Dimpled Chips

It seems perfectly reasonable that election officials in Palm Beach County, Fla. would have wanted a change in their voting equipment after the 2000 election. And touchscreen voting machines seemed like an obvious choice: Confusing butterfly ballots that had made the state a national laughing-stock were replaced by clear, well-labeled, brightly colored buttons; the machines were backed by the latest developments in counting technology (a field which has, somewhat counter-intuitively, apparently seen a fair bit of action in recent years); and most importantly of all, the nearest Chad would now be the one in North Africa.

What the election officials were probably not expecting, however, was the experience of one particular Palm Springs voter, who after patiently tapping through screen after screen of national and local officials fulfilling her civic duty was presented with an unsatisfying message of all too familiar a form: “Vote save error #1,” the machine said, “use back-up voting procedure.”

Voteprotect.org is a website run by a handful of nonprofit organizations including VerifiedVoting.org and the Electronic Frontier Foundation which collected and organized reports of voting irregularities during last Tuesday’s election. A cursory look at their data on problems related to the voting machines themselves reveals some interesting trends. The entire state of Massachusetts, where votes are recorded in large part by older optical scanning equipment, reported a total of 7 such incidents out of nearly 3 million ballots—one for every 500,000 or so votes cast. Palm Beach County had 27 machine related incidents out of their 550,000 votes—each voter there was roughly three times more likely to report trouble with their equipment than a voter was here.

These incidents ran the severity gamut. In Georgia, where all voting is done on touchscreen machines, voters complained of long lines due to malfunctioning machines or machines with dead batteries. There were complaints of slow machines, and machines which at first refused to accept the “smart cards” each voter used to identify themselves. Some machines crashed or went blank while they were being used.

Some machines, however, had bigger issues: “Voter’s machine defaulted to Republican candidate each time she voted for a Democrat,” one report from Cobb County, Georgia reads. “She told the precinct supervisor about the problem. It continued to happen 7 times.” Similar incidents occurred in reasonably large numbers—some voters tried to push a button for Kerry or Bush and found that the X would appear next to the name of the other candidate.

These problems were probably not due to a vast right-wing conspiracy in the voting machine industry. (Though it’s not entirely clear that such a conspiracy doesn’t exist—a board member of Diebold Election Systems, the company which makes most of the touchscreen voting systems that have been deployed, did at one point guarantee he would deliver Ohio to Bush in 2004. The promise sounds even more ominous in hindsight.) Rather, most of the issues surrounded poor “calibration” of the touchscreen inputs—the machines would register taps on one part of the screen as if they had been taps at some slightly different point. There were technicians on hand who could recalibrate the machines, and this tended to fix the problems for subsequent voters.

Still, as a result these machines relied on voters’ being sufficiently astute to notice when the confirmation said something different than what they had chosen, and sufficiently persistent to duke it out with the machines and complain to overworked officials when things went wrong. And for all the effort on the part of Florida officials to escape “close calls” due to fuzzy voting tools, these errors sound a lot like the dimpled chads they endeavored to replace.

Or they would, were it not for one more disquieting feature of most touchscreen voting equipment deployed in this election. Senator Kerry graciously conceded on Wednesday morning. But had he decided to fight it out and asked for hand recounts, it’s not clear what this would mean with respect to the new machines: They produce no printed receipt. In fact, they leave no paper trail at all. A lawsuit fought out in the Florida court system over the past six months tried to change this fact, but election officials have ultimately refused to deploy such equipment, calling it a frivolous expense.

I don’t mean to doubt that President Bush won this election fairly, and I don’t think touchscreen voting machines, for all their irregularities, tipped any balances. They even carried some ancillary benefits: Disabled persons, the blind in particular, were able to vote unassisted in a presidential election in large numbers for the first time in American history.

Still, we need to be able to trust the machines we use to vote. This doesn’t mean we need to understand how they work—no one should have to know that their voting machine employs strong cryptography to keep their votes safe. It does, however, mean that in our fleeting interaction with the machines, we have to be made confident they’re doing what they’re supposed to do. We don’t, for the most part, trust our computers. We save often, we’re told to run virus-scanning software—and still, we hear stories about lost theses and long treks in the snow to find working printers. But I think it’s quite clear at this point that while it takes a few days (and maybe a letter grade or two hit) to rewrite a history paper, it can take years to recover from a presidential election.

Matthew A. Gline ’06 is a physics concentrator in Quincy House. His column appears on alternate Mondays.