Skip to content Skip to footer

Voting Technology Research Gets In-Depth

As head of the Voter Technology Research Center at the University of Connecticut, professor Alexander Shvartsman directs a team of computer scientists granted special access to the optical scanners used to count the state’s paper ballots.

As head of the Voter Technology Research Center at the University of Connecticut, professor Alexander Shvartsman directs a team of computer scientists granted special access to the optical scanners used to count the state’s paper ballots.

Connecticut uses the Accuvote optical scanner by Premier Election Solutions — now Election Systems & Software — along with about 20 percent of all precincts in America. Despite certification in numerous states, it wasn’t until after the scanners were on the market that widespread reports of memory loss and other problems surfaced over the past four years.

In cases where devices experienced complete memory loss, Shvartsman and his team think they know why: the battery-warning signal.

The situation is typical in the world of electronic voting technology where outside computer scientists often discover and solve problems of faulty voting equipment before manufacturers do. The group from UConn presents its findings on Aug. 10 at the fifth annual Voting Technology Workshop as part of the Usinex Security Symposium in Washington D.C.

Shvartsman believes the problems with voting machines in general have occurred because vendors released flawed products prematurely. “We do not believe that they knew this stuff was not good, and they released it,” Shvartsman said. “I think they simply underestimated how tricky it is to get it done right and how important it is to perform really in-depth analysis and testing.”

In a December 2009 paper, Shvartsman and colleagues broadly criticized the majority of optical scanners on the market: “Despite the growing popularity of these machines, they are known to suffer from various security vulnerabilities that, if left unchecked, can compromise the integrity of elections in which the machines are used.”

Similar findings are common among computer scientists, who routinely say electronic voting systems could be made a lot safer than they are today, and that average citizens could independently verify the results. But for a host of reasons, sometimes out of a vendor’s control, virtually no one in the product world is turning their most progressive ideas into reality.

With midterm elections looming and states facing limited resources, few changes are unlikely to make the 2010 election season any smoother than in years past when widespread reports of election system problems continued.

Not since 2007 — when California, Ohio and Florida commissioned widespread reviews that led states to recycle millions of dollars in touch-screen machines — have voting machine vendors improved security in any meaningful way, said Dan Wallach, associate professor at Rice University. Wallach, who specializes in secure voting technology, is associate director of ACCURATE, a clearinghouse of election integrity research.

“The actual products that are certified and in use in the U.S. still have significant security problems that have not been fixed,” Wallach said. “The prototypes out there have really sophisticated security features, but these are not to be confused with things people are voting on.”

Kimberly Gurzick, spokeswoman for ES&S — which controls upwards of 60 percent of the election market in America — refutes Wallach’s claim about security. If there’s a better way to make elections more secure, ES&S would be glad to hear about it, she said. “We have professionals that go out and research the most secure systems and software and back-end tools to integrate into our system to make them as secure as possible.”

Wallach points to a prototype he and colleagues at Rice created in 2008 called VoteBox. This “direct-recording electronic” or touch-screen voting machine uses a type of encrypted programming arithmetic called end-to-end verification, which could be applied to any voting technology.

“It’s safe for the election authority to publish all of the ballots because they are encrypted. And then you can add them up yourself,” Wallach said. “You can’t see how any individual voted, but you can still make sure the totals are correct.”

But none of the leading voting machine makers have adopted it or many other technologies to improve existing election systems. Their reasons include a weak economy, market consolidation and dysfunctional regulation.

Regulation by the federal government and most states actually discourages making improvements to existing voting systems, said Marcus MacNeill, vice president of products and partnerships for Austin, Texas-based Hart InterCivic, which controls about 10 percent of the U.S. market. Instead of making it possible to approve incremental fixes to hardware or software similar to Windows updates on a personal computer, most states require an entirely new and costly certification process.

“If I want to deliver a traditional bug fix in the state of California, there is no process where I can submit that piece of software and someone at the state looks at it or it goes under some sort of test lab to field that change,” MacNeill said. “I literally have to put a new system through the federal certification process. Once you pass that, it goes through the state process.”

Requiring that vendors completely re-certify voting systems is designed to prevent malicious software disguised as a patch, but MacNeill says it’s too much.

Certification by the Federal Election Assistance Commission, required by a majority of states, costs about $1 million, and state certifications, such as California’s, run about $300,000, said MacNeill. He and others at Hart have been lobbying governments to make it easier to fix known bugs, but so far only the state of Washington has adopted a suitable process, he said.

MacNeill described a constructive relationship with computer scientists such as Wallach, who’s worked with the company on reviews of Hart machines.

“My object is to leverage as much as possible everything that’s coming out of the academic community,” MacNeill said. “Why? Because that offsets my development costs. The problem is I can’t take any of that and apply it to the systems I currently have fielded because of the nature of the regulatory process. Nobody has the money or compelling need.”

Beginning in 2003, the Help America Vote Act created a $3.65 billion frenzy of new voting technology, a staggering amount of money toward systems that in many parts of America hadn’t been upgraded for decades. Despite the spending, the result was a rush of new technology — much of it unwise — without the foresight to accommodate changes.

Money is once again the biggest barrier these days to improving voting systems, said Pam Smith, who runs the nonprofit Web site and advocacy group VerifiedVoting.org. The group tracks voting systems and method of auditing elections in every county in America in a push to require a voter-verified paper audit trail.

“Our biggest challenge right now, even in jurisdictions where you’ve convinced them that it’s a good idea, is now where are they going to get the money?” Smith said.

In the absence of federal legislation, 30 states now require or provide what would otherwise seem a no-brainer: a paper record, verified by individual voters and retained by election officials.

New Jersey, which passed a law requiring the state switch to a voter-verified system by 2008, still hasn’t made the change because of the cost. The state’s use of AVC Advantage paperless direct-recording electronic machines by Sequoia Voting Systems, meanwhile, led to a lawsuit.

In Maryland, state election officials bought electronic poll books by Premier rather than paper-based scanners, which ended up causing problems of their own. Last fall the state postponed a bid request for new scanners when Premier merged with ES&S, creating a sole-source bid, or at least the perception of one, at the time.

Market consolidation increased last year when Diebold sold its Premier unit to ES&S, which now controls more than 60 percent of the election market in the U.S. For antitrust reasons, ES&S sold off its intellectual property from the merger to Dominion Voting Systems in June, giving counties with existing Premier systems the choice of two vendors. A few months earlier, Dominion purchased Sequoia, the third largest American vendor, after ES&S and Premier.

“The vendors will admit they do what the customer wants and the customer is not me the research guy,” Wallach explained. “The customer is the county clerks who write the checks. The county clerks aren’t saying they want any of this stuff, so that’s not where the vendors are going. Ultimately that leads to a push on the standards process to require them to build better products, and the standards process is slogging through molasses right now.”

At ES&S, demand for paper-based systems shifted focus from touch-screen technology over the past five years, Gurzick said. “As a company, we shifted back to paper voting equipment because a lot of customers were really skeptical about using the touch-screen machines,” she said.

The company’s latest release, the DS200 IntElect digital scanner, can count hash marks or half-filled bubbles, and it takes a digital photograph of every ballot that can be stored electronically. Still, it too faced criticism: A report by the Florida Fair Elections Coalition on extremely high over-vote rates in the 2008 presidential election led to calls to decertify the machines in Florida, New York and Ohio.

Wallach called for a new round of comprehensive reviews on the next generation of voting equipment. “I would be shocked if they were significantly better from a security perspective,” he said. “They certainly won’t be anywhere near what the research community has been doing.”

In South Carolina, Brett Burefy with the South Carolina Progressive Network said his group had opposed the iVotronic touch-screen voting machine made by ES&S ever since the state first proposed buying them in 2004 for $38 million. South Carolina bought them anyway. In 2006, the same type of machines, which do not provide a voter-verified paper audit trail, led to a hotly disputed result in Florida’s District 13 race.

A series of mishaps in South Carolina in 2008 led to a near meltdown during a Republican primary. That same year, Burefy’s group participated in the national election protection hotline, as they had for the past four years. “We got more than a call a minute for 14 hours in South Carolina,” he said. “We had hoped it would have been a message to our Republican friends that the question of these machines and how they can’t be audited is not a partisan issue.”

Even Democrats in a state with a Republican-controlled legislature and governor’s office appear unwilling to push for change, even after the controversial results of the recent Democratic Senate primary where sleeper candidate Alvin Greene won the nomination. Losing candidate Vic Rawls believes it could be a computer glitch, but without a voter-verified paper audit trail there’s no way to know.

“A recount with these machines consists of no more than pushing a button and getting the same number you got to begin with,” Burefy said. “With the economic situation we’re in, you first need the political will, then there needs to be the money to do something. We haven’t even established the political will.”

As for the money, Burefy said, South Carolina declined to perform a $100,000 audit of the recent primary. And even if it did, the audit would be unreliable.

Founded in late 2007 by philanthropist Sara Miller-McCune, Miller-McCune is a nonprofit print and online magazine harnessing hard data and breaking research to support journalism that focuses on finding solutions to social problems. Supported by a combination of grants and advertising, Miller-McCune rejects any overriding ideology, believing that the best answers can come from anywhere.

We’re not going to stand for it. Are you?

You don’t bury your head in the sand. You know as well as we do what we’re facing as a country, as a people, and as a global community. Here at Truthout, we’re gearing up to meet these threats head on, but we need your support to do it: We must raise $21,000 before midnight to ensure we can keep publishing independent journalism that doesn’t shy away from difficult — and often dangerous — topics.

We can do this vital work because unlike most media, our journalism is free from government or corporate influence and censorship. But this is only sustainable if we have your support. If you like what you’re reading or just value what we do, will you take a few seconds to contribute to our work?