Saturday, January 24, 2009


A recent Slate article documents attempts by mathematicians to solve the problem of Gerrymandering. There are jokes about circles not tessellating well and talk of algorithms to measure compactness, competitiveness, fairness, county integrity . . . the list goes on. And, in my humble opinion, every single one of the yahoos involved misses the point.

As you no doubt recall, we tackled the issue several weeks ago, but the so-called experts haven't learned a thing. The problem isn't that these guys aren't good at math, it's that they've sacrificed what should be the one overarching goal of a redistricting scheme in favor of algorithmic showboating and myriad minor (and/or dubious) goals.
Let's be clear about that one primary goal. Currently some political party in power can Gerrymander the districts to favor their party — e.g., lump all the Republicans in one district and get small Democratic majorities in the others. We want to prevent this from happening. Period. The following other goals are comparatively inconsequential:
  • The idea of compactness is simple. Try to make the districts so that the people in it are as close to the center as possible. Back in the day this was hugely important. No one wanted to travel too far by mule to go hear some blowhards debate about taxes and tariffs and so there was some merit in making the district look more like a circle than a square. But now this is a nice-to-have provided we don't do anything too bizarre. Internet advertising and organizing have made compactness even less relevant.
  • Competitiveness and fairness are dubious goals. If it happens naturally that one party does better than its proportion of registered voters in the state suggests it should, well good for them. Let's not go out of our way to impose some sort of penalty on popular representatives. And all the fairness proposals have to do with making sure the Democrats and Republicans are evenly dispersed. We want to prevent the parties from making things uneven on purpose, but it's debatable if we want to codify the importance of these two parties in particular by mandating their proportions in congressional districts.
  • The idea behind county integrity is to match the districts as closely as possible to county line. Really? Who gives a shit? Does anyone even care what county they're in anymore? We shouldn't give one iota of thought to this absurd metric.
The metrics are wrong; but so is the general approach. One of the ideas is to let someone propose a redistricting scheme and then evaluate how "Gerrymandered" it is based on their scoring system. But the party in charge will no doubt still find some way to manipulate the districts in their favor and get a high "score" under the new fancy algorithm — but this time they'll have the cover of a good score. And all that these other metrics accomplish is to make any scheme and evaluation method more complex and therefore less transparent.

Far better to generate the districts according to an objective procedure. We can easily secure our primary goal with a simple, transparent scheme that can't be manipulated by the party in power.

If those involved were more interested in solving the problem and less concerned with showing off their algorithmic creativity, they'd adopt a system that was simpler, not more complicated. But no one likes the simple answer. It's harder to show off how smart you are with a simple answer.

Besides, Mithriblocks are compact enough and tessellate as well as anything. Of course, it doesn't really matter in the end. Those in power don't want the problem solved. They're better off adopting a fancy algorithm that they can manipulate, even if we're not . . .

1 comment:

Phutatorius said...

Yeah, so you can put Mithriblocks to use in Alaska, but you're still going to get a district that's the size of Alaska. So much for compactness, then.

I'm slowly coming around to this. My reservations were principally that the flipside of fairness/nonmanipulability (if that's a word) is arbitrariness. And if there is a substantive value worth fighting for here, it's that lines should be drawn to ensure the best representation of local interest.

But representing local interests isn't a constitutional value -- Article I only talks about how many seats to assign to each state. It doesn't discuss how those seats ought to be filled, and the more I think about it, I'm not so sure why representing local interests should be a transcendent concern.

So, OK. I think I'm sold.

But let me throw one more idea out there: if we're going to throw all result-oriented considerations out the window, how about a system in which each state's voters get to vote for all that state's representatives? Reps would have to deliver the goods to everyone in their state, and as a result I think you'd see a lot less pork.

Post a Comment