As of one DTL’s most decorated mappers, Blyden used yet another novel approach to help him stand out with his PA State House map. This entry explores how our geographic boundaries are outdated. We were fascinated by his analysis. His map required a great deal of work and research, and he achieved outstanding metrics.
My interest in redistricting
These DTL contests hit a trifecta of my interests. I am interested in politics, with a particular interest in issues of justice and political process, and how our election district maps are drawn is a core issue in having electoral justice and truly representative government. I enjoy games and puzzles – and I see DTL’s contests as being real-world puzzles to try to solve. And I like maps.
The problem of Gerrymandering
Having a single-winner district system creates a spoiler effect, and thus a tendency toward a two-party system, limiting voters’ choices. Moreover, because some districts tend to be dominated by one party and other districts the other party, gerrymandering deliberately worsens a problem that exists even in the absence of gerrymandering: The political minorities of many districts are effectively disenfranchised from having meaningful votes.
Eliminating gerrymandering will go a long way toward reducing these problems, but even elimination of gerrymandering will not completely solve the problem. So long as there are districts and an uneven distribution of the population, some people’s votes will be worth more than other people’s votes. In all probability, at least a few people who will lack a meaningful vote, by virtue of being a minority in their district.
Towards breaking the geographic assumption
Implicit in the concept of geographic legislative districts is an assumption of shared political interests or shared political identity based in geography. This assumption likely starts in the feudal origins of British parliament, transferred into U.S. electoral process. The geographic assumption might have still made sense when the U.S> was founded two and a half centuries ago. The nation was overwhelmingly agrarian. Government was of, by, and for the landowning class of men. The U.S. was much more homogeneous occupationally and demographically than today. Communication was at the speed of horses and sailing ship. Modern voting technologies were not yet imagined.
Today, we live in a very different world. Only about 1% of our population are farmers. Electronic communication, especially broadcast media and the internet,enable cultural similarities that cut across geographic location with far greater diversification within local populations. This means that people who share political identity are more likely to be geographically dispersed rather than concentrated in a single area.
Does the geographic assumption still make sense? Political science has given us knowledge that was unavailable to the founders about the adverse effects of voting systems, and modern technology has made it easy to calculate tallies of transferrable votes that would have been extremely challenging in past centuries. I think it is time that we use this knowledge and technology to at least partly liberate our legislative bodies from the geographic constraint.
Most state legislatures, including that of Pennsylvania, are bicameral. Why not convert the larger of these bodies, the House, to a single, state-wide district from which the representatives are elected proportionally, using a single transferable voting (STV) or similar electoral system? That way virtually everyone’s vote would count and everyone’s vote would have nearly equal impact.
The values that guided my state House and Senate maps were similar. In addition to the required goals of set number of districts (50 and 203, respectively), each entirely contiguous, and populations within a couple percent of the target population, I was striving mainly to make compact districts with high population equivalence, with “clean” boundary lines.
I did not pay attention to the partisan mix, which I consider another form of gerrymandering. Nor did I pay much attention to the minority-majority composition.
In each map I also have in mind the idea I call regionality: that districts should reflect internal political commonalities; that the parts of a district should “go together naturally” in our sense of geographic-based identities that come from common geographic, economic, and social factors.
For the House map, I started that with the same idea of avoiding county splits, but that soon appeared to be a hopeless, both because the population equivalence of the much smaller districts is very sensitive to inclusion of particular blocks – single block populations often exceed 5% of the district target population – and because trying to conform to county boundaries led to some very strange districts.
So I made a decision not to worry about county splits and instead try to focus on population clusters, i.e. sets of neighboring municipalities, etc, even where such clusters ran across county lines.
Input from Other Sources
For House map districts I did consult a few sources. When I did the Pittsburgh area, I used a Pittsburgh Neighborhoods map as a model, as much as population constraints allowed. For Philadelphia, I used Google maps and aerial photography to make districts that conformed as closely as I could to recognized Philly neighborhoods, and put district boundaries either on natural breaks in the residential population (e.g. riparian parks, industrial areas, etc) or else on key streets:
The House map was extremely challenging. Partly this is because the districts are small relative to some of the blocks. I tried to get each population within about 1% of the target population, i.e. ± 630. There are seven or eight districts where this goal was not met. A much bigger challenge is how hard it is to monitor how much over or under target population one is across all districts. It is much easier to discover at the “end” of the process that one has either too little or too much population left to complete a district.
Even recognizing this as a potential problem as I went along, I was only tracking it by glance, and ran into this problem. With so many districts, and each district as sensitive as these are to changing even a block or two, that required a lot of further adjustments!
With only 17 or 18 districts in the CD redistricting, there were larger populations that were much more robust to changing a block group or two, so this was a very minor issue. Even with 50 districts in the Senate map it was very manageable.
With 203 small districts that are thus very sensitive to changes, this became the most challenging part of the entire map drawing process, and made the House map even more time-consuming and laborious than you might expect from the mere number of districts.