For election officials in Fort Lauderdale, Florida, November 2 had passed with merciful ease. The balloting was deemed an administrative success -- until two days later, that is, when election workers noticed a mathematical oddity: As they canvassed more votes, the tallies in certain races had decreased. After some hand-wringing, election officials discovered that the Election Systems & Software (ES&S) machines used by Broward County to count absentee ballots were inexplicably programmed to subtract votes once the 16-bit devices hit a ceiling of 32,767 votes.
According to Rebecca Mercuri, a computer-security expert at Harvard's Radcliffe Institute for Advanced Study, this mistake was a common error made by “dumb freshman computer-science majors” and other neophyte programmers. In this case, ES&S had known about the problem for at least two years, but neither the company nor state or local election officials had aggressively sought to correct it. In a statement to Florida newspapers shortly after the election, the company issued a meek mea culpa: “While the county bears the ultimate responsibility for programming the ballot and structuring the precincts, we … regret any confusion the discrepancy in early vote totals has caused.”
But the problem with the computer error wasn't just “confusion”; it was the potential disenfranchisement of some of Florida's voters. And the problems with the new machines didn't occur only in Florida; they happened throughout the country.
On November 2, 30 percent of all voters used Direct Recording Electronic (DRE) machines, which use ATM–style touchscreen technology and generally leave no paper trail for recount. Without printed receipts as a backup, a malfunctioning DRE machine can permanently lose votes, a possibility that voting-rights activists repeatedly warned of before the balloting. In this election, only one state, Nevada, mandated that the machines produce paper receipts so that election officials could conduct a recount if necessary. And on election day, the Verified Voting Foundation, a nonprofit organization that advocates for more reliable electronic-voting machines, registered more than 1,800 machine malfunctions, nearly half involving DREs.
Carteret County, North Carolina, used DRE machines manufactured by the UniLect Corporation, with no paper record. The county lost more than 4,500 votes due to computer error -- a significant number given that some races were tight, with the statewide contest for agriculture commissioner decided by fewer than 2,400 votes. With no hope of recovering the lost votes, both sides are engaged in a bitter legal dispute, and the state is without an agriculture commissioner–elect.
Those monitoring elections say that incidents like the one in Carteret County, and the myriad other election-day machine malfunctions recorded by the Verified Voting Foundation, do not suggest a pattern of widespread fraud. Contrary to Internet rumors, there is no hard evidence of a nefarious election-day conspiracy plot involving the much-maligned Diebold Inc., or any other machine vendor, to hand deliver the election to George W. Bush. But with no auditable paper trail, experts say, there's really no way to know. As Chellie Pingree, president of the advocacy organization Common Cause, pointed out at a recent press conference, “the absence of a meltdown should not be a measure of success.”
Indeed, it was a national meltdown that ushered in the age of e-voting. In the wake of Florida's 2000 fiasco with hanging, dimpled, and pregnant chads, election officials sought to improve their election equipment. The Help America Vote Act (HAVA) authorized federal funding for states wishing to purchase new machines. Not surprisingly, the promise of money opened the floodgates for e-voting vendors wanting to get in on the action.
HAVA was underfunded from the outset. If money flowed at all to the states to update their machines, it was just a trickle, so many opted to purchase new machines with their own funds in hopes that they would eventually be reimbursed by the federal government via HAVA. They spent millions of dollars on e-voting technology, their decision of which system to purchase often based primarily on a single company's intense lobbying effort. In New York state, for example, the election-machine manufacturer Sequoia Voting Systems hired a former aide to House Leader Joe Bruno to lobby its cause. Lo and behold, the only municipality in New York to buy touchscreen machines was Bruno's home district of Clifton Park, which opted for Sequoia's model. Similarly, in January 2003, just one month before joining Sequoia as a paid consultant, California Secretary of State Bill Jones wrote letters to each member of the Santa Clara Board of Supervisors to reassure them that Sequoia's DRE machines were safe. Sequoia ultimately won Santa Clara's $19 million contract.
While a certain amount of government-industrial overlap is to be expected, what is startling about the voting-machine industry is the degree to which this symbiosis has been institutionalized. This is due, in large part, to a curious nonprofit entity called the Election Center and its versatile executive director, Doug Lewis. The Election Center's members include approximately 1,000 dues-paying state and local election-administration officials, as well some voting-machine vendors. The center provides a host of services for its members, informing them of new developments in election law, sponsoring professional development conferences, and offering training workshops for new election officials. In advance of the last election, the center also performed a quasi-oversight role over the machine-testing process. Specifically, the Election Center, in conjunction with the National Association of State Election Directors, selected which private labs would test new voting-machine technologies.
But in the eyes of many voting-rights activists, the Election Center (and Lewis in particular) acts as a tireless advocate for the industry's interests. In March, The Philadelphia Inquirer reported that the center has received tens of thousands of dollars from the major voting-machine vendors in the United States. Lewis also had a hand in forming the e-voting industry's trade association. In August of 2003, Lewis and Harris Miller, president of the Information Technology Association of America (ITAA), the country's largest IT trade association, hosted a conference call with the presidents of the major e-voting-machine vendors. Academics, such as the indefatigable paper-trail advocate David Dill, a computer scientist at Stanford University, had been publicly questioning the security and viability of DRE systems, and the press was beginning to catch on. In the conference call, Lewis, Miller, and the executives banded together to form a coherent public-relations counteroffensive under the auspices of a new trade association, later called the Electronic Technology Council, to be created as a subsidiary of the ITAA; membership was to be around $100,000 per company. On the council's Web site, an official statement of neutrality on the issue of voter-verified paper ballots is quickly followed by a long list of reasons why such a requirement would, in fact, be onerous.
Considering that Lewis' organization's members are election officials who serve a public trust, Lewis' critics cite his participation in creating the Electronic Technology Council as an apparent conflict of interest. If true, this has only become compounded since June 2004, when Lewis was elected chairman of the Board of Advisors for the Election Assistance Commission, a federal body created by HAVA to oversee the bill's implementation.
Lewis, however, isn't alone in blurring the line between government and the voting-machine industry. Industry representatives have inserted themselves directly into the emerging process of codifying the performance standards by which new machines must operate. Recognizing that the performance and testing standards for e-voting machines still lagged far behind innovations in technology, the authors of HAVA mandated that the Election Assistance commission pass a new set of guidelines by 2006. Though these standards are only voluntary, in practice they will provide the basic guidelines to which new machines need to adhere to win certification in most states. The commission passed on the responsibility of creating these guidelines to the Institute of Electrical and Electronics Engineers (IEEE), a professional society of high-tech engineers.
Alas, says Mercuri, the computer-security expert who is also one of about 30 members of the IEEE's voting-equipment-standards working group, this process has been hijacked by industry representatives who serve on the working group and are manipulating the standards to serve their own profit-expanding ends. A longtime employee of the prominent e-voting vendor ES&S, Herb Deutsch, now chairs the working group. According to Mercuri, Deutsch and his allies are deliberately making standards overly complex in order to dissuade any new startup company from developing new technologies. And, she says, they are not making systems very secure. Nonetheless, it now appears that the standards the group is proposing are likely to be approved by the Election Assistance Commission in 2006, without the security provisions demanded by independent academics, computer-programming experts, and activists.
Seeing as the 2006 voting-machine standards are not likely to require that touchscreen machines print voter-verified receipts, and that many secretaries of state have already shown themselves to be cozy with the industry, the legislative process remains one of the last remaining recourses for mandating that fallible machinery be backed up by a paper trail. Some states, such as New York, are working on laws that would require this. However, such an effort is unlikely in all the states, leaving voters unprotected against industry encroachment into the voting booth. Members of Congress ought to stand behind their Democratic colleague Representative Rush Holt in calling for national legislation that would mandate a paper receipt. Without such legislation, there is no telling -- literally -- how many votes may be lost in future elections.
Mark Goldberg is a Prospect writing fellow.