Prevent unwanted presidencies with paper ballots

While attending MIT's Emerging Technologies Conference in Cambridge today, I quickly found out that this country is still plagued with many of the same electronic ballot problems as it had in the presidential election of 2004, and now there seems to be a move afoot -- as odd as it may sound -- to get back to paper to ensure accuracy and legitimacy of election results.

California Secretary of State, Debra Bowen, joined three other e-voting experts at MIT's Kresge Auditorium to address the public's concern with the accuracy of today's polling systems. Bowen, who took office in 2006, ordered a complete review of the state's voting technology, which produced some surprising revelations as to problems many states, not just California, may face come November. Bowen said the review was in response to a "backlash" against electronic voting systems.

Prior to 2000, people already were voicing concerns about new electronic voting systems but it was the "Help America Vote Act", passed by Congress in 2002 that brought the issue to a head. The bill included close to $4 billion in federal money to encourage states to buy or upgrade voting technology. "Many jurisdictions not knowing what technology to get, took that as a signal to buy DREs," according to Doug Chapin, director of electionline.org, which has conducted research and analysis of states’ election technology since 2000. DREs or direct-recording electronics instantly record votes through the use of mechanical buttons or touch screen displays. But, Chapin said DRE technology in many cases didn't pass security muster. Bowen testified to that fact, saying "all of the systems [in California} had security issues."

For example, California's touch screen machines were using two security stickers on one side of the machine to prevent tampering with inner components, but, Bowen said, with a screw driver "you can remove two screws from the other side, open the clamshell ... and have total access to everything there and close it back up again with the security stickers intact."

In other instances, Bowen's office found that voting machines were being delivered to polling places two or three weeks in advance, giving anyone with access to a memory card the potential to virally replicate or upload software as a means of changing votes as they were cast.

Yet another problem with touch screens, Bowen said, was residents would step up to vote and see the wrong ballot in front of them, a problem stemming from poorly designed software and the fact that California had 330 different ballots because of local elections.

From L-R, Jason Pontin, editor in chief of MIT's Technology Review; Dera Bowen, California Secretary of State; Doug Chapin, director of electiononline.org; Ronald L. Rivest, prof. of electrical engineering and computer science at MIT; Pamela Smith, president of Verified Voting Foundation

One method of addressing software issues associated with the vast majority of proprietary e-voting applications out there is to move to using open source, especially for applications residing on optical scanners, which have been particularly troublesom. The concern is that IT administrators can't look at the software to correct errors or tweak it for a particular county's needs. Open source would go a long ways to disclosing problems associated with today's propretary e-voting applications, Bowen said.

"In any election -- it's Murphy's Law -- there's going to be a problem," Bowen said. "The question is, is it the type of problem you can recover from." Bowen and others on the panel said a more basic requirement of any e-voting system, no matter the technology used, is there must be a paper trail left behind for auditing purposes after the electronic count.

Ronald L. Rivest, a professor of electrical engineering and computer science at MIT, said the two biggest issues facing e-voting technology are maintaining the privacy of the voters and the verifiability of the results. Rivest, who has a background in cryptology and computer security, said the public considers the hand count the gold standard, but that method is no longer viable considering America now has the "most complicated ballot system in the world."

Bowen agreed noting that California often has 120 races in a single election. "If you had to start a hand count of 120 different races at 9 o'clock at night, I think we'd introduce more errors," Bowen said.

All of the panelists agreed the country remains a hodge-podge of voting systems. The most widely used voting method is the paper ballot, only today it's a preprinted list of candidates and initiatives, the choice for which can be indicated with some type of marker that is then recorded electronically with optical scanning equipment. The next most popular voting method is electronic touch screens, which make up about 30% of the systems throughout the country. And, yes, there are still hand-counted ballots in several states. California is no exception and has its own patchwork of voting systems, with around 30 counties using electronically scanned paper ballots and about 27 counties still using DREs.

Broken technology has also created confusion among voters as states have gone back and forth between voting systems in a search for the right one. In Florida, for example, since 2000, counties have gone from using paper punch cards to DREs back to paper ballots in combination with electronic optical scanners. Confusion can often lead to casting the wrong vote.

While there are calls by the public and officials to develop a ubiquitous Internet-based voting system, such a method would still lack any reliable way of verifying who is casting the vote on the other end of the pipe.

"Because we don't have any kind of national ID card, we have no method for doing that," Bowen said. A much more pragmatic problem is avoiding the selling of votes through Internet casting because if someone is giving a PIN number, as is used with Internet stock proxies, there's no way of verifying the number actually came from the voter.

Another wrench in the works of e-voting is that there's no standardized method to choose technology, not just state to state but also county to county. "In California, as in most states -- if not all -- it's not the secretary of state who purchases voting technology. It's the counties," Bowen said.

There are 58 counties in California, ranging from 16 million people in Los Angeles County to just over 1,200 residents in Alpine County. So you're basically asking a county IT person who may or may not have crypto or physical security experience to purchase a voting system. The software is proprietary and in most cases the person purchasing it has no legal right to review the software, Bowen said.

Then there are the politics of e-voting.

"The CEOs of one of the major voting machine companies was featured in a fund raising ad in which he pledged to deliver the state of Ohio to President Bush," electionline.org's Chapin said. And, that is the point at which the electronic voting debate took a sharp turn to focus on verifiability.

Another problem that e-voting systems have introduced is "exact match" voter registration lists. Voters with unusual sir names, Latino names or just names with unusual spellings can confuse verification applications, Bowen said. Even someone who simply uses the name John on their driver's license but registered to vote under the name of Jonathan, could experience problems. "Computers are not good at knowing that John and Jonathan are the same person," Bowen said. "California went through this in 2006 and the exact match list resulted in 26% of all voters registered in Los Angeles County not going on the [registration] list."

"We have to get this under control in terms of uniformity," Bowen said.

Perhaps one solution to e-voting technology issue is to simply allow a do-over, a mulligan if a voter slips up and makes a mistake. The need for precision in any e-voting technology is extremely high, and as technology is changed out or upgraded, there is a greater chance for mistakes. All of which bolsters the argument for not abandoning paper.

Chapin believes that a paper audit wouldn't mean pollsters counting every ballot, but that taking a statistically significant review of the ballots in any given county would be an acceptable way of measuring the accuracy of an election.

FREE Computerworld Insider Guide: IT Certification Study Tips
Editors' Picks
Join the discussion
Be the first to comment on this article. Our Commenting Policies