Voting Machines Cause Landslide of Problems

If this past summer’s primary elections are any indication, Nov. 7 could be a frustrating, if not scandalous, election day for the estimated 80 million to 90 million Americans who will vote.

Volunteers manning voting precincts nationwide reported dozens of problems with e-voting machines, about two-thirds of which are optical scan models (voters fill in a bubble on a paper ballot that is then scanned by a computer) or touch-screen computers.

During the spring primaries in Council Bluffs, Iowa, candidates’ names did not align with the correct bubble on the optical form, and votes were incorrectly assigned. In Montgomery County, Md., during September primaries, the board of elections failed to include 13,000 access cards in the packet of voting materials each precinct receives, leaving voters to wait many hours to vote. Problems such as these—as well as imagined ones—have caused lawsuits to be filed in nine states. Expect more, experts say. As many as 80 percent of voters nationwide are expected to use an e-voting machine this fall, up from 68 percent in the 2004 November elections, according to the Elections Data Services.

Four years after a federal law made billions of dollars available for new e-voting machines, numerous studies released this year continue to find security flaws in the machines. In September, Princeton University professor Edward Felten released a study of the Diebold AccuVote-TS voting machine, showing that his students, working in a lab setting, could install malicious code, modify votes and the logs tracking those votes, and spread viruses from machine to machine. Diebold says that model, two generations old, has been replaced by newer ones, but Felten remains a voting machine security skeptic.

The Brennan Center for Justice released a report in July that found 120 security threats, including states failing to install intrusion detection software. “Today the state of electronic voting security is not good,” said David Wagner, a computer scientist at the University of California, Berkeley, at a congressional hearing on the subject this summer. “With this technology, we cannot be certain that our elections have not been corrupted.”

The best way to double-check whether the machines tally votes correctly is for voters to also fill out a paper ballot—and for states to conduct a manual audit, Wagner testified. Only 13 states do so, however. Yet even paper ballots have proven unreliable, as Cuyahoga County, Ohio, found out this summer when it hired the Election Science Institute to study its May 2 primary election. ESI found about 10 percent of the paper ballots were “either destroyed, blank, illegible, missing, taped together or otherwise compromised.”

This underscores the biggest problem for e-voting, says Doug Jones, a computer science professor at the University of Iowa who has studied e-voting machines: Counties, which manage most of the elections nationwide, do not have the necessary managerial expertise. For example, states and counties rely on inadequate security documents written by the voting machine manufacturers.

Vendors also need to greatly improve design, given how rarely voters and workers touch machines, according to Wagner.

“You have to design the machines so things are so self-evident that learning is not necessary,” Jones says. “Until then, you can expect a lot of problems and a lot of controversy.”