Students working with the University of Michigan’s Center for Local, State and Urban Policy (CLOSUP) have created a system to help a Michigan commission collect and analyze public comments about drawing Michigan’s congressional and state legislative districts.
By using a committee of citizens to draw district maps, Michigan is setting a precedent that other states across the U.S. can follow," Gelman said.
The story begins in 2018, when Michigan voters approved a ballot initiative that moved responsibility for drawing Michigan’s congressional and state legislative districts from the Michigan Legislature to a newly-formed Michigan Independent Citizens Redistricting Commission. Starting in 2020, the bipartisan commission accepted public input–nearly 30,000 comments–and considered communities of interest for the first time before drawing Michigan’s districts.
“This approach to redistricting instead of doing it through the legislature is a very big deal as an attempt to fix what's broken in our democracy,” said Thomas Ivacko, who recently retired as executive director of CLOSUP and managed the team of students. “When the legislature managed redistricting, it was all done behind closed doors and the new maps were designed to keep the current majority party of the legislature in power. This process is a 180 degree change – it's all about transparency and maximizing public input.”
The commission completed its first set of maps in 2021, and they were used in the 2022 election. But last year a federal court ordered the commission to redraw 13 Detroit-area districts after finding they used race as a predominant factor while redistricting, which violates the 14th Amendment.
That’s where the CLOSUP students come in. The Commission realized they needed a more systematic approach to handling the deluge of public input, and it hired the CLOSUP team that includes Edward Plaut (JD ’24), Elizabeth Gelman (MPP ’24), and Danielle Hamer (MPP ’24) to collect and categorize public comments about redrawing the 13 districts. The students compiled comments from in-person and Zoom public meetings, emails and letters, and online portals into a single database to understand patterns and trends.
“The commission received a tsunami of input, which is a huge win for the public and our democracy, but they had no system in place to handle this massive amount of information,” Ivacko said. “This project addresses that shortcoming and provides an innovative model for how to make a significant improvement for next time around.”
The system the students created is notable because it uses artificial intelligence to code comments. The student team used the U-M’s Chat GPT tool to separate comments into categories, Hamer explained. In total, they categorized more than 1,400 comments using more than 4,000 codes. An important component of the project was developing and implementing best practices for the use of artificial intelligence.
“We gave AI all of the information we had, and then we double-checked the AI’s output for each comment for possible errors,” she said. “Two years ago, this level of analysis wouldn’t have been possible.”
The AI approach was also crucial because the team had only a few weeks to produce two reports to inform the Commission’s deliberations, and the volume of public input would have been otherwise unmanageable on such a tight timeline.
The team also reported each comment without revisions or judgment, Gelman said.
“We respect the public commenters,” she explained. “We don’t know more than they know. If they’ve taken the time to comment, it’s clearly something that matters to them. It’s not our responsibility to judge whether it’s right, wrong, or something else.”
The database of comments was designed to help the commission see the larger picture, and make decisions informed by qualitative data, Plaut explained.
“How did we filter out comments that weren’t informed? We didn’t,” he said. “There’s no competency test for voting; why should we hold this process to a different standard? In truth, it doesn’t matter if a comment is imprecise. Not every commenter will have profound or even correct insight. But if enough people are saying the same things in the same way, the data will speak for itself. In the aggregate, if enough Michiganders speak about what’s important to them, we can get a good idea about how they feel about their communities.”
By using a committee of citizens to draw district maps, Michigan is setting a precedent that other states across the U.S. can follow, Gelman said. “California and New York are doing something similar, but they have much bigger budgets,” she said. “This is a great model for smaller states with smaller bureaucratic structures.”
Overall, moving away from having state legislatures drawing district maps is an important step toward strengthening democracy in the U.S., Plaut said. “I think it’s pretty easy to understand when state politicians get to draw district lines, it affects election results, and that’s not fair,” he said. “It misaligned incentives in the most obvious way possible.”
Written by Sheri Hall