Mathematicians are deploying algorithms to stop gerrymandering

Thomas Hofeller, long the official redistricting director of Republican National Committees, was one of these users for decades. He died in 2018.Gerrymandering is the practice of packingscattering votes across different districts for one party, diluting their power and stuffing likeminded voters into one district, which wastes their power. Austin, Texas is divided into six districts. It is the largest US city without an anchor district.The Republicans Redistricting Majority Project (or REDMAP) was the catalyst for the full threat in 2010. It spent $30 million to win down-ballot state legislative elections, winning in Florida, North Carolina and Wisconsin. David Daley, author Ratf**ked:The True Story Behind America's Secret Plan to Steal Democracy, says that their wins in 2010 gave them power to draw maps in 2011."What was once a dark art, is now a dark science." MICHAEL LIThe technology had improved by leaps and bounds in the time since the previous redistricting cycle, which only accelerated the outcome. He says that it made the gerrymanders who were drawn in that year more durable and lasting than any other gerrymanders. The sophistication and speed of computers and the data available make it possible for partisan mapmakers put their maps through 60 to 70 iterations, and then to refine and optimize the partisan performance.Michael Li, a redistricting expert from the Brennan Center for Justice at New York University's law school, says: "What used to be a black art is now a black science." And when they are applied in an election, he says that it is almost impossible to overcome.A mathematical microscopeMattingly and his Duke team have been staying up late writing code that they expect will produce a huge win, algorithmicallypreparing for real-life application of their latest tool, which debuted in a paper (currently under review) with the technically heady title Multi-Scale Merge-Split Markov Chain Monte Carlo for Redistricting.However, technical discourse advancement is not the main priority. Mattingly and his fellows hope to educate politicians and the general public, lawyers, judges, mathematicians, scientists, and anyone else who is interested in democracy. Mattingly delivered a lecture in July with a shorter title: Can you hear what the people want to do?Many gerrymanders are known to have misshapen districts. Mattingly and Greg Herschlag, his key collaborator, provided expert testimony in several of the ensuing lawsuits. There have been many legal challenges in the United States over the past decade.Mattingly states that such disfigured areas make great posters, coffee cups, and T-shirts. However, stopping gerrymandering will not stop the use of strange geometries.These North Carolina congressional district maps show how geometry isn't a reliable indicator of gerrymandering. The courts deemed the NC 2012 map with its strange district boundaries to be a racist gerrymander. Although the NC 2016 map is more readable and manageable than the NC 2012 map, it was rejected by courts as a constitutional political gerrymander. Jonathan Mattingly of Duke and his team analysed the maps and found that they were identical in terms of their partisan outcomes. The NC 2020 map was drawn by a court-appointed expert. JONATHAN MATTINGLYA number of mathematical scientists have developed tools that can be used simultaneously to test for extreme outliers. Based on the Markov chain Monte Carlo algorithm, this mathematical method generates a random sample map from a variety of possible maps. It also reflects the probability that any map drawn will meet various policy considerations.Ensemble maps encode various principles that are used to draw districts. They also account for how these principles interact with the state's geopolitical geography. These principles, which vary from one state to the next, include keeping districts compact and connected, making them approximately equal in population, as well as preserving counties, municipalities and communities that share common interests. District maps must be compliant with the US Constitution as well as the Voting Rights Amendment of 1965.Mattingly and his colleagues will download the 2020 Census Bureau data and run their algorithm to generate typical, nonpartisan district plans in North Carolina. They will use historical voting patterns to identify benchmarks from the vast array of maps. They will assess whether the maps will produce different election outcomes, such as the number of seats won or lost by Democrats or Republicans. Given plausible voting patterns and a 50-50 split, it is unlikely that a neutral map would result in Republicans winning 10 seats while Democrats would win three.Mattingly says that we used computational mathematics to predict the outcomes of unbiased maps. Then we could compare it with a specific map.They will announce their findings by mid-September and hope that state legislators will follow the warnings. They will analyze the new district maps and engage in political discussions with the public. If the maps are found to have been gerrymandered again, there will be additional lawsuits in which mathematicians again play a central part.Mattingly states that I don't want to convince anyone that something is wrong. I want them to see a map, understand its properties, and draw their own conclusions.Jonathan Mattingly, an applied mathematician from Duke University, is Jonathan Mattingly. COURTESY PHOTOIn 2017, and 2019, Mattingly testified that they had analyzed two more North Carolina district maps. The court found that the maps were discriminatory against Democrats, and agreed to a partisan gerrymander analysis. Wes Pegden, a Carnegie Mellon University mathematician, testified in a Pennsylvania case using a similar methodology. The court found that the map discriminated against Republicans.Li says courts have struggled for years to determine how to measure partisan-gerrymandering. The breakthrough came when courts used these tools to create maps.The North Carolina case was brought before the US Supreme Court together with a Maryland case in 2019. Eric Lander, a mathematician, geneticist, and professor at Harvard, noted in a brief that computer technology had caught up to the problem it created. A test that asks: What fraction of redistricting plan are less extreme than the proposed one? This is a simple, quantitative mathematical question that has a correct answer.A majority of justices ruled otherwise.Li says that the Supreme Court's five justices are the only ones who seem to struggle understanding how math and models work. It was easy for federal and state courts to apply it. This was no more difficult than complex cases involving sex discrimination or complex securities fraud. Five justices of Supreme Court disagreed.They also stated, "This is not for our to fix" This is for the states and this is for Congress. It is not for us, Li.It will matter.Daley believes that the Supreme Court decision gives state legislators a green signal and no speed limit in regards to the type of partisan gerrymanders they can enact during map-making later in the month.