with huge swathes of the surrounding rural areas. This leaves the liberal residents of Houston underrepresented and disenfranchised.

I maintain that the solution to this particular problem is more clustering by demographics and ideology, regardless of the resulting district shape.

One of the fundamental flaws in our system of apportionment is the geographical bias. In all things, votes are assigned based on population in a particular geographical area. This inevitably leads to the "tyranny of the majority" so reviled by the founders. Liberal voters in Texas have basically no federal representation. Conservative voters in Washington are in a similar situation.
Conversely, voters whose opinions are in line with that of the majority have a voice, but little choice. The result of a typically gerrymandered district is one in which the incumbent seat is safe because the majority is a relatively narrow one. Not so narrow as to be at risk, but narrow enough to discourage ideologically similar (same-party) candidates from running a credible primary challenge.

To overcome these problems, voting districts should be arranged with minimal regard to geography. Rather, census data should be used to cluster ideologically and demographically similar voters. As people with similar ideologies tend to cluster geographically, the resulting districts would likely be non-contiguous, but more representative overall. As a Software Engineer, such a solution would be readily implementable in software. Of course, as politicians like to appear to be "doing things" that won't happen anytime soon. Regardless, the end result would be that demographically, but not necessarily geographically interests would be better represented. The cost to the majority party would be relatively minor as they would give up some seats in the legislature, but gain in that the existing seats would be even more safe than
they were previously.

As the districts would now favor one party overwhelmingly, it would certainly increase the "pandering" in campaigns. But this would now be a good thing. Rather than completely disregarding the views of the disenfranchised minority, candidates would be speaking to the whole of their selected electorate. Since the national parties wouldn't have to worry as much about the result, this scheme would counter-intuitively lead to an increase in the overall
competitiveness of elections as same (or similar) party candidates would be free to challenge the incumbent. Sadly for the likelihood of implementation, this would reduce the power of the national parties and incumbents who would be charged with taking the first steps.

Ultimately, doing away with primarily geographical constraints would lead to increased representation among all voters and a decrease in national party (and corresponding "big money") influence. The increased competition at a local level would trickle up to Congress as a whole and lead to less intra-party cohesion and greater bipartisanship as individual representatives would be more beholden to the voters than the national party.