Preventing Partisan Gerrymandering with Partisan Gerrymandering


Disclaimer: this is not a serious proposal. At best, it’s a thought experiment, but only if you don’t think too much about it.

Florida is interesting in that they are just one of three states (according to Justin Levitt) that explicitly allow for “floterial” districts. A floterial district is one that overlaps one or more other districts in a map. A formerly common example of this type of district arose when a state gained a member of the House, but failed to pass a new congressional map; the new member would be elected by the entire state, at large. These fell out of use with the 1960s reapportionment revolution (though are still the federal statutory remedy should a state fail to pass a map while gaining a seat), when redistricting became mandatory after every census. The only current usage of floterials I’m aware of is in the New Hampshire State House – they place a high value in not splitting towns, and in complying with equal population requirements, they make up population deviations by drawing floterial districts over adjacent underpopulated districts.

But floterial districts could be an interesting way of combating gerrymandering, by allowing both parties to gerrymander. What would happen if you allowed Republicans and Democrats to each draw their own statewide map with half the districts required, and overlaid them into a single, full district plan? Would the net result end up being relatively fair? In the spirit of the re-redistricting of the Florida congressional map that’s going on now, I gave it a shot with Florida’s 27 districts, with a 13-district Democratic gerrymander, a 13-district Republican gerrymander, and a statewide at-large district.

I used the voting districts loaded into Florida’s MyDistrictBuilder app as the unit I built on. Census blocks would allow for more gerrymandering and exact population equality, but I spent more time on this than I should have as it was. I did keep district variation within 1000 people above or below the target, which is pretty darn close, considering the target was about 1.45 million.

Democratic Map


District % Democrat Party
1 54.4% D
2 32.0% R
3 52.2% D
4 51.8% D
5 51.9% D
6 52.2% D
7 50.4% D
8 50.1% D
9 51.5% D
10 52.5% D
11 52.2% D
12 50.4% D
13 48.8% R

% Democrat is the average of the vote share received by Obama in 2012 and Alex Sink in 2010, with third-party votes excluded. This has the upside of making above 50% a Democratic district, below 50% a Republican district.

I managed to draw an 11-2 Democratic map. This is harder than it looks – the state is unbalanced when it comes to political geography, with a ton of Democrats in the Miami area, and just a handful of solid Democratic pockets up north. Thus, even with a highly packed Republican north Florida district (#2), there’s a lot of stretching districts a long way to Miami to hit >50% Democrat.

With more time, a 12-1 map is probably doable, since District 13 is so close, but Democrats would probably prefer a safer 11-2 map instead – I’ll admit, these margins ended up being a lot closer than I had originally planned.

Republican Map



District % Democrat Party
1 41.4% R
2 47.3% R
3 48.3% R
4 46.6% R
5 47.1% R
6 46.2% R
7 47.3% R
8 48.3% R
9 47.9% R
10 47.5% R
11 66.9% D
12 72.8% D
13 50.2% D

This is a 10-3 Republican map, with better margins on the whole than the Democratic map, and an 11-2 plan is definitely possible given District 13. Compared to the Democratic map, this one is downright compact, again thanks to the political geography; give up a couple solid Democratic districts in Miami, crack the rest, and it’s pretty simple to draw a bunch of safe-but-not-too-safe Republican districts.


The 27th district is statewide, and it actually ends up being the closest district in the map, partisanship-wise, given the measure I’ve been using. The 50/50 nature of the state makes it especially suited for this idea, since if you allowed both parties to have free rein over drawing maps, neither has an advantage – their only limitation is how safe they want to draw the districts for their own party. There’s a strong likelihood that you’d end up with a 13-13 map, with a highly competitive at-large seat.

Setting aside the statewide district, an interesting consequence is that most of the residents of the state would be represented by one Democrat and one Republican, with just 24% residing in two districts with the same partisanship. This makes it a weird twist on Thomas Brunell’s observation that voters are more likely to be dissatisfied when they’re on the losing side of a congressional election, which leads him to suggest that partisan packing, when done for both sides, may actually be a good thing.

The reason for the disclaimer at the start of this post is that these maps obviously break just about every other good redistricting practice. Compactness is abysmal, counties and cities are split, communities of interest are thrown out the window. There are no black majority-minority districts – none are even close – although there are still three Hispanic majority-minority districts like there are in the map in place now.

If one were to take this idea seriously, one approach would be to draw a black majority-minority district at a standard population to help comply with Voting Rights Act requirements, and then have two 13-district maps drawn by each party with the rest of the state, but constrain them under the Fair Districts Florida rules (other than the ban on drawing a district to benefit a party). It might even result in the lack of a court challenge, at least from the parties themselves.

Leave a Reply

Your email address will not be published. Required fields are marked *