Not an hour into her first day as the first female speaker of the House, Nancy Pelosi (D-Calif.) received a chorus of boos from Republicans.
The inauspicious start of her historic tenure began when she and fellow Democrats noted that Congress might unseat Vern Buchanan (R-Fla.). The Florida secretary of state certified Buchanan's election, won by a margin of 368 votes, even though electronic-voting machines had lost 18,000 votes, 13 percent of the race's total, without a trace. The election is still being challenged in court by Buchanan's opponent, Christine Jennings.
The ruptured election is the most glaring example of e-voting failure at a time when the technology is under fire in the court of public opinion. Politicians, researchers, commentators, voters' rights groups and bloggers have all railed on e-voting in the wake of the scandal. But other studies, from equally credible sources, show something lost in the shuffle of the Florida election scandal: widespread improvements in vote counting.
| ||||||||||||||||||||||
Angry voters
It was 44 degrees and drizzling in Austin, Texas, on election night 2000. Thousands of true believers in then-governor George W. Bush's presidential bid had gathered around the Capitol despite the weather, along with a stadium's worth of international media and acres of satellite trucks. A wheel-chair bound mother cradled her infant in her arms as they waited for Bush to appear, delivering either good news or bad. And wait they did, until after midnight as, one by one, the media called the election — for Gore. And then, for Bush. And then neither.
Bush never appeared that night and his home-state followers were far from the only Americans to be upset by the election that both sides still claim to have won.
The whole election hinged on one state, Florida, where old-fashioned punch-card voting machines made Americans cringe during the recount.
In the wake of the 2000 election, voters' rights groups like VotersUnite! cropped up from coast to coast, advocating for every vote to be counted. In 2002, Congress passed the Help America Vote Act, budgeting $2 billion toward simplifying voting procedures and helping guarantee votes would be counted. And then Diebold Inc. jumped feet first into the touchscreen voting business, sending sales reps county to county promising that touchscreen technology would improve future elections.
And there were improvements, big ones, though it was hard to tell from TV.
For the better part of the 2006 election cycle, CNN's Lou Dobbs explained that e-voting machines, especially those from Diebold, would unmake democracy.
"As of right now, there is little assurance your vote will count," Dobbs said in September. "As we've been reporting almost nightly on my broadcast for more than a year, electronic-voting machines are placing our democracy at risk."
Then the Buchanan election fouled in Florida. It was a familiar problem in a familiar place with an unfamiliar twist. This time, there was no paper trail. The Congressional votes of 18,000 voters just disappeared and, unlike in 2000, there was no way to find them. An attempt to recreate the errors failed. In December, a Florida circuit court denied Jennings' attempt to open the machines' source code for examination.
The story pervaded daily news, and it wasn't long before the watchdog groups decried touchscreen voting. In January 2007, four groups advocating the need to make every vote count, VotersUnite!, VoteTrustUSA, Voter Action and Pollworkers for Democracy, released a collaborative report, entitled "E-Voting Failures in the 2006 Mid-Term Election," which detailed more than 1,000 complaints of e-voting machine failures, including direct-recording electronic machines (touchscreen machines), optical-scan systems (fill-in-the-bubble, SAT-style ballots) and electronic-ballot markers (usually used to help disabled voters mark ballots correctly).
Of the complaints reported, touchscreen machines received the lion's share at 760 complaints, compared to 209 for optical scanners and 57 for electronic-ballot markers.
The report highlighted some of the biggest problems:
From Johnson County, Ind., a poll worker reports about the ES&S iVotronic:
"After we used the procedure that was given to the inspector, three machines out of four were not available at the start of voting."
From Prince George County, Md., a voter reports about the Diebold TS:
"At 7 A.M. only one machine is working. Nobody seemed to know how to fix the other one. Long lines of people trying to get to work."
From Indian River County, Fla., a voter's experience with the Sequoia Edge:
"Family of three went to vote. Father and particularly daughter had trouble with touch-screen voting machines. Daughter had to touch desired candidate five times to get the right candidate. The machine kept going back to the candidate that she did not want (Katherine Harris). She and the father were both eventually able to vote correctly."
But that was just half the story.
Corporate whipping boy
Diebold, with $2.5 billion in annual sales coming mainly from its ATM business, has been a lightning rod of criticism in the e-voting industry. Perhaps it was because of former chief executive Walden O'Dell's adamant public support for Bush; or perhaps it's just Diebold is the biggest target in the industry. 
Not only did Dobbs continually drub Diebold on TV, BlackBoxVoting.org devoted long articles and photo shoots to picking apart every flaw in its e-voting machines, down to weak tabs holding in the power sockets.
Meanwhile, a report by researchers at Princeton University declared that Diebold's TS model machines opened the doors for fraud on a massive scale. The report describes multiple security vulnerabilities in the Diebold TS. For example, it asserts that the door to the memory card slot can be consistently picked by a researcher with modest locksmithing skills in less than 10 seconds. The report also states that most of the keys are the same or similar.
"With lever machines or paper ballots, a person with little technical skill can commit fraud at a local level by tampering with a few machines or stuffing some ballot boxes, but this is difficult to do on a large scale," J. Alex Halderman, one of the report's authors, said in an interview. "On the other hand, attacks on e-voting machines are more expensive, requiring a high level of technical expertise, but they can affect an election at a much larger scale, since they can be carried out against many voting machines with relatively little incremental effort."
Ironically, given the heat they had taken before the election, Diebold's machines weren't used in Sarasota County, Fla., the one place where voting-machine trouble uprooted a national election in 2006. And Diebold Election Systems spokesman David Bear is quick to point out the fallacies of the Princeton research.
"The biggest flaw in the Princeton report is the whole report is based on the assumption of complete, unfettered access to a system, and the willingness of someone to commit a felony and go to prison," he said. "What system is safe from that? Voting centers are living, breathing environments. There are checks and balances, and that's the problem with ‘What if?' scenarios."
Bear doesn't deny that one key fits many locks. But, he says, the study fails to acknowledge the logistical challenges of running an election, and other problems that could just as easily stop votes.
"I can see that using a different key for each device makes it more secure," Bear said. "But let's play this out. Look at Georgia, where they have 22,000 voting units. In almost all voting jurisdictions they have a Democrat and a Republican having a key. So you would have 22,000 districts with 44,000 keys. You're not going to have somebody coming in and picking a lock. You're going to have someone not being able to vote at all because you're going to lose one of 44,000 keys."
And if it seems conspicuous that a company known for its long-lasting, high-security ATMs would make voting machines with less-than-stellar plastic tabs on its electrical sockets, there's a reason for that, too: Counties can't afford top-quality equipment. Massachusetts Institute of Technology professor of political science Stephen Ansolabehere, who co-authored a group of reports as a participant in the Caltech/MIT Voting Technology Project, said during an interview that the project calculated the total amount spent on elections across the country.
"We went out and made an estimate in 2001 of how much money was spent on elections, and the total, all the office staff and everything, was only about $1 billion," Ansolabehere said, "which is about 1/15 of the amount spent on sewers. This is not a big, profitable industry."
A vast improvement
Ansolabehere, who also studied votes lost in elections between 1980 and 2000, said Americans had better chances of having their votes counted in 2006 than ever before.
"We miss about 1 percent of the votes in this country," Ansolabehere said. "Historically, we were missing 2 or 3 percent, so we've really cut the problem in half since 2000."
One state the Caltech/MIT project studied, Georgia, it rated as having the second-highest undervote rate (the portion of voters who conspicuously didn't vote for a presidential candidate) in 2000. That rate dropped from 3.5 percent then to .39 percent in 2004 after the implementation of Diebold's e-voting machines statewide.
According to the Georgia Secretary of State, the system saved 103,000 ballots from being improperly counted in 2004.
Still, security concerns and arguments over reliability issues abound in the media with no clear solution. The loss of 18,000 votes with no paper trail, and vendors' unwillingness to publicly expose their source codes (which the judge in the Buchanan case ruled would "gut" the companies' trade-secret protections), has spawned a new argument about transparency and accountability.
One possible solution, a mandatory paper receipt for voters, is often debated. Ansolabehere said the problem with a paper receipt is that it can lead to voter fraud: voters selling their receipts to candidates, for example, or employers and unions checking their employees' receipts. And creating a completely secure transaction without the option of a receipt, which has never been accomplished for financial transactions, has become a white whale for programmers.
Bear points out, though, that many Diebold TS machines generate receipts that voters can review but not remove, though many locales choose not to use them.
"There is almost this belief that technology has increased the likelihood of problems with elections," Bear said. "We count votes more accurately than we ever have before. It would be naïve to think there haven't been issues with elections through time. But, for the most part, the vast majority were inadvertent. They have not been ill-intended. We've lessened the likelihood of those inadvertent and ill-intended issues."
