Unfortunately, the machines have "a fatal disadvantage," says Rep. Rush Holt of New Jersey, who's sponsoring legislation on the issue. "They're unverifiable. When a voter votes, he or she has no way of knowing whether the vote is recorded." After you punch the buttons to choose your candidates, you may get a final screen that reflects your choices - but there's no way to tell that those choices are the ones that ultimately get reported in the final tally. You simply have to trust that the software inside the machine is doing its job.
It gets scarier. The best minds in the computer-security world contend that the voting terminals can't be trusted. Listen, for example, to Avi Rubin, a computer-security expert and professor at Johns Hopkins University who was slipped a copy of Diebold's source code earlier this year. After he and his students examined it, he concluded that the protections against fraud and tampering were strictly amateur hour. "Anyone in my basic security classes would have done better," he says. The cryptography was weak and poorly implemented, and the smart-card system that supposedly increased security actually created new vulnerabilities. Rubin's paper concluded that the Diebold system was "far below even the most minimal security standards." Naturally, Diebold disagrees with Rubin. "We're very confident of accuracy and security in our system," says director of Diebold Election Systems Mark Radke.
Newsweek discovers the chaos in US voting machines. At last.