If quirky ATM machines randomly erased the accounts of 1 percent of their patrons, they’d quickly be taken out of service. Yet touch-screen electronic-voting machines that recorded no vote for 1 percent of voters in Florida’s Democratic primary in March are set to be used again in nearly a quarter of the state’s 67 counties. And Florida is hardly alone in switching to this relatively untested technology: 20 percent of the nation’s 3,114 counties will use electronic voting machines in November. That rate of adoption alarms some experts. “We’re not ready,” warns Douglas W. Jones, a computer scientist at the University of Iowa in Iowa City. “I’m suspicious if we’ll ever be ready.”

In Florida’s case, it’s unlikely that one in a hundred people who went into electronic polling booths in March deliberately cast no vote, given that voters using ballots read by optical scanners – like those used to score standardized tests – cast far fewer so-called undervotes. But if the machines were at fault, no one will ever know for sure, since they produced no paper trail or independently auditable record. And that’s one of the biggest concerns about current electronic voting technology, says Martha Mahoney, a professor of law at the University of Miami and a member of the Miami-Dade Election Reform Coalition, an activist group. The machines “lack transparency for the voter,” she says. “The voter casts a ballot but doesn’t see how the ballot is recorded.”

Computer scientists would relish the opportunity to help guard against errors or tampering by scrutinizing these electronic voting machines for bugs and vulnerabilities to hacking. But most electronic voting machines run proprietary software, which manufacturers are reluctant to release to any except their own testing authorities. Consequently, there is no independent verification that the code works as advertised and that nobody has tampered with it, says Michael Wertheimer, a cryptologist at computer consulting firm RABA Technologies in Columbia, MD.

Some experts contend there are ways to make electronic elections safer while still maintaining voter anonymity. Ted Selker, a professor at the MIT Media Lab who directs the Caltech-MIT Voting Technology Project, points out that many electronic voting machines are already equipped with headphones to help vision-impaired voters. This feature, Selker says, could be used to verify all voters’ ballots; software independent of the voting machine could read off the name of each candidate a voter selects, and a tape of the day’s votes could serve as an independent record in contested elections.

But while ideas like these could help make future election results more trustworthy, it’s probably too late to replace or modify the equipment counties will put into action on November 2. Indeed, laments Wertheimer, with so little time remaining and so much money already spent on new voting equipment, many election officials don’t want to hear that the technology may be flawed. “With every day that passes,” he says, “people will say, ‘So what? We can’t change anything before the election.’”