A carnation-shaped gerrymander grows in Georgia

The new state Senate proposal adds Black-majority districts - but still keeps a powerful partisan advantage.

District boundaries in the United States do a lot of jobs: represent counties, cities, and communities; give voice to racial and ethnic groups; and give weight to the major political parties. They influence all these forms of representation, while keeping a roughly equal number of people per district and have to be unbroken contiguous form, ideally of compact shape. It calls to mind the saying about Ginger Rogers’s job being harder than Fred Astaire’s: she had to do everything backward and in high heels.

In math and science, we would call this a problem of optimization under constraints. Such optimization has some basic universal properties. One is that it’s hard to maximize two goals at the same time. For example, if you’re a partisan and you want to get as many seats as possible for your side, then you may end up drawing some funny-looking districts.

Except for the Ohio 9th District, these Congressional districts are surrounded by land.

(Gerrymandered districts aren’t always funny-shaped. Sometimes geography can do the job for you. For example, most cities vote overwhelmingly pro-Democratic, and if their population is small enough, Republicans can gain an advantage just by encircling the city within a single district.)

As constraints are added, gaining a partisan advantage becomes harder. This week the Georgia General Assembly, which is controlled by Republicans, is responding to a new constraint: a federal court’s order to give Black communities more representation. The court ruled that the existing legislative maps (both Senate and House) violate Section 2 of the Voting Rights Act.

The overturned Senate map satisfied both anti-Black and anti-Democrat goals. Because Black communities vote overwhelmingly Democratic, you’d think that redrawing the maps would compel Republicans to give up some seats. As it turns out, that’s not necessarily the case. But they have to get artistic with the district boundaries.

You can look at the Republicans’ proposed remedial plan here (if you want all the legislative proposals, Republican and Democratic, Senate, House, and Congressional, they will all be found here). The remedial plan has 16 majority-Black districts, compared with the overturned plan’s 14 such districts.

There is some odd stuff going on here. It’s a blooming carnation of gerrymandering. (Or maybe it’s a sea anemone!). Note the oblong petal-shaped districts radiating out from the center, where Atlanta is. These districts scoop up suburban Democrats of other races, mostly white.

In addition, the new Districts 17 (63.6% Black Voting Age Population, BVAP), 38 (60.9% BVAP), 39 (55.4% BVAP), and 43 (63.8% BVAP) have above 55% BVAP, a sign that Black voters have been packed into a few districts.

As a result, the proposed map performs quite similarly as the old map:

• Partisan performance: Both maps have 33 Republican-leaning and 23 Democratic-leaning districts.

• Lack of competition: Based on past voting performance, no districts have expected Democrat-versus-Republican margins within 7 percentage points. Only two districts are within 10 points. 

• Partisan bias: Both plans allow Republicans to win a majority of the chamber with as little as 41% of the vote. Conversely, Democrats require 59% of the vote for a majority. 

Computer simulations done by the Electoral Innovation Lab’s Ethan Arsht show that a party-blind process would on average lead to 25-26 Democratic-leaning seats, and about 7 districts that are competitive within 10 percentage points. In other words, both the old and new plans are consistent with an intent to increase Republican advantage and reduce competition.

How did they do it?

Here are the old 15 districts:

Georgia State Senate districts affected by the federal court order: Districts 6, 10, 17, 25, 28, 30, 33, 35, 38, 39, 41, 42, 43, 44, and 55. Atlanta is split among Districts 8, 38, 39, 42, and other districts not shown.

And here are the new proposed districts.

Now the carnation petals are a bit easier to see. What is the racial composition of the petals?

Here, green shading indicates Black populations; blue shading, white populations. This makes it easier to see that Districts 10, 35, 38, 39, 41, and 44 join Black-dominated populations with Democratic-voting white communities.

Partisan lean of the same 15 districts. Note the packing of Democrats into the majority-Black districts.

These petal-shaped districts were not necessary to satisfy the conditions of the court order – but they were essential for purposes of maintaining partisan advantage. 

We can quantify this strategy using a well-known measure, the Reock (REE-ok) score. Named after its inventor Ernest Reock, this score is computed by dividing the area of the district by the area of the smallest circle that would completely enclose it. A perfect circle has a Reock score of 1; a district of mangled shape has a Reock score closer to zero.

The new districts have lower Reock scores overall:

We don’t know yet whether this plan will pass into law. If the federal court signs off on it, then what we have is a clever kind of gerrymander made possible in a swing state: packing minority voters and similar-voting white voters into long urban/suburban districts shaped like flower petals. Call it a Carnation-mander.

In addition to this plan, there’s also a state House plan to consider. And both proposals have alternatives proposed by Democrats. But maybe that’s enough for now!