But Who’s Counting?
Electronic voting machines are supposed to restore integrity to the electoral process—but critics say they merely open a new backdoor to error and fraud
By Travis Durfee

On Nov. 7, 2002, CNN reported that electronic voting systems in South Florida’s Broward County crashed, causing 103,222 ballots to be lost on election night. County election officials said the missing votes did not affect the outcome of any races, but one couldn’t help but remember that George W. Bush was handed the 2002 presidency based on a slimmer margin.

Also last November, the electronic voting systems used in Tarrant County, Texas (which encompasses Fort Worth, the state’s sixth largest city), would not allow voters to register ballots cast down straight Republican or Democratic party lines. The glitch, which election officials blamed on human error, wasn’t caught until Election Day, even though technicians had tested the machines in October 2002. It delayed the counting of 300,000 ballots for the governor’s race and a U.S. Senate seat.

In Georgia, a state that employs 22,000 electronic voting machines, incumbent Democratic Sen. Max Cleland lost a close election to his Republican challenger, Rep. Saxby Chambliss, despite polling information showing Cleland in the lead heading into Election Day. Four days prior to Election Day 2002, a USA Today poll showed Cleland leading by 49 to 44 percent over Chambliss. The day after the election, Cox News Service reported that Hotline, a political news service, “recalled a series of polls Wednesday showing that Chambliss had been ahead in none of them.”

Reports from various New Jersey newspapers stated that the electronic voting machines in 30 of 46 voting districts in Cherry Hill, N.J., malfunctioned in November 2002, locking the lever to vote for Democratic mayoral candidate Bernie Platt for up to four hours.

Electronic voting machines in Texas produced another oddity: Sen. Jeff Wentworth, Rep. Carter Casteel and county Judge Danny Scheel all won their respective elections in Comal County by the total of 18,181 votes. “Isn’t that the weirdest thing?” Comal County Clerk Joy Streater told the San Antonio Express-News shortly after the election.

Are electronic voting machines getting a bad rap? Are those opposed to computerized voting systems nothing but a lousy bunch of Luddites with nothing better to do than claim the sky is falling? Or are the above cases merely the tip of the iceberg, portents of a coming wave of vote-counting error and fraud as we enter the era of computerized voting?

Proponents of the machines don’t think so. They insist that these incidents are isolated and can be attributed to human error. They are quick to note the number of security steps and certifications each machine is required to pass before it can be used in an election. They claim it would be extremely difficult for an individual to bypass these security measures and actually rig an election. They call such a scenario improbable—but not impossible.

For the last decade, critics of electronic voting machines have been harping on issues of security and machine malfunction, and proponents of computerized voting can’t seem to write them off. Considering that the detractors are some of the most respected names in computer science, it’s no wonder.

The 35 days from Nov. 7 to Dec. 12, 2000, will be remembered as some the most regrettable in the history of our nation’s electoral process. It was over those 35 days that the 43rd president of the Unites States of America was ultimately decided, not by the voters, but by the Supreme Court. Despite cries of confusing ballots, misleading directions at polling places and wrongly disenfranchised voters, the nation’s highest court decided that the questionable ballots of nearly 200,000 Floridians would not be counted in an election that was decided by far fewer votes.

Fearing that the integrity of our nation’s electoral process was severely tarnished, Congress produced a panacea, the Help America Vote Act—a far-reaching, multibillion-dollar proposal to modernize and standardize election systems throughout the 50 states. Congress hoped the legislation would ensure that the ballot and polling-place confusion that tainted the 43rd presidency would never happen again.

HAVA requires states to replace their old mechanical and punch-card voting machines—devices long criticized for being improperly equipped for persons living with disabilities—with electronic voting machines. Just as word-processing software’s ease of execution and versatility eclipsed the typewriter’s mere functionality, so will electronic voting machine technology do for the lever and punch-card machines of old, goes HAVA’s logic.

But critics of the switch in technology say that these direct-recording electronic voting machines, or DREs, are more prone to malfunction and open to fraud than their predecessors. Critics paint worst-case scenarios where the machines crash and lose thousands of votes, or a rogue computer programmer installs an undetected code into the machines that throws an election.

David Dill, a professor of computer science at Stanford University, maintains Verifiedvoting.com, a Web site that details the faults of electronic voting machines.

“Computerized voting equipment is inherently subject to programming error, equipment malfunction, and malicious tampering,” begins an online petition on Dill’s site.

The petition has drawn 1,794 signatures in all, 992 of which belonging to computer scientists from academia (Massachusetts Institute of Technology, Carnegie Mellon University and Princeton University, to name a few), and researchers from the private sector (Microsoft, Adobe Systems and Sun Microsystems). Representatives from research groups, law firms, and citizen-, civil- and disabled-rights groups round out the list.

Rebecca Mercuri, one of the petition’s signatories and an assistant professor of computer science at Bryn Mawr College has been one of the most consistent and prolific critics of direct-recording electronic voting machines over the past 10 years.

Mercuri advocates DREs that produce a verifiable paper audit trail. After casting their ballot, voters would receive an ATM-like receipt verifying their choices. The piece of paper would then be deposited in a lockbox, and the votes could be recounted in the event that a DRE loses its memory or the results from an electronic machine are challenged.

“If we had a piece of paper that said this is who I voted for, I see it and put it in the ballot box and that is what is used to count the elections, that would be fine,” Mercuri says. “But no, we’re just supposed to trust the machines.”

We shouldn’t trust the machines? Despite the millions of dollars companies have spent researching and developing software for these machines, they are still faulty? Mercuri says that the best electronic voting machines are only as good as the standard against which they are tested, and the current standard used by the Federal Election Commission, which certifies DREs, isn’t good enough.

According to Mercuri, a DRE’s software and codes are currently tested against an obsolete, decade-old security criterion, and the machines are tested only once with a hypothetical code. Since new candidates, parties and races ensure that a machine’s code is often rewritten for each election, Mercuri sees plenty of opportunities for election fraud.

Under those standards, Mercuri says it wouldn’t be much of a challenge for an ill-intentioned, modern-day programmer to install a patch, or a rogue code, onto a DRE, altering the outcome of an election one way or another.

“When a person writes computer code, it is written in a symbolic notation that might even look like English—like, ‘If this happens; then do this,’ ‘If this button is pressed; then add this to the total,’” Mercuri says. “But then that English-like code has to be transformed by another program into something a computer can read, and in the translation process a permutation can take place that adds something to the code that could enable what they call a Trojan horse. This is not fiction; this is fact and it is very easy to do.

“In the process of doing this transformation,” Mercuri continues, “you can add something to the code that says, ‘If you press this button three times; then increment this guy by 10 percent and decrement this other guy by 10 percent.’ That is a very simplistic example, but that’s the gist.”

To find such a directive, Mercuri says an auditor would have to review not only the computer’s code, but the code of the translating software, to ensure that everything is working properly. Considering that such directives are quite small, 200 characters or fewer, it would be virtually impossible to find.

“You can look at the program code until you’re blue in the face, but there is almost no way of knowing whether these things have been added in the transfer process,” Mercuri says.

Proponents of electronic voting machines say Mercuri’s claims sound too much like a conspiracy theory; that it would be very difficult for even one of the machines to be tampered with. But they stop short of saying it is impossible.

“The people that are not for electronic voting machines are the theorists from the data-processing world, so they’re pretty hard to argue with,” says Larry Tonelli, a salesman with Sequoia Voting Systems, one of two manufacturers certified to provide New York with electronic voting machine. “To tamper with the machines is possible, but it would require total collusion from so many people involved in the election process.”

Jim Allen, who engineers voting machines for Sequoia, says that critics like Mercuri have made a profession out of discrediting his work, and he is none too happy.

“I’ll tell you that they are the biggest bunch of bullshitters out there,” Allen says. “All the years we’ve tried to keep the voting machines sacrosanct, yet these people want to discredit everything we’ve done. It is destroying the sanctity of voting.”

But Mercuri says she is simply fulfilling her responsibility as a computer scientist and relaying the facts.

“One of the papers in California said we are creating a UFO effect, she says, “But these concerns are real. There are certain scientific facts out there that you can’t avoid, and just because someone is stating true scientific facts, they shouldn’t be criticized for that. Unfortunately that has been the case.”

Despite concerns that smoke and mirrors are being used behind the scenes to count votes, the machines themselves don’t really look that evil. They’re pretty slick, actually.

Metroland recently took a tour of Harvard Custom Manufacturing (a Sequoia manufacturing plant in Owego), spoke with Sequoia engineers, and tested a few of the company’s voting machines.

Sequoia’s machine, the AVC Advantage, is one of two full-face, electronic voting machines certified for use in New York, the other being Electronic Systems & Software’s iVotronicLS. Representatives from ES&S did not return calls for comment for this story.

The Advantage has a toylike quality about it; the machine’s burgundy-trimmed, navy-blue plastic casing resembles the kind of durable material used to house children’s computer equipment in museums and toy stores in the mall.

The machine stands at almost 6-and-a-half feet tall, and the ballot faces the voter at a 45-degree angle. Unlike the lever machines, the ballot face on the Advantage can be lowered and adjusted to 90 degrees, ensuring wheelchair-bound voters privacy in the booth. The Advantage ensures voter privacy—no, not by force field—but by a wall formed by the two plastic flaps used to cover the ballot face in storage, and a plain cloth curtain.

The ballot itself would look very similar to the one already in place in New York, and its size could be adjusted for easier reading depending on the number of candidates in any given election. The Advantage can be programmed to accommodate up to 504 ballot positions, so the piece of paper the ballot is printed on can be customized for each election—the fewer the races, the larger the font for each candidate and race.

Voters would actually interact with the ballot by pressing a touch pad, similar to those on cash registers in fast food joints, over a candidate’s name. Voters would cast a vote by pressing the name of a candidate, and then a little green “X” would light up to show a voter his or her selection. Unlike the lever machines, votes could be changed as often as the voter would like, and write-ins can be typed into a keypad at the bottom of the ballot. Voters would log their selections by bopping the flashing red button to register the votes. The Advantage emits a cell-phone-like ring, and voilà, your vote has been cast, 21st-century style.

As HAVA requires of all new voting machines, the Advantage also has a number of features to ensure access for physically disabled and voters with English- language deficiencies. The ballots in the machines are already written in English and Spanish, and the machines can be programmed to read ballots in Cantonese, Pakistani or Somali. Sequoia says they are working on “sip and puff” technology, which would allow quadriplegic voters to vote by sipping and puffing a burst of air into a straw fitted to the voting machine.

But for all of the Advantage’s features to ensure its usability for all persons living with disabilities, there was at least one obvious bug upon inspection last week—the machine’s audio component for visually impaired voters. A prerecorded voice instructs the visually impaired voters to scroll through the choices using the “green and yellow triangular buttons” and make their selections by pressing the “large, circular, red button.”

 

Due to an antiquated statute in the state’s election law, only two vendors are currently certified to provide New York with its updated voting technology, Electronic Systems & Software, based in Omaha, Neb., and Sequoia Voting Systems, with headquarters in Oakland, Calif. Both companies have received their share of bad press.

Machines built by ES&S are at the heart of one of the better-known cases of an election falling under question of fraud. It was on the Omaha-built machines in 1996 that Chuck Hagel was overwhelmingly elected to the U.S. House of Representatives, becoming the first Republican candidate from Nebraska to do so in 24 years—even though press reports state that Hagel was trailing in Election Day polls.

It gets more interesting: Hagel actually had a substantial financial interest in the company on whose machines he was elected, a matter he failed to disclose to the Federal Elections Commission.

Bev Harris, an independent investigative journalist who maintains the Web site Blackboxvoting.com, has reported on Hagel’s apparent conflict of interest as one of a number of questionable cases involving electronic voting machines. She has written a book on the matter under the same name as her Web site, which is scheduled for publication later this year.

Sequoia has its political ties as well. The company has received a fair amount of criticism in New York for hiring a lobbyist, which on the surface seems rather odd since lobbying is a legal activity. But a closer look at how the state has been preparing to implement HAVA’s reforms lends some light to the matter.

Although the state has been preparing for HAVA for the past two years, all of its major decisions have been made in the last few months by the New York State Commission on HAVA Implementation—a task force widely criticized for being stacked with political appointments and lacking diverse citizen representation. In fact, Lee Daghlian, spokesman for the New York state Board of Elections, so much as admitted that committees designed to reform elections law have been politically stacked in the past, and the HAVA task force was no different.

“When we had these discussions about the plans to implement the Motor Voter law in the early ’90s, the conversation was similar,” Daghlian told Metroland [Newsfront, June 26].

“There was a Democratic governor at the time, so most of the department heads that were on the task force were Democrats. So it makes sense this time that [the majority of task force members] were Republican because we have a Republican governor.”

Considering the task force’s admittedly Republican slant and New York’s nonexistent regulations on lobbying for state contracts, Sequoia’s selection of Jeff Buley—who serves as a legal consultant for the state’s Republican Committee and was counsel for Gov. George E. Pataki’s reelection bid last year—as its chief lobbyist seems rather dubious.

Neal Rosenstein, government reform specialist with the New York Public Interest Research Group, said he wouldn’t be surprised if Sequoia landed a large share of the $140 million New York state has to spend on voting machines, given the work of the task force so far.

“That’s business as usual in Albany,” says NYPIRG’s Rosenstein. “The decision should be made on the basis of who has the best machine, not who has the most politically connected contractor.”

Responding to criticisms that his company has hired a Republican-connected lobbyist as it seeks its share of a lucrative state contract, Sequoia’s Tonelli responds, “We hired Jeff O’Dwyer as a Democratic lobbyist, too. But nobody seems to write about that.”

Over the past few months, election officials and citizens groups across the country have been forming committees and drafting plans to ensure that their state complies with HAVA by Sept. 1. On that date, the beginning of the next federal fiscal year, federal election officials will evaluate individual states’ plans before doling out federal funding.

But HAVA doesn’t offer states much guidance on what has fast become one of the election reforms’ most contentious issues—the reliability of electronic voting machines. HAVA simply requires states to update their technology and leaves discretion on security issues up to the states. In New York, the jury is still out on what should be done to ensure the security and reliability of electronic voting machines.

Both the New York state Senate and Assembly each passed its own legislation amending the state’s election laws and allocating funds for HAVA compliance before the end of session, but a compromise wasn’t reached. The Senate’s legislation didn’t even address many of HAVA’s more complex issues, like voting machines or a statewide voter- registration database.

The Assembly passed legislation sponsored by Keith Wright (D-NYC) requiring new voting machines to produce a verifiable paper audit trail, and for 2 percent of these receipts to be checked against the electronic records as a safeguard. NYPIRG’s Rosenstein, who represents the New York State Citizens Coalition on HAVA Implementation, supports the Assembly legislation.

“No electronic technology can be safe and secure, and voter-verified paper audit trails are the most obvious backup that you can employ,” Rosenstein says. “It is easy and clear for the voter to understand, and it is reassuring to the voter.”

But a draft of the state’s plan for HAVA compliance, written by the state Board of Elections, doesn’t mention the need for a verifiable paper trail. “You don’t get a piece of paper out of the mechanical machine either,” says the Daghlian from the state Board of Elections. The state Legislature may address the issue when it reconvenes in the fall, Daghlian says, but the Board of Elections is ready to move forward regardless.

Sequoia’s Tonelli says his company’s machines can easily be retrofitted with printers to produce a verifiable paper trail, but he doesn’t see a need for them. “I thought the whole idea of this election reform was to get away from paper to begin with,” he says.

Though Rosenstein says his group supports the idea of a verifiable paper audit trail, he readily admits that it is not a cure-all. Visually disabled voters would still be shortchanged by election reform if they had to rely on someone reading from a piece of paper that said who they voted for. Unless voting machines were equipped with data-to-voice technology, so they could hear verification of their vote, visually disabled voters would lose the right to privacy that is one of HAVA’s cornerstones.

Access issues aside, it doesn’t take Rosenstein too long to think of other reasons why a paper trail may not be in the best interests of the public.

“Maybe these paper printers [for the audit trail] will be the worst thing in the world,” Rosenstein says. “The paper could get damp sitting in these warehouses, and it’ll jam in the printers and create longer lines and cost more money, or after every election people will want a recount: ‘Show me the paper, show me the paper.’”

Rosenstein wonders what effect all the speculation about the security of electronic voting machines will have on voter confidence, one of the issues HAVA set out to address. Whatever decision they may make, Rosenstein says that the legislators, election officials and gadflies who influence the process must know that the integrity of our nation’s electoral process is at stake.

“Voter confidence in the integrity of the elections is probably just as important as some of these fears and concerns about people tampering with the systems,” Rosenstein says. “This could lead to the public’s increased indifference about elections, which may ultimately lead to a lower voter turnout.”