Caltech, MIT elect to improve voting system
Javier Marquez

Whether you used the ancient lever-operated machines this past election day, the more modern optical scanning and touch-screen computers, or the primitive punch cards with their infamous chads, those who study the American voting process agree: the hodgepodge of voting systems in use today is not worthy of a democracy like the United States.

Indeed, as the aftermath of the November presidential election illustrated so clearly (and painfully) to the country and the world, the American system of recording ballots is due for a major overhaul. “It is embarrassing to America when technology fails and puts democracy to such a test as it did in the last month,” President David Baltimore said in reference to the five weeks of debate and uncertainty.
Due to voter error as well as faulty balloting mechanisms, thousands of votes in the state of Florida went uncounted. At stake were the coveted 25 electoral votes that, based on the number of popular votes gained in the state, would determine whether George W. Bush or Al Gore secured the nation’s next presidency.

In a December 14 teleconference that linked the Sherman Fairchild Library with MIT, Baltimore and MIT President Charles Vest announced that the two institutes will join together to investigate the various methods by which Americans cast their ballots. Utilizing their findings, the researchers will apply their creativity and technology to developing specific guidelines and standards for improving the entire voting and record-keeping enterprise.

This effort will harness the combined ingenuity of Caltech’s Thomas Palfrey, professor of economics and political science; Jehoshua Bruck, professor of computation and neural systems and electrical engineering; and Michael Alvarez, associate professor of political science. The MIT team comprises Stephen Ansolabehere, professor of political science, and Nicholas Negroponte, chairman of MIT’s media lab.

Equipped with a Carnegie Corporation grant estimated to total $250,000, the scientists have already begun examining reams upon reams of voting records supplied to them by county governments across the nation.

“A lot of what we’re doing right now is information collection of two sorts,” Palfrey said. “One is voting data—we’re mainly trying to identify voter error, and that’s done by looking at what’s called overcounts and undercounts. We’re also looking at recounts and reversals—that is, when they change a vote as the result of a recount.”

Undercounts and overcounts, he explained, are discrepancies between the number of voters and the total votes cast. An undercount results when a voter skips or doesn’t select a candidate running for office, so that the ballot doesn’t register a vote. Conversely, when a voter selects more than one candidate for the same office, such as when he or she punches a hole next to the names of two candidates, an overcount has been committed.

“The other information we’re collecting is county-by-county-level data for the entire U.S. regarding what voting system they use,” Palfrey added. “Some use paper ballots, some use prescored punchcards or unscored punchcards. There’s electronic voting and the lever system. That data is readily available.”

Once the information is collected from all available sources, it will be subjected to rigorous analysis, Palfrey said. The scientists, along with social science PhD candidates Catherine Wilson and Tara Butterfield, will correlate rates of voter error with the types of balloting systems, across geographical lines.

“Phase one is getting the data and getting the facts: what’s good and what’s bad about these systems,” Palfrey said. “That will generate a report in four or five months and we’ll include broad recommendations.”

This initial survey will evaluate which systems can be relied on to provide more accurate counts, something that hasn’t previously been done on this scale. “As far as we can tell there haven’t been any sweeping voting studies that cover the nation,” Alvarez said. “There are ATM-style voting systems in use, and the upside is we think they will produce the lowest voter error.”
He added the caveat that a single, uniform method of voting will probably not meet the nation’s needs. Because of this country’s size, its heterogeneous population, and accessibility laws that guarantee every member of the electorate the right to vote, there will always be more than one way to cast a ballot, he said.

“In parts of the country, some of these systems just won’t work,” Alvarez said. “In Arizona recently, they had four types of voting available, including Internet voting, traditional polling places, and vote-by-mail systems. But you can’t use electronic systems everywhere. For example, there are some Navajo reservations where they have no electricity. There’s a need for flexibility.”
•