Quotes from Experts

Redistricting tools and gerrymandering

Reporters can use the video clips, audio, and comments below in news stories, with attribution to the scientist who made them.

Journalists: Get Email Updates

What are Quotes from Experts?

November 2, 2022


Does partisan gerrymandering really matter that much? Whom does it harm, and how?


Wendy K. Tam Cho, Ph.D.

“That’s actually a legal term. And so without going into the legalities of it, the popular understanding is that it’s when parties entrench power unfairly. So when you entrench power you basically take power away from the voters. So voters, they vote, and it doesn’t really matter—the legislature, and the legislature is not responsive to how the voters vote. And that encourages the wrong incentives, and it basically creates a democracy that none of us signed up for, right? It’s basically not democracy.” (Posted November 2, 2022 | Download video)

Wendy K. Tam Cho, Ph.D.
Professor of political science, statistics, mathematics, computer science, Asian American studies, and law, University of Illinois at Urbana-Champaign.

Daryl Deford, Ph.D.

“I think we saw a bunch of examples in the last cycle, where it certainly made an impact. And you could look at states where courts found that maps had been partisanly gerrymandered, and the citizens felt like they weren’t getting good representation in those districts. So that’s one harm that’s very local.

“But there’s also a bigger harm in terms of people’s feelings of unfairness about the overall system. They are voting in something that seems to be an unfair layout.” (Posted November 2, 2022 | Download video)

Daryl Deford, Ph.D.
Assistant professor of data analytics, Washington State University

Maxwell Palmer, Ph.D.

“Partisan gerrymandering is when one party is able to draw legislative districts to systematically advantage themselves at the expense of the other party. And the consequence is that voters who support candidates of the other party are less able to see their preferences reflected in the legislature, both political preferences and policy preferences. It also means that a party can win control of the chamber—a majority of the seats—without winning a majority of the vote. So, for example, in Wisconsin the state house was gerrymandered such that Republican candidates, even without winning a majority of the vote statewide, can still win a majority of the seats in the chamber and win control of it, even if a majority of voters prefer Democratic state representatives.” (Posted November 2, 2022 | Download video)

Maxwell Palmer, Ph.D.
Associate professor of political science, Boston University

What tools are available to evaluate the level of partisan bias in district boundaries?


Wendy K. Tam Cho, Ph.D.

“Our general tool is just a computer. And computer basically is something that helps us measure different aspects of a map. So how many people are in a district, how competitive the district is, whether we’ve kept cities together, whether the districts are compact, and it also helps us with tradeoffs. So if I have to keep this city together, what happens to the district, what happens to competitiveness? So there are a lot of complex interplay of these factors, and the computer just basically helps us understand those types of things.” (Posted November 2, 2022 | Download video)

Wendy K. Tam Cho, Ph.D.
Professor of political science, statistics, mathematics, computer science, Asian American studies, and law, University of Illinois at Urbana-Champaign.

Daryl Deford, Ph.D.

“This is a question that political scientists have been thinking about for a long time, and they’ve designed a bunch of really useful measures to try to capture it. So those include things like the mean-median difference, which computes the difference between the average voting behavior across the state—which is the mean—and the median performance of a district in the state as a measure of fairness or bias. Another common measure is the efficiency gap, which measures the difference between the number of wasted votes for each party, again across the state.

“And more recently mathematicians and computer scientists have built methods to generate large ensembles, or big collections of plans that satisfy a version of the state’s criteria for making a redistricting plan. You can apply those exact same measures of political imbalance that the political scientists generate to these big collections of plans, to try to understand a baseline for fair behavior in a state, taking into account where voters live and the overall political geography of the state.” (Posted November 2, 2022 | Download video)

Daryl Deford, Ph.D.
Assistant professor of data analytics, Washington State University

Maxwell Palmer, Ph.D.

“There are many different tools and measures available to measure partisan bias. Some are relatively simple measures. They try to calculate something based on the share of the votes received in each district and then the outcomes—who actually wins each district. Others are more complicated and use computer simulations to simulate a large numbers of maps and compare how computer-drawn maps differ from the actually drawn maps. (Posted November 2, 2022 | Download video)

Maxwell Palmer, Ph.D.
Associate professor of political science, Boston University

In what ways are computer-assisted tools currently being used to increase partisan advantages in drawing boundaries?


Wendy K. Tam Cho, Ph.D.

“If you have information on the partisan balance in different districts, you can basically change it, right? And while you’re changing it you can keep track of the other things you might be interested in in the district: how competitive it is; whether you want it to be competitive or not; whether you want to keep districts at a certain level of compactness, meaning a nice shape; whether you want to keep the cities together. You know, you can kind of keep track of all these things and you can basically set them at whatever levels you want while creating whatever partisan advantage you want.” (Posted November 2, 2022 | Download video)

Wendy K. Tam Cho, Ph.D.
Professor of political science, statistics, mathematics, computer science, Asian American studies, and law, University of Illinois at Urbana-Champaign.

Daryl Deford, Ph.D.

“This is something that I’ve thought about a lot, that we’ve thought about a lot when I worked at the Metric Geometry and Gerrymandering Group at Tufts University, because we released a package for analyzing redistricting, and doing this ensemble process. And something that we were really concerned about was that if we make this open-source tool, to what uses is it going to be put, right? What sort of things might people do with it once it’s out there?

“And we had a lot of discussions about this and about the ethics of it and settled on the side that openness was an important part of our process—and also that we were responding to a landscape in terms of gerrymandering, where a lot of gerrymandering that was happening behind closed doors and using either closed-source or proprietary data. It was a little difficult to respond to. And so we were hoping to encourage more citizen activism. We were hoping to encourage making these tools available to the public to analyze the plans in their states. And there’s no question that a lot of the gerrymanders that exist had some optimization and techniques applied to them, I think. Whether those were computational or whether they were by hand, I think is less relevant than the fact that it occurred, and that we were providing a tool to help analyze that and detect it where it had happened.” (Posted November 2, 2022 | Download video)

Daryl Deford, Ph.D.
Assistant professor of data analytics, Washington State University

Might algorithmic tools be able to draw district lines more fairly? What would the strengths and limitations be?


Wendy K. Tam Cho, Ph.D.

“Algorithms can be used to make a lot of things more fair. They can also make things less fair. I think there’s been a lot of discussion recently about, say, facial recognition algorithms or, say, who gets a mortgage and who doesn’t get a mortgage. A lot of these things can be facilitated by algorithms. They can also hurt fairness by using the algorithm instead of using a human. Redistricting algorithms are the same. The technology itself is neither good nor bad. It’s really about how they’re used. And I think what we need to do with algorithms in general—and certainly with redistricting ones, as well—is to think about what we’re creating, how we’re using them, and how they actually change the process of redistricting. And if we’re able to change the process in a positive way, then the algorithm can do good. But if it changes the incentives in a bad way or just creates another tool for someone to use to do bad things, then we’ve actually done something bad with the algorithms. (Posted November 2, 2022 | Download video)

Wendy K. Tam Cho, Ph.D.
Professor of political science, statistics, mathematics, computer science, Asian American studies, and law, University of Illinois at Urbana-Champaign.

Daryl Deford, Ph.D.

“Redistricting in the U.S. is really fundamentally a human problem. So it’s not one that we should expect to outsource to the machines or outsource to computer algorithms. There’s an enormous number of things to balance here that differ state by state, in terms of priorities, in terms of the history of the state, the local communities that are supposed to be preserved. And so I think these computational tools are great for helping to explore this space, to understand baselines, to detect potential bad behavior.

“But at the end of the day this really is a domain that should be handled by humans and should have a lot of input from political scientists and legal scholars. And one of the big lessons, I think, on the computational and mathematical side that we’ve discovered is that there are an enormous amount of unintended consequences for trying to overoptimize for specific criteria or try to dial up the importance of individual rules over others. And these tradeoffs and places where the metrics conflict are really places that need to be addressed by human beings as part of the process outlined by the states—and not just outsourced to an artificial intelligence or a computer algorithm or something like that.” (Posted November 2, 2022 | Download video)

Daryl Deford, Ph.D.
Assistant professor of data analytics, Washington State University

Maxwell Palmer, Ph.D.

“Algorithmic tools seem really appealing to us. The idea that we could let a computer decide this really contentious process and come up with a fair or ideal map would be great. It would remove the political conflict, the debates about redistricting and gerrymandering. But to me the challenge is that the computer has to optimize for something. And what we’re choosing to prioritize in our algorithm is a choice about political and policy values. Do we want to see certain communities represented well, do we want to draw districts that look really good, do we want to preserve past districts as much as possible? Whatever we’re optimizing for would require some sort of political debate or resolution to decide how we should program this algorithm in the first place.” (Posted November 2, 2022 | Download Video)

Maxwell Palmer, Ph.D.
Associate professor of political science, Boston University

Wendy K. Tam Cho, Ph.D.
Professor of political science, statistics, mathematics, computer science, Asian American studies, and law, University of Illinois at Urbana-Champaign.

None.

Daryl Deford, Ph.D.
Assistant professor of data analytics, Washington State University

Dr. DeFord conducts research on applications of mathematical and computational methods for analyzing political redistricting. He also helped develop the GerryChain software package for redistricting ensemble analysis. During the 2020 redistricting cycle he served as an expert in court cases in Wisconsin and Pennsylvania and also collaborated with mathematicians in Colorado to consult for the Colorado Independent Legislative Redistricting Commission during the line drawing process.

Maxwell Palmer, Ph.D.
Associate professor of political science, Boston University

Dr. Palmer frequently serves as a consultant and expert witness on voting rights and redistricting litigation. In the past year he has testified in Caster v. Merrill before the U.S. District Court for the Northern District of Alabama; Pendergrass v. Raffensperger and Grant v. Raffensperger before the U.S. District Court for the Northern District of Georgia; and Galmon, et al. v. Ardoin before the U.S. District Court for the Middle District of Louisiana. He also served as the independent racially polarized voting analyst for the Virginia Redistricting Commission in 2021, and has worked as a consultant to the United State Department of Justice on several matters.