On Tuesday, 4 out of 5 voters across America will use touch-screen machines or optical scanners to cast their ballots. More than one-third of the electorate will vote electronically for the first time, raising questions about whether voters, machines, and poll workers are up to the task.
After the disputed 2000 Presidential election and Florida's infamous "hanging chad" punch-card recount, the federal government allocated $3 billion to help states buy new voting machines and created the Election Assistance Commission to oversee the sweeping change.
"I'm confident with all the machines that are being put out there, whether it's one company or another," says EAC Chairman Paul DeGregorio. "In the end, you can trust the results."
But states with the new machines are hedging their bets. According to the non-partisan Electionline.org, 18 of the 33 states using touch screen machines require them to generate a paper trail – a kind of cash register scroll covered in clear plastic – that voters can see and verify before they punch the "vote" button that records their vote. Nowhere does a voter receive a printout of their vote like an ATM receipt.
To the ire of many voting watchdog groups, 15 states using touch screen machines do not require a voter verifiable paper trail.
Changes — Too Much Too Soon?
Electionline.org Director Doug Chapin found half the states that made the switch to machines did so in the past two years – potentially increasing the chance of mishaps. Connecticut, for example, will be using optical scanners in 25 towns for the first time on Election Day.
"Lots of jurisdictions around the country have been asked to swallow a lot of change in 2006, and many of them are going to suffer indigestion," Chapin says.
In fact, the more low-tech optical scanners are the most popular machines. Fifty-six percent of U.S. counties, with 49% of the nation's registered
voters, will be using optical scan equipment, while only 36% of counties, with 38% of the voters will use the touch screens, according to Election Data Services.
With scanners, a voter typically fills out their choices with a pen on a paper ballot – typically ovals as on a standardized test. The scanner counts the votes and stores the paper ballots, forming an automatic paper trail for a recount.
Fears the machines may fail has driven up voter anxiety. So have reports of machine snafus in primaries this year — paper jams, power outages, service issues, and setbacks tabulating results.
In Maryland, the governor and his challenger asked supporters to vote by paper absentee ballots, and the state ordered enough for half its voters. In Colorado, a judge ruled its states machines should never have been certified.
Ohio-based Diebold, the nation's leading vote machine manufacturer, has supplied 38 states with 136,000 touch screen units and 24,000 optical scanners.
"Diebold does not control the elections. We provide the election equipment," says Diebold Marketing Director Mark Radke.
During a show-and-tell at company headquarters, Radke demonstrated how a plastic voter access card is ejected and voided so you cannot vote again, and how a security sticker covers the slot for the machine's memory card.
"Even if that memory card disappears for some reason, you still have all the votes that have been cast on that machine in your internal, non-volatile memory," Radke says.
Radke likes to point out that machines help voters avoid mistakes like "overvoting" – picking two candidates – and "undervoting" – failing to pick anyone for an office – because the machine prompts a voter to recheck such choices.
Although touch screen machines are neither connected to one another or the Internet, if machines were to switch votes due to sabotage or tampering, you might never know.
"I think we need to replace them and not use systems that are requiring trust in software that can never be trusted," says John Hopkins University computer scientist Avi Rubin, author of "Brave New Ballot," a critique of the machines, particularly Diebold's.
"If the systems go down because of something like a power outage or a power surge or a glitch, it's very possible that event would cause the hard discs and the flashcards inside of them to fail," Rubin says.
Princeton University computer science professor Edward Felten has demonstrated a machine could be vulnerable to an electronic Trojan horse — a computer virus that could steal votes and tilt an election. Felten obtained an old Diebold machine and tampered with its memory card in his office, causing votes in a mock presidential election to switch from George Washington to Benedict Arnold.
"A single person might be able to steal an election in a nearly undetectable way in a real election in the United States," Felten says. "Either you switch to a backup system, like a low-tech paper ballot, of you just cross your fingers and hope for the best."
Troubling to some experts is how voting machine vendors – including ES&S and Sequoia, each supplying more than 15 states — chose and pay so-called Independent Testing Agencies to certify their machines. Only three labs do most of testing – Systest in Colorado, Ciber and Wyle, both in Alabama – to determine whether machines satisfy government standards.
"I think that our current testing model is all wrong. We have some standards that have been produced that are not adequate enough to have a secure system," Rubin says.
10 Percent Failure Rate?
Carnegie Mellon University computer science professor Michael Shamos says the federal standards are too low, for example permitting a "mean time between failure" of only 162 hours. Under that standard, you could expect 10 percent of the electronic voting machines to fail, Shamos says.
Next year, the EAC plans to impose tougher standards. Chairman DeGregorio says he is far more worried about human errors than mechanical ones. Counties have been holding a plethora of classes to make sure the nation's 1.4 million poll workers, whose average age is a very un-computer generation 70, are properly trained to manage the machines.
"The concern with new technology is not so much whether the machines work, but whether the people operating them can make them work," Chapin says. "We've seen problems across the country with poll workers have trouble getting machines started, getting results out of them at the end of the day, with voters having trouble making their selections or verifying that their selections are correct."
Some elections officials admit glitches –either computer or human – often go unreported.
"Folks are scared about letting you know what happened. It is very public process." Alice Miller, Executive Director, Washington, D.C. Board of Elections and Ethics told a recent EAC hearing. "They don't want their mistakes to be out in a public forum."
Not helping matters is the sheer hodgepodge of voting machines in state after state. Only a handful deploy the same machine model statewide. California, Pennsylvania, and Virginia use six different types of machines.
"The learning curve is very high," says Lehigh County, Penn. Elections chief Stacy Sterner. And so is the scrutiny, with thousands of partisans and self-appointed watchdogs some armed with video cameras, signing up to be poll observers. "I worry about people coming in and trying to catch them in a mistake," Sterner says.
"A lot of people don't like change," says retiree Dee Hauze, a Pennsylvania poll worker. "Once everybody gets used to it, it's gonna be very good, because you can't make a mistake, and if you do, you can erase it."
In bellwether Ohio, where law dictates the paper trail carries the day in a recount, 47 of the 88 counties contract with Diebold. Stark County is confident the machines will work, having conducted five recounts since it began using the machines.
"We've had to count the paper audit trail and compare it to machine tallies, and not once has a single vote varied," says board of elections director Jeffrey Matthews. "People need to take a deep breath and rely on the professionalism of election officials."
Diebold's Radke admits he'll be holding his breath on Election Day. "Me and a lot of politicians." he says.