E-voting goes (mostly) smoothly, but suspicions, critics remain
BY Michael Hardy Federal Computer Week
Published on Nov. 8, 2004
Electronic voting technology got its first major test in the Nov. 2 election, and for the most part, the machines seemed to work well.

Voters and activist groups nationwide reported a slew of problems, such as machines failing to boot up or freezing while in use. Some voters using touch screens said the machines refused to accept their votes for particular candidates or failed to offer complete ballots that included local races.

In certain cases, machines that failed to start forced polling places to turn voters away, said Cindy Cohn, legal director of the Electronic Frontier Foundation, a Washington, D.C., interest group.

Some potentially serious problems were reported late last week, as election officials continued counting and checking votes. One Ohio precinct, where only 638 ballots were cast, reported 4,258 votes for President Bush, according to an Associated Press report. Officials who found the error said Bush actually got 365 votes to John Kerry's 260.

Equally troubling, voters in some precincts claimed that the machines showed them summary screens reflecting different votes than the ones they had cast, but they were able to change to their preferred candidates.

"The thing that is worrisome about these [complaints] is that we've seen it across all systems," Cohn said. "It doesn't seem to be limited to one [brand] of machine." She said many voters may not have noticed that their confirmation screens listed incorrect choices.

Many of those who believe the e-voting systems are still too prone to errors and vulnerable to tampering said the technology did not prove itself Nov. 2.

"A lot of people are missing the big picture," said Aviel Rubin, a critic of e-voting and a computer science professor at Johns Hopkins University. "They're focused on whether the machines worked" from the voter's point of view.

But if something affected the final vote tallies, whether it was deliberate or accidental, voters would almost certainly never notice, he said.

"A smart person who wanted to rig the election would do it for a close race," he said. "This was the perfect opportunity. We don't have any evidence that it happened, but we don't know."

Rubin co-authored a 2003 report detailing security vulnerabilities in one version of Diebold Inc.'s Diebold Election Systems' voting machine software. Diebold is one of several firms that develops software for touch-screen voting machines and the back-end tabulation systems. Rubin has been a proponent of enabling the machines to create paper records that voters would verify before leaving. Polling places could use the records if a manual recount is needed.

Only Nevada officials require the paper trail on touch-screen machines. That may change before the 2006 midterm elections.

"I think we have an opportunity before us to do a lot before the '06 election, and I'm going to devote myself to that," Rubin said. "Unfortunately, we get a lot of resistance from election officials and vendors who are using this election to argue that the machines worked perfectly and should always be used in every election."

David Bear, a spokesman for Diebold, argued that the Nov. 2 election experience validates electronic voting.

"This election did a lot," he said. "You see overwhelming acceptance, even preference, from voters. I think the national dialogue is good. Where the problem comes up ? and, unfortunately, I think this happened some before this election ? is when the dialogue serves only to confuse and frighten voters."

Before the election, some computer hackers had threatened to corrupt vote tallies in some obvious way to prove the technology's vulnerabilities, but that did not happen, said Herbert Thompson, director of security technology and research at Security Innovation Inc.

"I think that the technology just isn't ready yet to depend on it as our sole and official record of the will of the people," he said. "I think that eventually we'll have systems that are provably reliable and provably secure. But we're not there yet."