Ensuring that elections are fair and equitable is fundamental to democracy — yet easier said than done, as MIT students discovered this fall in a new class, co-listed as 6.S897 and 17.S952 (Elections and Voting Technology).
Taught jointly by Charles Stewart III, the Kenan Sahin Distinguished Professor of Political Science, and Ronald Rivest, the Vannevar Bush Professor of Electrical Engineering and Computer Science (EECS), the class explored challenges embedded in election systems from both the technical and the political science perspective — providing students with new insights into the complexities of a system many thought they understood.
“I had this idea we could fix voting so easily by using electronic voting machines, but to learn there are huge security concerns was really interesting for me,” said Megan Goldberg, a PhD student in political science whose research focuses on political behavior.
“You come from the technical side and think you can just throw some standard piece of cryptography at the problem, and then you start to understand all the constraints,” said Ben Kraft ’15, a mathematics major. For example, votes need to be kept secret while being tracked in such a way that no one can vote twice; the system must be easy to use; and the technology must be robust enough that it will not fail on election day. “You add all of these things together and [the problem] becomes more interesting,” Kraft said.
Voting Technology Project
The fall class was an outgrowth of Stewart and Rivest’s long collaboration on the Caltech/MIT Voting Technology Project, a research and policymaking initiative that emerged following the 2000 presidential election, which was plagued by charges that ballot and voting machine errors had upset results. The class covered many of the challenges presented by modern election systems — from the fairness of redistricting to the readability of ballots to the security of electronic voting.
Guest lecturers added depth by sharing their expertise on a wide variety of topics. For example, Philip B. Stark, chair of the Department of Statistics at the University of California at Berkeley, discussed machine auditing of vote tallies; Dan Wallach, professor of computer science at Rice University, spoke about the design of a new voting system, StarVote, to be used in Travis County, Texas; ballot designer Dana Chisnell explored issues related to usability; and David Jefferson, security researcher at Lawrence Livermore National Laboratory, discussed the risks of voting over the Internet.
Rivest pointed out that classes like this one are useful for future engineers and scientists because they provide a broader context for their work. “Computer scientists are increasingly seeing systems they work with being involved in socially impactful situations, and voting is just one of them,” he said. “Understanding how engineering work fits into this fuller context is tremendously important [for engineering students] as they move forward in their careers.”
Election-day observation
Timed to coincide with the U.S. midterm elections, the class also gave students the chance to see voting in action and to learn just how messy even a rather well-run election can be. Kraft, for example, found that despite all the talk in class about the need for privacy, only about a third of voters he observed in Cambridge bothered to use an envelope to shield their ballots.
Goldberg noted that at the Boston polling site she visited, the ballot box filled up much more quickly than expected. “Massachusetts law suggests you should only empty the box at the end of the day, but I was there two hours and they emptied it twice,” she said.
“It’s interesting to see how people try to deploy the technologies we’re discussing,” said Marco Pedroso ’14, whose research focuses on election security issues. Noting that this class was very different from his usual graduate-level classes in EECS, he added, “Political science talks more about implementation.”
Evaluating real-world election issues
Indeed, real-world election issues were at the heart of the team projects that wrapped up the class. Inspired by a dispute that arose in Wisconsin in which Republicans claimed that a new ballot would favor Democratic candidates, for example, Goldberg and her teammate, EECS graduate student Alexandra Hsu, compared election results to measure the impact of different design choices.
The pair found no merit to the claim of partisanship but did find an uptick in votes for candidates whose names appeared at the tops of columns. The students are planning to research this matter further during Independent Activities Period.
Other student teams investigated the metrics of gerrymandering; Brazil’s ballot numbering system; how best to determine the margin of victory in a proportional voting system; cryptographic techniques for shielding voting data; and using an algorithm to draw legislative districts.
Experience in interdisciplinary collaboration
Reflecting on their experience at the end of the term, both professors and students said the interdisciplinary class had helped them see elections from new angles.
“I think that political science brings in a concern for the particularity of modern elections. Computer science is more involved in abstract data,” Rivest said. For that reason, he said, teaching the class jointly proved “an interesting exercise.”
Stewart added that “the other interesting thing that comes together in a class like this is that computer scientists are very concerned that a voter can be assured of the accuracy of the count — without trusting anyone — whereas election officials are in a world where at the end of the day you have to trust people.”
Sharing and analyzing multiple forms of knowledge and perspective was part of what made this new class offering special. The course — which attracted students from a range of majors — also reflected a burgeoning awareness that many major issues are best addressed by collaborative teams that include experts from the STEM (science, technology, engineering, mathematics), humanities, arts, and social-science fields.
“This was a great opportunity to learn what other students are doing and also to see across departments that we have common research interests,” Goldberg said. “That was probably my favorite thing about this class.”
_______________________________________________________________________________
