The “Solution” to Gerrymandering

Some people highlighted Alex Krogh-Grabbe’s maps of the Providence mayoral primary but of far more interest to me was this earlier post on algorithmic redistricting. In the original Wonkblog post from which Krogh-Grabbe draws from, they highlight the work of Massachusetts software engineer Brain Olson who created a program that creates “optimally compact” districts. Wonkblog titles this the “solution” to gerrymandering, but Krogh-Grabbe is skeptical.

Instead he offers a great critique. District compactness isn’t the sole concept by which we determine district quality, and there are plenty of other features that have to be taken into account, among them respecting city and town boundaries, the need for majority-minority districts, and respecting “communities of interest.” Krogh-Grabbe also points out that “computers are only as objective as the people who program them, and only as accurate as the data that they use…” but the use of algorithms should be a bit more transparent than the use of consultants, bipartisan commissions, or independent commissions.

In someways, Krogh-Grabbe is making similar arguments to this May 22 post by Georgetown University’s Jonathan Ladd who argues “gerrymandering in a useless concept.” Ladd makes the point that it’s virtually impossible to satisfy all the different goals of redistricting. The Baker v. Carr decision (which ended the practice of RI state senators representing geographical areas rather than people) requires we have districts of equal population, but we weight the importance of the other goals is up for debate. Ladd argues that redistricting is ultimately a choice among values.

For instance, the demand for political competition means you simply can’t draw compact districts and call it a day. Voters tense to cluster together with like-minded voters, and so you see Democratic voters packed into dense urban areas and Republicans spread out more throughout the country. So compact districts will favor Republican candidates, even if more votes are cast for Democrats (case in point, the 2012 House elections). This is also why people get confused about why maps of US presidential elections by county look so overwhelmingly Republican. Democrats are packed into dense areas while Republicans are spread out in open country.

Now, there are ways to sidestep this stuff. The shortest split-line method removes the fancy computer program, throws out the rules, and makes geometrical districts based on the shortest lines necessary to split two populations equally until you have enough districts. Alternatively, you can pretty much ignore whatever happens in the district-based elections and use something like mixed-member proportional representation.

If there is a Constitutional Convention in Rhode Island, this is the stuff it should really be debating, because changes like these will have a far greater impact on the responsiveness of legislators to their constituents then Ethics Commission oversight (a good thing) or the line-item veto (meh).