The Right to Vote

by Alexander Keyssar

  Those who worried that HAVA was an “anemic piece of legislation” (as the New York Times later characterized it) were not reassured by the early history of the Election Assistance Commission. It took more than a year for the four members of the EAC (two Democrats and two Republicans) to be appointed and confirmed by the Senate. In addition, the agency was initially so underfunded and understaffed that it had difficulty meeting deadlines set by HAVA or offering timely guidance to the states. (Indeed, lacking office space, the commission occasionally had to meet at a Starbucks in Washington.) In April 2005, after repeated public and private complaints, the first chairman of the EAC—Republican Deforest B. Soaries, Jr., a minister and former New Jersey secretary of state—resigned from the commission, lamenting the lack of support for the agency and its mission on the part of both Congress and the administration.31

  Things did improve after the rocky start. By 2007, the EAC had disbursed roughly $3 billion to the states, most of which was being used to purchase new voting machines and develop statewide registration databases. The agency also made considerable progress in establishing “voluntary voting system guidelines” for new technology as well as a process for certifying election systems and accrediting laboratories that tested them. In addition, the EAC sponsored a series of research reports to assist the states, on topics such as provisional ballots, ballot design, and the desirability of “postage-free” absentee ballots. Nonetheless, the implementation of HAVA was uneven, and the EAC had difficulty fulfilling the diverse mandates that it had been given. By February 2008, only eight states had filed certifications with the EAC that they had met all of the HAVA requirements; six months later, the commission announced that backlogs at testing laboratories would prevent the EAC from certifying repairs or from providing software patches to flawed machines in time for the November election. Importantly, the EAC lacked the regulatory authority to compel the states to comply with the legislation, although the Department of Justice— which had that authority—did file suit against some of the most flagrant laggards, including New York and New Jersey.32

  Despite the passage of the Help America Vote Act, thus, election administration remained decentralized, at heart a state rather than federal responsibility. On most matters, the roles of the federal government and the EAC were purely advisory, and even where HAVA had established requirements, states could devise their own strategies for meeting them. All states, for example, were obliged to make it possible for voters to cast provisional ballots, but the rules determining whether those ballots would actually be counted varied across the country. Congress had acted, but the real action—with respect both to regulation and innovation—remained in the states.33

  Technological Progress

  Florida—hardly a shock—was among the first states to purchase new voting machines after the election debacle of 2000. The state’s sweeping election reform law, passed in May 2001, required all counties to cease using punch cards, mechanical lever machines, or hand-counted paper ballots by 2002; it also gave $24 million to the counties towards the purchase of optical scanning systems or new, computerized touch-screen systems, as soon as they were certified by the state.34 Certification of the new DRE (direct recording electronic) technology was completed within a matter of months, and fifteen counties, including some of the largest in the state, opted to purchase these machines; by doing so, county officials reasoned, they would acquire the most advanced technology available and put the problems of 2000 safely behind them.

  But the new technology did not put an end to Florida’s electoral woes. In the 2002 gubernatorial primary, problems with electronic machines delayed the count and cast doubt (yet again) on the outcome of a close election; one precinct reported no votes cast, although it had 1,000 eligible voters. Two years later, there were numerous reports of machine malfunctions in a March primary, and during the 2004 general election officials noticed that vote totals began to go down, rather than up, as absentee ballots were counted. In 2005, in a one-ballot election in Miami-Dade County, five hundred votes were mistakenly tossed out by the DRE machines. Then, most dramatically, in 2006 in Sarasota, electronic voting machines recorded an 18,000-vote “undercount” in the Congressional election to replace Katherine Harris, the former secretary of state who played a large role in the 2000 election. Although the margin between the two candidates was fewer than five hundred votes, no recount was possible in the absence of physical ballots—which led both to litigation and to a shadow cast over the election’s outcome. In 2007, Charlie Crist, the recently elected Republican governor, signed into law a new bill that provided $28 million to replace all of the DRE machines (except those needed for voters with disabilities) with optical scanners in time for the 2008 election.35

  Once again, Florida’s experience was more paradigmatic than unique. In the wake of the 2000 election and particularly after the passage of HAVA, state and local officials across the nation sought to replace their existing, often antiquated voting machines with new, improved technology. Many were drawn to the DRE devices, which were being aggressively promoted by four manufacturers: Diebold Election Systems, Sequoia Voting Systems, Electronic Systems and Software, and Hart InterCivic. The lure of these machines was obvious, almost irresistible, in an era when every office desk had a computer and personal banking was conducted largely through ATM terminals. Computerized voting promised to be accurate, crisp, fast, and reliable, and there would be none of the messiness of handling paper ballots with ambiguous markings or dangling chads of paper. In addition, touch-screen machines could easily be made accessible to voters with disabilities, as mandated by HAVA, and they could readily be programmed to meet the non-English language requirements of the Voting Rights Act. By 2006, according to an EAC survey, DRE machines had been placed into use in more than half of the counties in the United States; some states, such as Georgia, had converted entirely to the new electronic technology.36

  From the beginning, however, there were skeptics who worried about both the reliability and the security of computer-based voting systems. As untold thousands of PC users learned every year, computers did crash, and stand-alone DRE devices provided no alternative record (or physical ballots) that could be counted in the event of a malfunction. The computers, moreover, were “black boxes” that did not permit election officials or inspectors to determine, with certainty, that the preferences expressed by a voter were in fact those recorded in the electronic tallies. This concern was heightened by the refusal of DRE manufacturers to open their software code to independent experts; it was heightened even further when the code to a Diebold machine was inadvertently made public, examined by experts, and found to contain disturbing flaws. Nor did it help the cause of DRE devices when the chief executive of Diebold, a Republican fundraiser, publicly expressed his determination to “deliver” Ohio’s 2004 electoral votes to President Bush.37

  A series of formal investigations of electronic voting technology confirmed many apprehensions about the security of the systems. One study, commissioned by Maryland in 2004, found that the 16,000 Diebold machines it had purchased could be broken into and tampered with by computer experts with remarkable speed and ease; among other problems, all of the machines could be opened with the same key, readily available in hardware stores. In 2006, a team of Princeton University computer scientists released a paper (and a video) revealing that they had developed “vote-stealing” software that could be installed on a Diebold machine in less than a minute and that could change votes without being detected; the machines were also vulnerable to viruses that could spread from one machine to another during elections. A later investigation carried out in Ohio found that handheld devices could plug false vote counts into some electronic machines and that “malignant software” could be introduced into servers at boards of election. “The biggest problem is security against tampering,” declared Lowell Finley, a California election lawyer and co-director of Voter Action, a nonprofit “election integrity” organization founded in 2005.38
>
  While the concerns of computer security experts were becoming better documented and more widely known, instances of machine malfunction proliferated. Although there may well have been an improvement in the overall performance of voting technology after 2000, there were also numerous reports of failure, some of them troubling. 39 In the fall of 2003, in Indiana, an electronic voting system indicated that 144,000 votes had been cast in a jurisdiction that had only 19,000 voters. In 2004, in North Carolina, a close race for state agricultural commissioner was mired in confusion for a year when machines failed to record 4,500 votes because the machines’ ballot capacity had been exceeded. In Albuquerque, a Sequoia machine lost 13,000 votes in one election, and dozens of voters filed affidavits claiming that they had touched one candidate’s name on a computer screen only to see the computer select the candidate’s opponent.40 In Ohio, during the presidential election of 2004, a computer error awarded President Bush an extra 3,900 votes in a precinct that had only 800 registered voters. Across the nation, moreover, voters and election officials were encountering problems emanating from a poor fit between the new technology and the humans who staffed polling places—many of whom were elderly and nearly all of whom lacked expertise in information technology.41

  This problematic track record began to sway public opinion, leading some state officials to pause before updating their technology and spending additional (or any) HAVA funds. Opinion polls revealed a growing public mistrust of electronic machines, and by July 2006, activist and citizens’ groups had filed lawsuits in nine states (including California, Colorado, New Mexico, Ohio, and Pennsylvania) to prevent their continued use.42 The manufacturers of DRE machines responded to the criticism with claims that their systems were more accurate than other technologies and that the fears of tampering were based on purely hypothetical (and extremely unlikely) scenarios; to press their case, they mounted extensive lobbying operations and hired several ex-secretaries of state to assist their efforts.43 DRE technology was also supported by some groups representing citizens with disabilities, since touch-screen machines, with special add-on features, were more accessible to many disabled voters than were most alternative technologies, particularly those based on paper ballots.44

  By the end of 2007, however, most state officials, independent experts, and election integrity organizations had concluded that paperless DRE technology posed an unacceptably high risk of producing electoral outcomes that were inaccurate, disputed, or perceived as illegitimate—whether through machine error, human error, or malfeasance. The most readily available method of lowering that risk was to require all DRE devices to produce a “voter verified paper trail (VVPAT),” a printout that each voter could inspect to make sure that her vote had been recorded correctly before it was added to the electronic totals; the paper printout would then be preserved for potential use in recounts. Between 2004 and 2007, a majority of states passed laws requiring VVPAT as part of their voting systems, and federal legislation mandating or encouraging such paper trails was strongly promoted in Congress, particularly by New Jersey Representative Rush Holt. Yet confidence in VVPAT as a solution to the problems with DRE machines was never universal, and by 2007 it had begun to ebb among election-integrity experts. Voter-verified paper trails did not satisfy core security concerns since there could be no certainty that the printed preferences were identical to those recorded electronically. Moreover, both research and experience demonstrated that the printing technology itself could be unreliable, as well as clumsy and expensive.45

  As a result, some experts and activist organizations argued that DRE technology was too inherently insecure and potentially problem-ridden to be deployed in elections. They maintained that states should cease buying DRE machines and replace those already in use with optical scanning systems—in which voters actually marked a paper ballot that could be quickly read by a machine and would be available in the event of a malfunction or a need for a recount. The preference for optical scanning systems, which were already in use in many states, was also voiced by the National Institute of Standards and Technology, which advised the EAC. Optical scan technology had its critics, who pointed out that imperfectly marked ballots could be misread by the machines, but the flaws of optical scanning systems seemed preferable to the hazards of DRE machines. “Paper ballots don’t break down or fail to boot up,” declared Lowell Finley, who in 2007 was appointed deputy secretary of state for voting technology and policy in California. “They also cost less, are more accessible, and provide the added benefit of a permanent record.” Some who shared this analysis continued, nonetheless, to advocate VVPAT because they believed both that it was politically unrealistic to insist that states and counties junk their DRE machines and that it would be better to have machines with paper trails than completely paperless systems.46

  The view that DRE technology was fundamentally flawed also began to gain traction among state officials. Even before Florida acted, New Mexico, in 2006, responded to a series of technological mishaps by deciding to abandon its DRE machines and return to paper ballots counted by optical scanners. In August of 2007, Debra Bowen, the secretary of state of California, decertified DRE machines in thirty-nine counties (including Los Angeles), launching what would become a lengthy process of negotiation with county officials and manufacturers regarding which systems (and with what additional security features) could remain in use in the state. Colorado followed a similar path, heading—with some zigzags—toward paper-based technology. In January 2008 in Ohio, after an official report concluded that all five electronic voting systems deployed in the state had serious security flaws, Secretary of State Jennifer Brunner ordered that paper ballots be made available to any voters who requested them in the upcoming primary election; she also recommended the replacement of all the DRE machines then in use. A few months later, officials in Iowa announced that their state would return to paper and optical scanners in time for the fall 2008 elections. Election administrators who had never gone down the electronic path in the first place congratulated themselves on their foresight.47

  Nonetheless, DRE technology (with and without paper trails) remained in place in jurisdictions across the country. Some state officials retained full confidence in the machines, and many others were understandably reluctant to jettison expensive, recently purchased equipment that they had already spent months or years training local administrators to utilize.48 During the 2008 elections, roughly a third of all voters (less than in 2006) cast their ballots on DRE machines, while more than half utilized paper ballots counted by optical scanners. An enduring feature of the electoral system was that numerous different technologies remained in use: although punch cards and lever machines were far less common than they had been in 2000, Americans in 2008 continued to vote in diverse ways, varying with the state and county in which they lived.49

  Whatever the trepidations of public officials and voting experts, there were no nationally reverberant failures of the nation’s voting technology in 2008. During the weeks of “early voting” leading up to the November 4 election, there were numerous reports of DRE machines “flipping” votes (from one candidate to another), but election administrators were able to solve, or contain, that problem. On Election Day itself, machines malfunctioned in scattered locales around the country although not in sufficient numbers to generate extremely long lines or uncertainties about the tally (except in a handful of local races). The core flaws and vulnerabilities of DRE technology had not vanished, but thanks to the preparation and increased experience of many election officials, as well as the red flags run up in advance by skeptical computer experts and activist groups, no crisis occurred. Then too, the national election was not close: Barack Obama’s margin of victory over John McCain in many states was substantial, and, unlike the elections of 2000 and 2004, the presidency never hinged on the outcome in any single state.50

  Despite the relatively uneventful election of 2008, the bloom had irretrievably come off the rose of electronic voting. “It is
hard to believe now what a darling touch-screen voting was seven years ago,” concluded an article in Time in November 2007. Touch-screen or DRE devices had offered a tempting technological fix for the public loss of confidence precipitated by the chaos of the 2000 presidential election. The embrace of such a fix sprang from a deep current in American political culture, a longstanding impulse to find technological solutions to complex, multidimensional issues. Yet machines alone could not solve problems that were political and institutional as well as technological. The headlong, if decentralized, rush to purchase new voting machines had spawned a new round of irritating problems: with the support and encouragement of Congress, many states had become the proud owners of voting machines and software that had not been adequately tested and were clearly, perhaps fatally, flawed. Election administrators and politicians had learned a great deal about voting technology during the early years of the twenty-first century, and they had participated in a remarkable undulation of technological preferences, yet even after 2008 the threat of another “Florida” continued to lurk on the horizon.51

  Crime and Punishment

  In the summer of 1997, a small group of inmates at Norfolk State Prison in Massachusetts formed a political action committee to influence public debate about criminal justice and social welfare issues. As had been true of inmates in the commonwealth since the American Revolution, the men residing in Norfolk were legal voters. An underlying goal of the political action group, according to one of its founders, Joe Labriola (a decorated Vietnam veteran who was serving a life sentence for murder), was “to make prisoners understand that we can make changes by using the vote.”52

  Within days of its founding, the prisoners’ PAC drew fire from politicians as well as victims’ rights groups. Acting Governor (and Republican gubernatorial candidate) Paul Cellucci led the charge, declaring that “the idea of prisoners organizing politically is to me repugnant.” Cellucci and his allies in the state legislature then filed an amendment to the state constitution to prohibit prisoners from voting. As required by law, the proposed amendment was submitted to two successive sessions of the state legislature (in 1998 and 2000), both of which endorsed it by large majorities. The final step in the amendment process was taken on November 7, 2000, when the state’s electorate voted on the measure. Despite opposition from leading newspapers and liberal organizations, it passed easily, winning almost two-thirds of the votes cast. “The hammer sure came down in a way we didn’t expect,” concluded one of the original members of the PAC. For the first time in the state’s history, Massachusetts had amended its constitution to narrow the breadth of the franchise. Once the amendment took effect, Maine and Vermont became the only states in the nation that permitted inmates to cast ballots.53