My past few weeks have been filled with thoughts of redistricting – I gave a talk at MathFest titled “Ready for redistricting 2020?” and, then, spent an exhilarating week at Tufts at the Geometry of Redistricting Workshop. This workshop was organized and hosted by the Metric Geometry and Gerrymandering Group, a Boston-based team of mathematicians run by Moon Duchin of Tufts University. The team did an extraordinary, amazing, marvelous, phenomenal (your choice of superlative adjective) job of organizing the five-day event. The logistics were complicated. And, how great it was to be with dozens of experts, all interested in the topic, but bringing varied backgrounds and training with them. We were mathematicians, statisticians, computer scientists, and operations researchers; we were economists and political scientists; we were civil rights lawyers and legal experts on redistricting. We ranged in age, and we work in academia, at law firms, in the private sector, and for non-profits. [At left is Robert Cheetham, President and CEO of Azavea, talking about their product DistrictBuilder.]
As the workshop website says, the first three days were open to the public, and the last two were spent in specialized training (a teaching track, a computing track, and an expert witness training track). The public days were filled with morning lectures and afternoon deep delves into topics ranging from computational, to social, to mathematical, to legal.
The morning sessions the first three days were so popular that they were held at the historic Somerville Theater at nearby Davis Square. The theater opened in 1914 and, according to its website was “designed for stage shows, vaudeville, opera, and that new fad – motion pictures – the theatre was only one of the Hobbs Building attractions, which also included a basement café; bowling alley and billiards hall; ten ground floor storefronts; and the Hobbs Crystal Ballroom on the second floor.” It was fun to see workshop participants in such a splendid space, enjoying popcorn and frozen thin mints (you know who you are!). And, Moon had to clear us out of there immediately after our talks were over at 12:45 so that we did not hold up the early matinee viewing of Dunkirk each day.
So, why were we brought together? First, the next round of redistricting will commence soon after the decennial census and subsequent reapportionment of Congress. Second, the Supreme Court has announced that it will consider whether or not Wisconsin’s congressional maps have been gerrymandered on partisan grounds. Both are BIG deals.
Advocates in Washington are concerned about the upcoming census – the President’s proposed FY18 budget includes a 2% increase in Census Bureau funding, considered “woefully inadequate” to enable the bureau to prepare for the 2020 census. Previous increases in the eighth year of the ten-year census cycle have ranged from 61 to 96%.
Why are Wisconsin maps under scrutiny? In 2012, Republicans won 60 out of 99 seats in the State Assembly while only receiving 48.6% of the statewide vote. In 2014 and 2016, Republicans extended their advantage. The Supreme Court has resisted setting a standard by which to decide that a partisan gerrymander is unconstitutional, but they may do that in this case. Justice Ruth Bader Ginsberg has said that this is the most important case the court will hear next term. Indeed, if the Wisconsin map is found to be unconstitutional on the partisan argument, partisan map-drawers’ hands will be tied in the next round of redistricting following the 2020 census.
Of course, I wouldn’t be writing this if mathematics were not relevant and, indeed, redistricting uses a lot of math. Measures of compactness are by now considered traditional for detecting gerrymandering (though not agreed-upon nor well-understood); the newish efficiency gap is a central character in the Wisconsin drama; and computer simulations (to generate lots and lots of maps that could be used) are being used and developed, and may point the way to some future clarity for measuring partisan bias in the redistricting process.