In the weeks following the disputed 2000 presidential election, Riverside County Registrar Mischelle Townsend became a celebrity of sorts. Media from around the nation converged on her California county, touting Townsend's brand-new touch-screen voting machines as a vastly superior alternative to the troublesome old punch-card machines and their infamous hanging chads.
These days, though, Townsend finds herself defending the electronic machines against charges that they are vulnerable to security attack. Critics contend that without a voter-verified paper trail, the new machines are not trustworthy.
Once viewed as a panacea, electronic voting machines are now a lightning rod for controversy about how to keep election counts clean and honest. Congress passed the Help America Vote Act in 2002 to speed up modernization of the election process, but despite authorization of $3.9 billion in federal funds to assist states with the job, only a small portion of the money has been disbursed.
Even so, many states have made progress since the 2000 elections. According to Election Data Services, 309 counties, representing 12 percent of the voting-age population, used electronic voting machines in the 2000 election. This year, 669 counties, representing nearly 29 percent of the voting-age population, will use them. Some other counties have new machines that are not electronic, such as the optical-scan devices in the state of Michigan. "Many voters will return to the polls in 2004 and see changes," says Doug Chapin, director of the Election Reform Information Project. "Almost as many will return and see no changes."
Of the changes, electronic voting machines have received the lion's share of attention. Since the 1960s, there have been murmurs of dissent about any use of computers to count or record votes. But security concerns drew national attention in 2002, when Professor David Dill of Stanford University organized other prominent computer scientists via his Web site, www.verifiedvoting.org.
They expressed concern that electronic machines were an open invitation to malicious hackers, whether they be mischievous teenagers, foreign agents or renegade insiders. Meanwhile, programming code from Diebold, one of the largest vendors of electronic voting machines, appeared on the Internet. Dill contacted Avi Rubin, a professor at Johns Hopkins University, and asked him to conduct a security analysis.
Rubin's report provided additional fodder for making machine security a nationwide issue. He noted many points of vulnerability that might be exploited by either insiders or outsiders, some as simple as the fact that all the security pass codes were set to the same PIN (later found to be 1111). Other flaws included the potential for counterfeit key cards, improper use of cryptography and the possibility for hacking while votes were being transmitted from precinct to central counting locations. "It was one of the worst I've ever seen," says Rubin, who specializes in evaluating security systems. "There were mistakes that my first year students wouldn't make."
WHERE'S THE PAPER?
Diebold, the election machine vendor, didn't help its own credibility when the company's server was hacked and internal e-mails were released, one of which mentioned charging Maryland "out the yin-yang" if they insisted on adding a paper trail to their Diebold machines. Diebold CEO Walden O'Dell, a fundraiser for President Bush, also pledged to "deliver [Ohio's] electoral votes to the president next year," arousing immediate suspicion from conspiracy theorists. Diebold countered Rubin's charges by noting that he had served on the board of VoteHere.net, one of Diebold's competitors.
This only emboldened Rubin, Dill and 400 other computer scientists who signed an online petition arguing that, given the security dangers inherent in electronic voting, it is naive to think an electronic system can ever be managed securely. The only way to be safe, Dill says, is to have each machine print out a paper ballot after each vote is cast. The voter would review the ballot to make sure it is correct, and deposit it for eventual counting.
Dill also served on a California task force that studied the security of touch-screen voting. The task force came out sharply divided, declining to make a recommendation on the paper-trail controversy. After 6,000 people contacted Secretary of State Kevin Shelley, he issued an order requiring paper trails in every California county by 2006. Dean Heller, Nevada's secretary of state, quickly followed suit, moving up the deadline to require paper trails in time for the general election this November.
For Shelley and Heller, the problem with electronic voting machines is not just security but also voter confidence. They believe that a barrage of news stories pointing out security flaws has shaken public trust in electronic voting and that paper trails may be the only way to remedy the situation.
A few states have seen bills introduced in their legislatures to require paper trails, but none has passed thus far. An amendment to HAVA is also pending in Congress to require paper trails nationwide. That bill has attracted the support of 128 co-sponsors.
The paper trail advocates make a passionate and convincing case. But there is another side. Many voting officials, election machine vendors and some computer scientists complain that paper trails constitute an unnecessary complication and may create as many problems as they solve.
"I think that the paper verification system is kind of giving people a false sense of security," says Harris Miller, president of the Information Technology Association of America, which includes the Election Technology Council, a new trade group for voting machine manufacturers. Miller argues that the security issues presented by introducing paper back into the process dwarf those of electronic voting. "I can give you a receipt," says Miller, "but if I started out the day by stuffing the ballot box with 50 ballots for Bush, I haven't actually done anything to make the system secure."
Michael Shamos, a Carnegie Mellon computer science professor who has been certifying election machines for the states of Pennsylvania and Texas for more than 20 years, thinks many of his peers are blowing electronic security flaws far out of proportion. He points out that lever machines produce no paper trail, and that optical-scan and punch-card machines also rely on computers for counting.
Shamos agrees that the electronic system has flaws but insists that the likelihood of their being exploited in an actual election is very small. Research conducted for the state of Maryland and the Ohio secretary of state concluded that there are real problems with electronic voting, but they can be fixed. "Merely because you can mouth a scenario doesn't mean that we should shut down electronic voting completely," Shamos says.
It is not entirely clear whether paper would provide an additional layer of security. While a paper receipt would convince voters that the machine did record their vote as cast, that paper might be more vulnerable to manipulation than electronic machines. Theoretically, he says, subtle changes in the timing marks for optical-scan ballots could invalidate crucial votes for a particular candidate. "If you want to outlaw something, you should outlaw all uses of pieces of paper," Shamos says. "The idea that paper is somehow safer or more secure is ludicrous."
For many election officials, concerns about re-introducing paper to the process are primarily logistical. Would the first count come from the machine or the printed receipts? Which would represent the true count? Would the paper be counted by humans or by another machine? How could people be prevented from walking off with their receipts, and then selling them? How would the paper be kept secure? What would happen to all the paper after the election was over?
Even without the answers to those questions, election officials know a few things for sure. Machines equipped with paper trails cost significantly more. A paper trail requires printers. And introducing additional machinery into an already hectic polling place staffed by elderly poll workers makes them very nervous. "If we had to inject printers, we would have made banner headlines that made Florida pale in comparison," says Riverside County's Townsend, who will have to install a paper trail by 2006 if the secretary of state's plan stays on track.
Nevada is confronting some of those logistical issues as it pushes to implement a voter-verified paper trail system by November. No such system has been evaluated as yet by the federal government, so the state is planning to get private consultants to do a security assessment. But because of relatively late notice, printers for the paper trails aren't supposed to arrive until August--just days or weeks before early voting begins--with little time for poll-worker training. If the printers are late, Secretary of State Heller has announced that he will not use the electronic machines without a paper backup and will instead revert to optical-scan machines. But it might not be easy for local jurisdictions to print the optical-scan ballots in time.
Voting officials in Georgia have also spent a fair amount of time dealing with the paper-trail controversy. Shortly after the 2000 election, Georgia Secretary of State Cathy Cox announced that she would be purchasing electronic voting machines for the entire state, primarily to reduce a 2.5 percent rate of ballot spoilage that was recorded in the 2000 election. With the support of then-Governor Roy Barnes, Cox was able to secure the funding, purchase Diebold machines and put them in place by November 2002.
Georgia's first electronic election went fairly smoothly, with only minor mechanical glitches from the voting machines. The percentage of spoiled ballots was reduced to less than .09 percent, and about 90 percent of voters expressed high satisfaction when asked about the machines in independent surveys. But shortly after the election, the state started to take heat about security flaws and Diebold's missteps. Governor Barnes and U.S. Senator Max Cleland, both Democrats, were unexpectedly voted out of office during that election, and a theory circulated that the results had been tampered with to benefit Republican candidates.
Cox's office insists that an analysis of voting patterns would show that there was no tampering and that Diebold was a good choice as vendor. "Unquestionably, things have been said that we would have preferred not to see," says spokesman Chris Riggall. "But at the end of the day, we have no regrets in terms of the performance of the system."
SHOW US THE MONEY
Controversy aside, Georgia has at least met its commitment to polling place modernization. Few other states can make that claim. The Help America Vote Act set a deadline of November 2004 for all states to complete their election reform effort. In fact, many states will just be getting started then on some of the major components of the updating.
States have not been completely idle, however. They have made changes on most of the items that did not require much money. Almost all states will have provisional ballots this November, for example, meaning people who show up at the polling station but are not listed on the voter rolls will still be allowed to vote, and have their eligibility determined at a later date.
But on the expensive items--statewide voter databases and new voting equipment--many states have been reluctant to proceed without guaranteed money and federal guidance on voting machine standards. Although HAVA authorizes $3.86 billion in federal funds for election reform, only $650 million has been released to states, a large portion of which was for a buyout of punch-card machines.
The rest of the money has been delayed several times. First, the Bush administration waited nearly a year after passage of HAVA to appoint the members of the Election Assistance Commission, the body charged with distributing the money. It took two more months for members to be confirmed by the Senate, so the EAC didn't hold its first meeting until January 2004. At that point, the group discovered it didn't have enough money to print the state HAVA compliance plans, also a requirement before the money could be distributed. Eventually, a deal was struck with the General Services Administration, and the plans were printed in late March. However, HAVA stipulates that it must be at least another 45 days after the plans are printed before the money is handed out. Doug Lewis, director of the Election Center, predicts that the HAVA money will be released "no sooner than July."
According to Lewis, states are averaging about 18 months behind schedule. And while he says that a timely release of the money will give states "a shot at being compliant for 2006," he thinks there's a possibility that some of the statewide voter databases won't be ready until 2008. But Lewis says that the federal funding issues have left states little choice in the matter. "When you look at how state governments have buckled under the weight of their own debt structures," he says, "it's been very difficult for a governor or a legislature to say go out and do it on a hope and a prayer that we'll get the money."