E-Voting Enthusiasts Face Hurdles
By Caron Carlson
April 5, 2004

Internet voting in a U.S. presidential election is gone, but it is not forgotten. The security concerns that last month forced the Pentagon to cancel a project allowing some military personnel and other U.S. citizens to vote online from overseas in November dampened the Internet voting initiative, but it did not deter enthusiasts.

The Michigan Democratic Party, undeterred by the demise of the federal SERVE (Secure Electronic Registration and Voting Experiment) project, permitted online voting for the first time in its Feb. 7 caucus. The party hopes to repeat the experiment in four years.

"It was a success. It was secure, and that was the main focus," said Jason Moon, spokesman for the party in Lansing. "We're hopeful it can be part of our next caucus."

The system, operated by Election Services Corp., was protected by two firewalls and monitored around the clock, and it elicited twice as many votes as the 2000 caucus, according to Moon. "I'm sure there will be some changes," he said, "but no conclusions have been made."

For the computer scientists who put the kibosh on the SERVE project, the only change now needed is the abolishment of the Internet voting experiment. "People who understand security do not advocate Internet voting," said Barbara Simons, a computer scientist who served on the peer review panel that recommended shutting down SERVE in January. In Simons' view, the Michigan Democratic caucus results proved nothing.

"I don't think there was a whole lot of motivation for stealing that [Michigan] election. It was a foregone conclusion," said Simons in Palo Alto, Calif. "If [Sen. John] Kerry hadn't won, people would have known that things were screwed up."

The problem with testing Internet voting systems is that the stakes are too high in an election where there is a strong motivation for fraud, Simons said. Testing in an election that doesn't matter—as some have suggested SERVE should be tested—removes the conditions that could threaten the integrity of votes, she said.

"We have a logical problem here with Internet voting and with paperless touch-screen voting," Simons said. "There's no incentive for someone to subvert the votes in an election that doesn't count."

The SERVE project, led by Accenture Ltd., would have allowed 100,000 overseas citizens from 50 counties in seven states to vote in the primary and general elections. But two weeks before its scheduled debut in the South Carolina primary, the project came skidding to a halt. The peer review team said the system could cause automated vote buying and selling and privacy violations, and it even could reverse the outcome of elections—all of which could occur without anyone knowing.

Computer scientists never say never, but Simons and her colleagues concluded that SERVE is much too far ahead of its time.

So for now, electronic voting proponents and their adversaries have turned their attention away from the Internet and to e-voting at the polls. While Simons and her colleagues consider the security flaws of Internet voting insurmountable, they also see problems in e-voting at the polls, including proprietary software, lack of protection against insider fraud and lack of voter verifiability.

Software can contain bugs, and it can be tampered with, introducing the possibility that a printout of votes following an election does not represent actual votes in a touch-screen system, Simons said. That problem does not necessarily increase the likelihood of fraud, but it increases the appearance of the potential for fraud, she said.

"You're basically handing over your rights in a democracy to a handful of companies with secret code," Simons said. "Why is it that a bunch of geeks who are not known for being political activists have suddenly had their lives taken over by this issue? We're doing this because we care about democracy. I believe our democracy is at stake."

The computer scientists are not alone in their concern. Bills were introduced last year in the U.S. Senate and House of Representatives to improve vote verification where electronic systems are used. The legislation would require a manual recount of a percentage of votes, voter-verifiable paper ballots and open code.

Because of some state requirements, many e-voting technology providers are developing verifiable paper ballots to accompany the electronic process.