Voting Machinations

This Town Meeting Day in Burlington, all eyes will be focused on Vermont's first-ever exercise in Instant Runoff Voting (IRV). Election officials and voters will be wondering whether this procedure of ranking candidates in order of preference will go smoothly or result in confusion at the polls and controversy over the results.

At least one Vermont expert on election mechanics has a more fundamental concern, however. Richmond resident Gary Beckwith, a WebPages designer and IBM contractor, worries that returns in Burlington and many other places could be doctored by someone tampering with voting machines. To improve safeguards against electoral thievery, Beckwith is working to create a Vermont chapter of the national group United for Secure Elections, which sees a growing danger of vote fraud. (Check http://election.solarbus.org [1]. for information on Beckwith's efforts.)

In 70 towns and cities, including Burlington, around the state, voters deposit their marked paper ballots in a type of optical scanner that has proven vulnerable to hacking. Beckwith cites a recent test of the scanner in Florida in which hackers were able to falsify the results of a mock election while leaving no trace of their treachery. They did it by manipulating the memory card contained in the same Diebold, Inc. brand of scanner that will count most of the votes cast in Vermont on Town Meeting Day.

"These machines can be hacked to turn election losers into winners and winners into losers," Beckwith says. "It's possible," he offers as an example, "that the result of Bernie Sanders' Senate race could be rigged without any sign of that having happened."

Local and state election officials insist such concern is unfounded. Consultants hired to help implement Burlington's IRV initiative generally agree, suggesting that fraud is especially unlikely in this case.

"If there were some sinister group trying to steal an election, would they risk a long jail term to do it in a municipal election?" Terry Bouricius asks rhetorically. A former Burlington city councilor, Bouricius is now a partner in Election Solutions, the Connecticut-based consulting firm that signed a $5800 contract with the city to help conduct an IRV education campaign. Election Solutions also assists private groups and the public sector in designing secure ballots and transparent voting systems.

Caleb Kleppner, one of Bouricius' partners, points to multiple measures put into place to ensure that Burlington's vote is conducted fairly and counted accurately. The city plans to post the raw data from the March 7 election on its website along with software that will enable users to verify the vote count. Burlington elections director Jo LaMarche "has been making this process as good as it can be," Kleppner says. "Unlike a lot of election officials, she's trying to do it the right way."

Burlington's system of instant runoff voting may prove "a model that should be followed by other cities around the country," adds Kleppner, who has experience with IRV procedures in San Francisco. He and LaMarche express confidence that the multimedia voter-education drive will ensure that IRV balloting goes smoothly on Town Meeting Day.

But both Kleppner and Bouricius acknowledge that no voting system is foolproof. "With an election, one should always be suspicious," Bouricius cautions. Adds Kleppner, "The bottom line is that vendors [of voting machines] should not automatically be trusted."

LaMarche says she has seen no evidence of attempted vote tampering in her 21 years as Burlington's elections director. Diebold optical scanners have been used in Burlington since 1994 and have proven highly reliable, she adds. Several recounts took place during that time, with the actual marks on the paper ballots being compared to the electronic count by the optical scanners. "And we've seen very little, if any, discrepancy," LaMarche reports. The Diebold machines are clearly superior to the type the Queen City used prior to 1994, she notes.

As long as voters mark paper ballots and do not record their choices exclusively by electronic means, Vermont officials say, it will always be possible to corroborate the machines' tallies by doing an old-fashioned hand count. And voters in every one of Vermont's 246 municipalities do leave such a paper trail, says state elections director Kathy DeWolfe.

Vermont does not use the touch-screen electronic voting machines, also available from Diebold and other manufacturers, that have generated much greater controversy than the optical scanners. The accuracy and integrity of a touch-screen vote recorder cannot be checked against paper ballots because none is used in this type of voting. Vermont law requires that paper ballots be cast in all its elections.

"I'm completely confident we're using the best system available," DeWolfe says in regard to the Diebold scanners. Kleppner, too, says he has no qualms about the Diebold optical scanner, which, he notes, has been used without incident in IRV elections in Cambridge, Mass., for the past eight years.

*****

Is Beckwith, self-taught in the science of voting machines, simply being paranoid?

He emphasizes that his apprehension about cooked outcomes has nothing to do with the integrity of election officials in Burlington or the rest of Vermont. Beckwith's concerns are based on the secret design of Diebold's scanners and on the actions of the company's past and present executives.

Consider:

Walden O'Dell, then-CEO of Ohio-based Diebold, wrote an invitation to a Bush re-election fundraiser in 2003 stating that he intended to help "Ohio deliver its electoral votes to the president." Ohio did get counted in the red column the following year, with its result providing the decisive margin of victory for the Republican incumbent.

The uproar over O'Dell's comments, compounded by allegations of voting irregularities in some Ohio precincts in 2004, culminated with his resignation two months ago. Company officials privately acknowledged the impropriety of such partisan remarks by O'Dell, who was also a major donor to Bush's re-election bid.

But Diebold's new CEO, Thomas Swidarski, is also a Republican stalwart. Swidarski was one of about a dozen Diebold executives who helped fund the Bush-Cheney campaign in 2004, with Swidarski himself making the maximum individual contribution of $2000.

Diebold has since barred its top administrators from making political donations. But the Cleveland Plain Dealer reported recently that three Diebold executives not covered by the ban have continued to contribute to GOP candidates in Ohio.

Two groups of investors are meanwhile suing Diebold in federal court on the grounds that the company gave misleading assurances about the security of its voting machines. Those allegedly false claims led to artificial inflation of Diebold's share prices, the lawsuits charge. The disgruntled investors complain that Diebold is "unable to assure the quality and working order of its voting machine products."

Taking note of these developments, The New York Times criticized Diebold's "flawed approach to its business" in a December editorial. "The counting of votes is a public trust," the Times declared. "Diebold, whose machines count many votes, has never acted as if it understood this."

Using paper ballots isn't enough to allay misgivings about elections that are conducted even partly by electronic means, the Times added. "Paper trails are important," the editorial stated, "but they are no substitute for voting machine manufacturers of unquestioned integrity."

Beckwith may be a lone voice in Vermont raising questions about the reliability of Diebold election machines, and that's a little surprising; in other states many groups and individual activists have been striving to raise awareness about alleged flaws in the company's voting security systems. Avi Ruben of Johns Hopkins University, for example, says the Diebold voting system "is far below even the most minimal security standards applicable."

And then there's the infamous "Hursti hack."

Uncertain about the security of Diebold's optical scanners, the elections supervisor of Leon County, Fla., asked two expert hackers in December to test the machines' vulnerability to electronic tampering. One of them, software-security strategist Herbert Thompson, told the Associated Press he was "shocked" at how easily he was able to penetrate the defenses of the Diebold optical scanner and manipulate its vote count. Finnish computer whiz Harri Hursti meanwhile diddled the memory card inserted into each Diebold voting machine. Hursti hacked the card with a scanning device available on the Internet. He, too, was able to alter the outcome of a mock vote and leave no trace of his intrusion.

Leon County elections chief Ion Sancho then ruled that Diebold optical scanners could not be used in his locale.

Both hacks were successfully executed despite the fact that Diebold refuses to disclose information about the software used in its scanners and tabulators. The publicly held company says it treats this information as proprietary in order to protect its products against competitors and hackers. Critics contend, however, that Diebold has a duty to be more transparent about what's inside its machines because the company plays a key role in the democratic decision-making process in hundreds of localities.

"To the extent that any privately owned equipment is susceptible to tampering, there's going to be a certain risk of fraud being perpetrated and perhaps going undetected," suggests Kleppner.

Despite all these concerns, California officials last week conditionally certified Diebold machines for the state's elections this year.

Vermont elections director DeWolfe expresses full faith in the integrity of Diebold's scanners. And she contends that any glitches or deliberate misconduct in the use of the machines would almost certainly be detected by local election monitors in Vermont.

In regard to the Hursti hack, DeWolfe echoes Diebold officials, maintaining that this manipulation of mock vote results was achieved under circumstances that would not exist in an actual election. DeWolfe says that Leon County elections chief Sancho told her the Diebold machine produces a guaranteed accurate tabulation as long as no one is given access to the scanner's memory card. Diebold spokesman David Bear said in remarks quoted in The Washington Post that what Sancho did "is analogous to if I gave you the keys to my house and told you when I was gone." The Florida test was based on hackers having "complete unfettered access" to the equipment, Bear added, saying a responsible elections administrator would never allow that to happen.

In Burlington, LaMarche notes, only she and the seven ward clerks are permitted to handle the memory cards.

DeWolfe suggests there is a greater chance of local election monitors in Vermont colluding to carry out fraud than there is of a town clerk handing over a scanner's memory card to some unauthorized party. And the chance of collusion for the purposes of fraud on the part of Vermont election monitors is minuscule, she adds.

Between nine and 21 observers are present in most Vermont towns when votes are cast and counted, DeWolfe notes. "And these people know their towns quite well. They know what a vote count should look like."

DeWolfe also dismisses objections to Diebold executives' practice of contributing money to political candidates, all of whom appear to be Republicans. "I am certain that every company that has anything to do with selling election supplies includes employees who make political donations. That doesn't make the company fraudulent," DeWolfe declares.

Other Diebold defenders point out that no one has demonstrated an occurrence of fraud in any actual elections in the United States involving electronic voting machines. In some quarters, insinuations of election rigging are rejected outright as either fantasies spun by conspiracy theorists or whining by sore losers.

"The beauty of the American voting system is that it's done locally," says John Silvestro, president of LHS Associates, a Massachusetts-based firm that distributes Diebold election machines throughout New England. "The vote total is so comparatively small in each locality that any tampering couldn't impact the outcome of a statewide or national election."

During the 20 years he has been selling Diebold election products, Silvestro continues, "there have been nearly a thousand recounts involving our machines and there has never been one instance of anybody making an accusation of fraud."

In other cases not involving Diebold machines distributed by LHS Associates, however, there have been numerous allegations of tampering. The results of Ohio's vote in the 2004 presidential election prompted many such accusations, for example.

"Diebold is a really good company," Silvestro instists, noting that the firm sells "millions of ATMs worldwide that handle trillions of dollars."

Indeed, Diebold's business is focused on card-based financial and self-service systems, with electronic voting technology accounting for only a small segment of the company's operations. Diebold began making voting machines following Congress' move to help states and localities avoid a replay of the Florida vote-count debacle in the 2000 presidential election. The Help America Vote Act, approved in 2002, included $3 billion in funding for efforts to enhance voting security.

But Diebold has been experiencing financial difficulties of late, with its earnings dropping 76 percent, to $14.6 million in the fourth quarter of 2005. The disappointing results mainly reflected a companywide restructuring, but negative publicity around Diebold's voting machines may have deepened the earnings slump. Swidarski, the new CEO, expressed confidence in the machines' performance in a recent interview with the Associated Press. But he added that a plan to cut $100 million in costs over the next three years could lead Diebold to get out of the voting machine business.

Whatever the company decides, many election experts believe that electronic voting will become increasingly common in the United States -- with and without the backup of paper ballots. And that will only amplify the risk of voting fraud, Beckwith warns.

The underlying danger highlighted by the Hursti hack, he says, is that a manufacturer or distributor in cahoots with political swindlers could build into its scanners and tabulators undetectable means of altering vote counts. "If someone at LHS wanted to rig an election," Beckwith says hypothetically, "they could do something similar to what Hursti did to the memory card."

Before the first ballots are cast on Election Day, town clerks are instructed to "zero out" the optical scanners to ensure that they haven't pre-recorded any votes. But an expertly hacked memory card could actually be loaded with votes the zero test would not detect, Beckwith says.

Such manipulation is at least theoretically possible in Vermont, Beckwith adds, because the Diebold optical scanners may contain "executable or interpreted code on the memory cards." The presence of this code is what could allow a sophisticated Hursti-style hack to occur, Beckwith explains. The inclusion of the code in the memory cards would also put Diebold optical scanners out of compliance with voting system standards promulgated as guidelines by the Federal Election Commission in 2002, he adds.

The scanners and tabulators used in Vermont are certified to 1990 federal standards, says Silvestro, head of the firm that distributes the Diebold machines. "The new model in production is certified to the 2002 standards," he adds. But Silvestro did not directly respond to the issue of whether executable code is contained in scanners distributed by LHS. Vermont officials have also not indicated whether the code is present.

Silvestro argues that the scenario of pre-cooked memory cards is implausible because an in-house hacker would have no way of knowing how many phantom votes to conceal in a card. Too large a number would raise suspicions about voter turnout, and too small a number would not achieve the intended result of throwing an election to a candidate who had actually lost, Silvestro says.

That doesn't mean such an attempt would never occur or would always be unsuccessful, says Beckwith, who is frustrated that Vermont's election director has such unskeptical trust in Diebold's voting machines. The most practicable way to help ensure against electronic vote fraud, he suggests, is by requiring random audits of returns in a statistically significant proportion of precincts or wards. Under this procedure, scanners' tabulations would have to be checked against the paper ballots cast at randomly selected polling places.

Current Vermont law allows random audits to be carried out at the discretion of the secretary of state.

Beckwith isn't calling for scrapping scanners altogether and going to exclusively paper ballots counted entirely by hand. As Silvestro notes, "Machines have been shown to be more accurate than human beings in counting votes." For Beckwith, rather, the main aim is to come as close as possible to guaranteeing the integrity of the voting process in Vermont and every other state. "I want to live in a democracy," he says, "where everyone can have confidence that elections are conducted openly and honestly."