5 Keys to Campaign Strategy #4: Brittle Gerrymandering
How political parties can find — and exploit — breaking points in electoral systems
So far in this series, we’ve talked about a few essential concepts that political campaigns and politically-engaged citizens need to understand: the tradeoffs of persuasion vs. GOTV, how elasticity shapes swing states, and why undecided voters make polls a lot less reliable.
But we still haven’t addressed the elephant in the room: gerrymandering. This unfortunate reality of the American political system discourages many people considering running for office. “These sleazy politicians have predetermined the outcome of the race,” they think. “There’s no point running.”
In this post, we’ll see how gerrymanderers make trade-offs that make a lot of districts competitive under the right conditions, and how we can identify “leverage points” where a party could pick up half a dozen seats with just a small nudge.
How thinly do you spread your voters?
In 2018, the Republicans won the popular vote in the US House elections in North Carolina by just 2 points (50.4%–48.4%), yet they walked away with 10 of the 13 House seats. This lopsided win was all thanks to Republican-led gerrymandering.
This points to the central goal and challenge of gerrymandering: you want to win as many seats as possible, but there’s a limited supply of votes for your party. So how do gerrymanderers think about this puzzle? Let’s think through two different gerrymandering strategies with an example.
The benefits of spreading your votes out
Suppose you’re a Republican in charge of gerrymandering a state to favor your side. Your state has 10 Congressional districts and 100 voters, split up like this:
How many Republican seats can you squeeze out of your light-red state? One option would be to cram all the voters into single-party districts, like so:
The Republicans will win 6 of the 10 seats pretty much every time, while the Democrats will get just 4. That’s not bad, but you, the gerrymanderer, can exceed this. What if you spread your voters out so that you had a 6–4 edge in every single district?
In a normal year, you’d be expected to win every single district. That’s a 10–0 seat edge with just a 60–40 advantage in voters. Impressive! (This is a simplified version of what the Republican gerrymanderers in North Carolina did: they spread out their voters just right so they’d get a small edge in a bunch of districts.)
Robustness and brittleness
It should come as no surprise, then, that most gerrymanderers opt for the second approach: spreading your votes thinly lets you run up a huge number of seats in the average year.
But I’d argue that this thinly-spread strategy has a huge drawback: brittleness. Suppose for a second that Democrats have a good year and that 10 of the 20 voters who lean Republican start leaning Democratic instead. Instead of 40 hardcore Republicans, 20 “lean” Republicans, 20 “lean” Democrats, and 20 hardcore Democrats, we now have a 40–10–30–20 split.
If we’d spread our voters thinly, here’s what might happen:
In half the districts, the Democrats flipped enough votes to narrowly overcome our slight edge. Now, the Republicans have gone from being up 10–0 to being tied 5–5. Spreading votes out gave the Republicans an edge in a normal year but left them vulnerable to a “blue wave.” It made the gerrymander brittle.
Meanwhile, if we’d played it safe and packed voters into single-party districts, we might see a map like this:
Despite the shift toward the Democrats, the Republicans still win all six of their seats by a healthy margin. This gerrymander was robust to the “blue wave.”
In a normal year, the spread-out, brittle strategy was better: it got 10 seats to the robust strategy’s 6. But in a wave year, the robust strategy was better, keeping 6 seats to the brittle strategy’s 5. Gerrymanders, then, can choose between the high-risk, high-reward brittle strategy or the consistent, low-risk robust strategy.
Because most gerrymanderers choose the brittle strategy, this means that the opposition party can strategically search for places where gerrymanders are brittle. If they hammer hard enough, they can shatter their opponent’s advantages.
A case study in Texas
To understand how brittleness plays out in the real world and how you can design a strategy to exploit it, let’s head to the Lone Star state. Republicans have heavily gerrymandered the state and currently hold 23 US House seats to the Democrats’ 13 despite only winning the popular vote 50–47.
Partisan leans and national swings
Before we dig into Texas’s districts, it’s worthwhile to explain some key concepts. The United States is pretty evenly divided between Democratic supporters and Republican supporters, but not every place in the country is as evenly split. “Blood-red” places like Wyoming cast a much higher percentage of their votes for Republicans than the nation as a whole does, a “deep-blue” places like Vermont do the opposite.
To measure how differently an area votes than the nation at large, political scientists developed a term called partisan lean. New York’s 14th Congressional district, home to AOC, has a partisan lean of 29 points toward the Democrats, abbreviated as D+29. This means that if the national vote is 50–50, Democratic candidates could be expected to win by 29 points, or about 64.5–35.5. Meanwhile, Arizona’s 1st Congressional district in Phoenix is R+2, meaning that in a 50–50 national election, the Republicans would be expected to squeak out a 51–49 victory in the district.
Now, that doesn’t mean that Republicans will always win every R–leaning district and that Democrats will always win every D–leaning district. Partisan lean theory says that, if the national vote moves 1 point in either direction, the vote in each district will also move 1 point in that direction. If the shift is large enough, it could overcome the partisan lean and lead the opposite party to win. This is what happened in Arizona’s 1st district (AZ–01) in 2018, by the way: the national vote for the House favored the Democrats by about 8.5 points, a shift that overcame the district’s R+2 lean and helped a Democrat win election to the House.
(Note that 1-to-1 shifts don’t always happen; as we saw earlier, some districts are inelastic and won’t shift much even in a national “wave” election. Plus, some candidates are popular enough that they can shift the race several points toward them, even if the “fundamentals” don’t support them. Every district is different; this level of analysis uses the average district for simplicity.)
Texas’s moderately-red seats
Let’s look back at Texas. The Republicans managed to turn a small edge in the popular vote into a huge edge in seats. Now that we know about partisan lean, we can check out the partisan leans of Texas’s 36 districts to see how the Republicans pulled it off:
Two things might jump out at you:
- There are very few competitive seats. Only one seat (TX-23 at R+1) leans less than 5 points toward either party; just four seats are within 8 points.
- There are a lot of seats that lean modestly toward the Republicans, shown by the cluster of bars around the R+10 line. A full 12 districts — a third of the state’s total — lie between R+9 and R+13, inclusive.
Both of these things are by design. The Republicans roughly followed the brittle gerrymandering strategy we talked about earlier: they put just enough Republican voters in those R+10-ish districts so that their team would win, helping them squeeze out a bunch of modest victories. The Republicans probably figured that anything less than R+10 wasn’t strong enough to withstand a slight Democratic advantage, so they cleared out almost all of the light-red seats, leaving just TX-23 and a handful of others.
The snapping point
Due to the Republicans’ strategy, the Texas map has the interesting feature of being very unresponsive to small shifts toward the Democrats.
- In a normal election year, the Republicans would be expected to win 25 seats to the Democrats’ 11, since 25 seats are R+1 or more.
- If the Democrats have a good year and win the national vote by 4 points, all the seats in Texas would be expected to shift 4 points toward them. They’d be expected to win all the seats that are R+4 or less. Unfortunately for the Democrats, they’d only pick up TX-23, leaving the Republicans with a 24–12 advantage.
- Even if the Democrats won the national vote by 8 points, they’d still lose 22–14.
But after that, things get interesting.
- If the Democrats win the national vote by 10 points, they’ll take the lead, winning 17–16, with 3 tied districts (we assume R+10 districts would be toss-ups in a D+10 year).
- If the Democrats win the national vote by 12 points, they’ll take a commanding 22–12 lead, with 2 ties.
In short, the Republicans’ gerrymander holds up if the Democrats win by 8 points or less. But after that, the floodgates open. This graph shows how nothing major happens until Democrats get an 8-point edge, at which point they suddenly start winning bunches of seats with every additional percentage point:
Brittle gerrymanders can bend to a certain point — but, after that, they break in spectacular fashion. If you can find that point, you can do a lot of damage with just a little push.
Strategic plays
Brittle gerrymandering means that there’s a non-linear relationship between the popular vote margin and the seat distribution. If the Democrats start out even and gain two percentage points of the vote, they’ll only be expected to pick up one seat, but if they start out at a ten-point advantage and gain two more points, they’ll likely pick up five seats.
Political parties would thus do well to identify the “breakpoints” where brittle gerrymandering suddenly snaps and start shoveling resources there. Democrats look to have a 6-point national lead this year, and Texas is rapidly becoming a purple state, so with just a little more investment, the Democrats could earn a windfall of seats in the Lone Star state.
More generally, the non-linearity of gerrymandering means that it makes more sense for parties to focus their investments in states close to the breakpoints, and that it’s usually more effective to concentrate energy into a few states instead of spreading money and time out across many states. If you have the money to swing a race two points in your favor, you need to find the places where those two points will have the most impact.
Finally, what does this mean for gerrymanderers? I’m not a huge fan of gerrymandering in the first place, but I’d argue that gerrymanderers need to start accounting for the fact that we live in an era of political instability, one where red waves (2010, 2014, 2016) and blue waves (2008, 2012, 2018) happen all the time, and one where demographic changes are sweeping the nation. Instead of created flashy but frail gerrymanders, parties would do well to create gerrymanders that remain robust no matter how the tides turn.
The more general lesson:
Optimization makes you brittle. Robustness protects you from instability.
Up last
With this, we’ve developed a strong understanding of how campaigns can squeeze the most benefit out of limited time, money, and other resources.
To finish this series, we’ll take a look at the ongoing 2020 Presidential election and learn how we can apply some of the concepts we’ve developed so far to better understand the candidates’ strategies.
Stay tuned!
