Mathematicians are using algorithms to stop gerrymandering

For decades, one of those users was Thomas Hofeller, “Michelangelo of the modern gerrymander“He has long been the official redistribution director of the Republican National Commission, which died in 2018.

Gerrymandering schemes include “cracking” and “packing”: parties scattering votes across constituencies, thus diluting their power, and gathering the vote they want in a single constituency, wasting power they would have elsewhere. The city of Texas (Austin) is cracked, divided into six districts (it is the largest city in the U.S. that does not anchor the district).

In 2010, the full force of the threat was carried out with the Republican Republican Majority Project or REDMAP. He spent $ 30 million on statewide legislative voting races, and won results in Florida, North Carolina, Wisconsin, Michigan and Ohio. “The profits made in 2010 gave them the ability to draw maps in 2011,” says author David Daley. Ratf ** ked: The true story behind the secret plan to steal American democracy.

“What was dark art is dark science now.”

MICHAEL LI

That technology has made strides since the previous redistribution cycle surpassed the result. “The gerrymanders drawn in that year made them much more enduring and enduring than any other gerrymanders in our nation’s history,” he says. “The sophistication of computer software, the speed of computers, the amount of data available allow partisan mappers to place their maps in 60 or 70 different iterations and truly refine and optimize the partisan performance of these maps. “.

Michael Li, a reduction expert at the Brennan Center for Justice at New York University School of Law, says, “What was once dark art is now dark science.” When manipulated maps are implemented in elections, he says they are it is almost impossible to overcome.

Mathematical microscope

Mattingly and his Duke group They’re keeping writing late in hopes that they’ll get a “huge profit, algorithmically,” preparing the final tool for their actual application, which was released in a technically headlined article (currently under review)Multi-scale summation distribution to redistribute the Markov chain to Monte Carlo“.

Carrying out technical discourse, however, is not a top priority. Mattingly and his colleagues hope to educate politicians and the public, as well as lawyers, judges, mathematicians, scientists — anyone interested in the cause of democracy. In July, Mattingly gave a public speech with a more accessible name that cut faster: “Do you listen to the will of the people in the vote?”

Mishapen neighborhoods are often thought of as a brand of a gerrymander. With the 2012 map of North Carolina, the congressional districts were “very strange-looking beasts,” says Mattingly (along with his key collaborator, Greg Herschlag) who gave expert testimony in a number of subsequent lawsuits. Over the past decade, there have been legal challenges across the country — Illinois, Maryland, Ohio, Pennsylvania, Wisconsin.

Such distorted neighborhoods “make very nice posters and coffee cups and T-shirts,” Mattingly says, “it’s true that stopping weird geometries won’t stop gerrymandering.” And the truth is that with technologically sophisticated hand games, a gerrymandered map is hard to spot.

The tools developed by several mathematical scientists at the same time provide what is called the “extreme-outlier test”. The approach of each researcher is somewhat different, but the result is that a suspicious gerrymandered map is compared to a large collection or “set,” with an unbiased, neutral map. Mathematical method of work, based on the so-called Markov chain Monte Carlo algorithms—Creates a random sample of maps from a universe of possible maps, and a map drawn reflects the probability of satisfying political considerations.

Group maps are coded to capture the various principles used to draw ranges, taking these principles into account as they interact with the geopolitical geometry of a state. Principles (which vary from state to state) include criteria such as keeping districts relatively dense and connected, having an approximate population, and caring for regions, municipalities, and communities with common interests. And district maps must comply with the U.S. Constitution and the Voting Rights Act of 1965.

With the release of data from the Census Bureau 2020, Mattingly and his team will load the data set, run the algorithm, and create a collection of typical North Carolina non-partisan district plans. Based on this wide distribution of maps, and taking into account historical voting patterns, they will perceive the references that railings should have. For example, these maps will assess the relative probability that different election results will result — he says, the number of seats won by Democrats and Republicans — and by what margin: 50-50 split voting and credible voting patterns, unlikely a neutral map will give Republicans 10 seats and Democrats only three (as was the case with that 2012 map).

“We’re using computational math to find out what we expected as a result of unbiased maps, and then we can compare it to a particular map,” Mattingly says.

They will know their findings by mid-September, and then expect state lawmakers to pay attention to the railings. When new district maps are proposed in the fall, they will analyze the results and contact the public and political dialogues that will arise, and if the maps are suspected to be guerrilla again, there will be more litigation, and mathematicians will once again play a major role.

“I don’t want to convince someone that something is wrong,” Mattingly says. “I want to give them a microscope so they can see a map and understand its properties and then draw their own conclusions.”

COURTESY PHOTO

When Mattingly testified in 2017 and 2019, after examining the two subsequent iterations of the North Carolina district maps, the court accepted that the said maps were excessive partisanship, discriminating against Democrats. Wes Pegden Carnegie Mellon University mathematician testified using a similar method in a case in Pennsylvania; the court accepted that the aforementioned maps discriminated against Republicans.

“The courts have long struggled to measure the gerrymandering of partisanship,” Li says. “But then there seemed to be a breakthrough when he threw out maps from court to court using some of these new tools.”

When the North Carolina case reached the U.S. Supreme Court in 2019 (along with the Maryland case), mathematician and geneticist Eric Lander, a professor at Harvard and MIT who is now President Biden’s chief scientific adviser, briefly warned that “the problem that created computer technology has caught on.” The extreme standard seemed to him to be, “Which part of the redistribution plans are non-extreme than the proposed plan?” The question he asks, “is the correct and quantitative mathematical question. right answer “.

Most justice otherwise he concluded.

“The five Supreme Court judges are the only ones who had trouble seeing how math and models worked,” Li says. “They managed to apply to the state and other federal courts; that didn’t go beyond the intellectual capacity of the courts to handle it, rather than a complex case of sex discrimination or value fraud. But five Supreme Court judges said, ‘That’s very hard for us.’

“They also said,‘ This is not for us to fix, this is for states to fix; this is Congress to fix it; it’s not for us to fix, ”Li says.

Will it matter?

As Daley sees it, the Supreme Court’s decision leaves state lawmakers “green light and no speed limits” on the partisan gerrymanders they can make when making maps in the same month. “At the same time,” he says, “technology has improved to a point where we can use it now [it] to see through the gerrymanders driven by the technologies created by lawmakers ”.