Technical Knockout

by Kelen Tuttle

Computer voting technology is not the fix-all we believe it to be.

One needs only to mention the dubious results of the 2000 election to elicit nightmares of the ambiguous hanging chad, the lack of paper trails, and accusations of poor security and immoral election officials. And most of us are sick of hearing about it. But there’s still one problem so fundamental to the future of American democracy that it’s worth revisiting: that no matter how verifiable the paper trail or hacker-proof the hard drive, computer voting technology is worthless unless it enables voters to properly record their intentions.

Nearly 30 percent of precincts in the U.S. own electronic voting machines, and during each election a small but significant percentage of voters using them fail to vote as they intended. A clear demonstration of this phenomenon occurred in Wellington, Florida, during a 2002 mayoral runoff. With only one race on the ballot, 2522 people showed up at the polls, and 78 of those—a full three percent—failed to record any vote at all. To even the most naive observer, that seems highly improbable. More likely, these voters misunderstood how to record their votes into the machines, quite possibly altering this election’s close outcome.

This story is not unusual. According to an MIT study, it is commonplace for this many voters to accidentally misvote using electronic voting machines. In a laboratory test, the MIT researchers found that three percent of the participants who believed they had successfully voted for one candidate using commercially available machines either mistakenly failed to vote or voted for another candidate without ever realizing their mistake.

This need not be the case: the makers of electronic voting machines could easily redesign the electronic ballot format to reduce voter error. Right now, more than two-thirds of machines use ATM-like touch screens with buttons much too small and close together, especially the “cast ballot“ and “next screen“ buttons, says Whitney Quesenbery, president of the Usability Professionals Association and member of the U.S. Election Assistance Commission. This invites voters to push the wrong button, accidentally submitting their ballots before having the chance to vote in all elections. Quesenbery also dislikes the practice of displaying more than one race on each screen. The many boxes separated by only a thin black line may confuse voters, who may not understand which candidates belong in each race. But so far, the largest voting machine manufacturers, Diebold and Sequoia, have neglected to redesign their touch screens.

In response to the alarming number of misvotes, MIT media arts and sciences researcher Ted Selker spent the past two years designing new software that reduces voter mistakes with a simple and intuitive design. According to Sharon Cohen, one of Selker’s graduate students, less than half a percent of participants failed to vote or voted for the wrong candidate using Selker’s interface.

Unlike commercial models, Selker’s prototype displays a list of the candidates for each race on its own page, with tabs that link to the other races lining the bottom of the screen. After the voter has clicked on his preferred candidate, the corresponding tab changes color, indicating a successful vote. The voter then continues selecting tabs until he has finished voting. This system both prevents overvotes and limits the number of accidental abstentions. Although the program allows a voter to abstain from any race, the color change makes it obvious when he chooses to do so. The tabs also make it simple for a voter to return to a specific race at any time during the voting process to confirm or change a vote.

Unlike any commercially available machine, Selker’s design also requires the voter to view and approve a screen listing of all of his or her selections. “In testing, we found that few voters bother to view a review screen if it’s not required,” said Matthew Hawkenberry, a graduate student in Selker’s lab. By making the review screen mandatory, Selker’s interface gives voters one more chance to check the accuracy of their ballots.

Although MIT’s design is paying off in the lab, it’s useless unless voting machine manufacturers integrate Selker’s refinements into their machines. These companies are reluctant to spend the time or money needed to install such an upgrade, says former Sequoia consultant Craig Short. Because they already have five to ten year exclusive agreements with counties around the U.S., manufacturers have no financial motivation to redesign their software to take Selker‘s innovations into account.

Although voting machine manufacturers know precisely how to solve the problem of mistakenly cast ballots, American voters continue to use inferior technology. A small but significant percentage of votes continue to be cast in error in elections across the United States. But it doesn’t have to be this way. It might cost a bit more money, but voting machine manufacturers need to recognize their inferior product and upgrade their machines to fix the problem. Aren’t our elections worth it?