WORD COUNT 577                                                                                                                                                                            JULY 2, 2008

MANY VOTING MACHINES ARE STILL SUSPECT – by Dan Lopresti 

Pivotal events sear their images in our collective consciousness. Few Americans will forget scenes from the immediate aftermath of the 2000 U.S. Presidential election: confusion and consternation over initiated and aborted recounts, with officials in Florida grouped around crowded tables attempting to interpret a voter's intent in the face of “dimpled” and “hanging” chads. 

Recognizing a critical need to modernize the way we conduct our elections, Congress invested billions of dollars in the most sweeping overhaul in history, the Help America Vote Act, or HAVA. 

Unfortunately, while its goals are laudable, HAVA's implementation to date has been fundamentally flawed. We may soon be confronted by a different set of unforgettable images, this time involving electronic voting machines that purport to inform us who won and lost, but that we cannot completely trust because computer memories and programming are infinitely malleable. Indeed, as another presidential election draws close, it is quite likely we are worse-off than we were pre-HAVA. 

While the American public has a general awareness of the dangers posed by computer viruses, identity theft, “phishing” schemes, and the like, experts who study computer security understand a fuller range of issues that confront us. 

Left out of the debate when HAVA was first put in place, these professionals stepped forward when it became apparent that computer technology was being improperly applied. Driven by a belief in social responsibility and working with little or no outside assistance, they have identified a lengthy list of serious problems in commercial systems already used in elections. 

Poor software engineering practices, supposedly secure passwords widely publicized on the Internet, flimsy locks that can be picked in a few short seconds (or opened with ubiquitous minibar keys), and machines that are reprogrammable with only a minute or two of unfettered access, are just a few examples of the problems that have been uncovered. 

At the state and local government level, weak certification criteria and discredited testing authorities have only exacerbated the situation. These revelations have begun to take a toll, most recently causing the states of California and Ohio to discard most of their electronic voting systems after conducting the first comprehensive and truly independent top-to-bottom studies of such equipment. 

With the 2008 election just a few months away, the time to act is now. It may seem anachronistic, but many scientific experts believe that provisions for the inclusion of a physical record, in the form of hand- or machine-marked paper ballots, are central to guaranteeing safe and secure elections. Paper ballots provide verifiable evidence of a voter's intent that is readable by machines as well as by human auditors. 

The processing of such records is not without its own set of challenges; unlike the situation with purely electronic systems, these can be addressed through careful system design and procedural safeguards. 

Concerned computer scientists around the country are working to develop better election technologies, including methods for enhancing transparency and facilitating post-election recounts and audits. At Lehigh University, research funded by the National Science Foundation is examining the inclusion of paper as an integral component in trustworthy voting systems, including ways of mitigating the errors that arise in the computer processing of paper ballots and enabling extremely accurate manual recounts. 

We are indeed at a pivotal point in our nation's history. The decisions we make now will determine whether upcoming elections run smoothly or raise suspicion and incite divisiveness. Our voting machines must be above reproach because so much else depends on them. 

-- 

Dan Lopresti is associate professor of Computer Science and Engineering at Lehigh University and co-director of the Pattern Recognition Research Lab. He has authored more than 100 publications in journals and refereed conference proceedings on a wide range of topics and also holds 21 U.S. patents.  

# # # # #