Ballot Roulette: Computer Scientists and Mathematicians Look for Better Ways to Vote

Article excerpt

Two months ago, in primaries for governor and congressional and state legislative seats in Maryland, many trips to the polls became painful experiences. At hundreds of precincts in Montgomery County, for instance, new touch-screen voting machines sat useless for lack of plastic authorization cards needed to operate them. In many polling places, electronic poll books with lists of eligible voters froze or mistakenly claimed that new arrivals had already cast their ballots.

Maryland governor Robert L. Ehrlich Jr. has called for a return to paper ballots and is urging voters statewide to cast paper absentee ballots for next week's general election to avoid the computerized machines in polling places.

In Illinois in March, hundreds of precincts in Cook County reported difficulties with their electronic-voting systems. Snafus with electronic systems have also plagued contests this year in Iowa and Arkansas, not to mention the 2004 election, in which problems with electronic machines occurred in Ohio, North Carolina, Florida, and other states.

The technologies that underlie the U.S. voting system have undergone a huge change in the past 6 years. According to the Washington, D.C.-based Election Data Services, a company that tracks voting-machine trends, the percentage of citizens using computerized-voting machines has climbed from roughly 12 percent in 2000 to an expected 38 percent in this Tuesday's election.

Although the machines have gotten a bad rap, human foibles contributed to the recent problems, and the electronic systems are in some ways an improvement over older technology. But whether they are the best option remains to be seen, and the search for the most practical and secure voting technology goes on.

"Five to 10 years ago, computer scientists weren't paying attention" to the technology used in voting, notes computer scientist David A. Wagner of the University of California, Berkeley.

However, newly aware of the stakes, risks, and intellectual challenges associated with voting equipment, computer scientists and mathematicians specializing in encryption are now avidly taking part in the search for dependable and inviolable voting technology. These researchers are investigating existing systems, devising ways to improve them, and inventing entirely new approaches.

"In the long term, the goal is to ... make a voting system that's more reliable and secure than what we have now or have ever had. I think that's a very feasible goal," says computer-security specialist Edward W. Felten of Princeton University.

OPEN SESAME The technological transformation now under way in polling places has its roots back in 2000. That's when the close and pivotal presidential vote in Florida focused national attention on voting-system flaws. Those flaws included technological ones, such as confusing ballot layouts and balky punchcard ballots (remember "butterfly ballots" and "hanging chads") that made many voters' intentions uncertain.

Identifying the 2000 election debacle as partly a technology failure, Congress in 2002 passed the Help America Vote Act (HAVA), which pledged $3.9 billion to the states for modern voting equipment, voter education, and other election reforms. Under HAVA, many electoral districts across the country have purchased electronic-voting machines to replace punchcard equipment and mechanical voting machines. The electronic machines typically either scan a paper ballot that was marked by hand or record voters' selections made by means of buttons, a dial, or a touch screen.

The latter class of devices, known as direct-recording-electronic (DRE) machines, is the newer of the two electronic approaches and the one that's attracted the most criticism for reliability problems. But operational breakdowns aren't the only cause for concern. Several analyses dating back to 2003 have identified security vulnerabilities in DREs that could allow an attacker to secretly alter vote tallies or disrupt polling. …