The legal standard set in Rucho v Common Cause puts the onus on the plaintiffs in a Voting Rights Act (VRA) case to demonstrate that a map is a racial rather than a partisan gerrymander in order for it to be struck down as unlawful. This means plaintiffs have to 'do the homework' of the defendants for them: they must produce a map that is at least as politically gerrymandered as the defendants drew, while achieving better racial representation outcomes. This standard was successfully met by a group of mathematicians in the El Paso redistricting case, winning in federal court before the Supreme Court shadow docket put a hold on the decision, allowing the racial gerrymander to go through. The mathematicians who accomplished this used Monte Carlo Markov Chain (MCMC) methods to draw a so-called 'ensemble' of millions of alternative congressional maps for Texas, and used statistical methods to show that the enacted map was an outlier for racial representation among the maps that had similar partisan outcomes. In this talk I will go over these mathematical details, as well as describe the computational tools currently used in this field. Note: If you have a working Jupyter Notebook installation, you are encouraged to bring it to the talk, as you will have the opportunity to follow along the computational demos included in the talk.
The Mathematics of Redistricting—what they did in Texas
The Mathematics of Redistricting—what they did in Texas