Even if you have followed every strange twist and U-turn of the 2004–05 Washington gubernatorial election, would it surprise you to hear that more than 100,000 votes were never recounted? These are the electronic votes from the two counties, Snohomish and Yakima, that have computerized touch-screen voting in Washington. The computers do not include a mechanism to recount and compare votes for error. More than 2.7 million paper votes statewide were recounted by optical-reader machine and by hand, ultimately giving Democrat Christine Gregoire a hairbreadth 129-vote victory. But the 106,000 touch-screen ballots—constituting almost 4 percent of the state vote—were simply re-totaled without review and added in. A new study questions the validity of many of those touch-screen votes, suggesting that Gregoire should have beaten Republican Dino Rossi in the initial tally of ballots on Election Day. Rossi was on top at that point by 261 votes, before a final hand recount gave the election to Gregoire.
Officials, including the Republicans who are suing for a new election, don't seem too bothered by the unvetted touch-screen votes, despite promises to explore all the breakdowns of the state's ballot botch. Snohomish County elections manager Carolyn Diepenbrock, for one, says the election process was as honest and accurate as her county could make it. Last fall, she says, Democratic and Republican party representatives randomly selected some of the Sequoia-brand touch-screen machines to be used to test the precision of electronic voting. And on Nov. 2, the system worked, insists Diepenbrock. "Altogether, we deployed 956 machines, I think, out of the 1,000 we own. On the dozens of machines that were preselected by the parties, we ran parallel monitoring tests all Election Day. The sampling showed every vote was being recorded."
The nagging question: Were they recorded accurately? Diepenbrock and other officials in the state's third-largest county, run by Democrats, think so. About one-third of Snohomish County's voters—90,000 or so—used the computerized machines, and another 16,000 voted electronically in Yakima County. At least 28 Snohomish touch screens had "calibration issues," officials concede. That includes electronically shifting votes to a candidate other than the one selected by the touch-screen voter. Some screens also appeared with votes already selected by the computer. Similar problems cropped up in earlier elections in the county, which has employed touch screens for three years. But, "No votes were impacted by those errors," maintains Diepenbrock, because voters alerted poll workers to the glitches and the errant votes were corrected.
That's the opposite of findings by Paul Lehto and Jeffrey Hoffman, who have just completed a study of the Snohomish vote. They contend the outcome of the Nov. 2 election was affected by electronic irregularities. Their 29-page report found that problems of switched votes or machines freezing up occurred at more than 50 polling places. Under their scenario, Gregoire likely beat Rossi in Snohomish County or, at worst, lost by a much narrower margin—meaning she likely lost enough votes to account for Rossi's 261-vote statewide lead shortly after Election Day. Two-thirds of Snohomish County voted with paper absentee and provisional ballots, favoring Gregoire over Rossi by 97,044 to 95,228. The remaining one-third, voting electronically Nov. 2, favored Rossi—and by a much greater margin: 50,400 to 42,135. Though they can't prove it, Lehto and Hoffman think many Democratic votes were switched to the GOP. They believe it was "a mathematical impossibility that Gregoire's 1,800-vote lead on absentee paper ballots was completely overcome by an 8,000-vote Rossi landslide on Election Day on the Sequoia touch screens, ultimately leaving Republican Rossi with a 6,000-vote margin in traditionally Democratic Snohomish County."
Election Day trends statewide suggest that the Republican turnout on Nov. 2 was higher than the Democratic vote; thus, more Republicans than Dems would have voted via touch screen that day. But Lehto and Hoffman point out that Election Day touch-screen balloters favored Democratic presidential candidate John Kerry by a narrow margin in Snohomish County. (Including paper votes, Kerry won the county by 22,000.) That suggests "no credible argument can be made that Election Day was dominated by Republicans in any overall sense." Lehto and Hoffman theorize that computers might also have automatically assigned under-votes—when a voter fails to choose one candidate or another—to a candidate not of their choosing, and they say a forensic analysis or audit of the machines' programming could settle that and other claims. The county and the machine maker, Sequoia Voting Systems of Oakland, Calif., won't allow that. Lehto says he might sue. Sequoia, and particularly another touch-screen voting-machine maker, Diebold of North Canton, Ohio (whose CEO promised to do whatever it took to get George W. Bush elected president), have had to fend off accusations of electronic ballot rigging ever since the 2000 Florida election debacle. On its Web site, Sequoia says its systems are tamperproof, adding that "Sequoia's customers from Palm Beach County, Florida, to the suburban Seattle county of Snohomish, Washington, continue to enjoy the accolades that result from having carefree, accurate, electronic elections in partnership with our seasoned team of dedicated technicians and former election administrators."
Lehto, a business law and consumer fraud attorney in Everett who is also a Democrat, and his brother-in-law, Hoffman, a professor of engineering technology at Northern Michigan University, spent the past couple of months going over the county's records, printouts, and other election data, some of it obtained through public records requests. Their study findings, issued in December, got lost among the recount chaos. "I personally am surprised that the Republicans are shouting fraud from the rooftops," Lehto says, "and yet the Lehto and Hoffman study is non-news for the mainstream media." Lehto, of course, has an ax to grind politically and believes electronic voting machines in the U.S. can be fraudulently programmed to favor one candidate—usually the Republican. But he doesn't come off as wildly fanatical about his biases. "It's phenomenal," says Lehto, "how lax so many people are about the security of elections—as if individuals will assassinate presidents and governors but would never resort to nearly undetectable nonviolent vote rigging."
Elections manager Diepenbrock, who read the touch-screen study, "would not agree with Mr. Lehto's analysis." She believes party vote trends are reflected in the process of voting and notes that Rossi ended up winning the county after the final, hand recount as well. (In Snohomish County, that was 145,628 to 139,189, although, again, the computer votes weren't really recounted.) Diepenbrock does support a proposal for what is called a paper verifiable audit trail, which can be later used for verification if a recount is needed. Another technology, in which an encrypted printout is issued by the machine to the voter, who can later check the vote on the Internet to see if his vote was recorded correctly downstream, is being used in other states. The Legislature is weighing such options, but if it approves a system, Diepenbrock says, "it won't mean anything if they don't fund it. The counties don't have the money."
Electronic verification of touch-screen votes was supposed to become a state requirement by 2006, but the paper trail might not gain legitimacy, as some hoped, says Lehto. "Or it may only be an audit tool and not a [legally verifiable] ballot at all," under proposed legislation in Olympia. Actually, says John Gideon, information manager for the national voter-rights group VotersUnite.org (which has the Lehto-Hoffman study posted at its Web site), an idea for another verification process, the voter-verified paper ballot, might also be losing ground. Officials had promised to implement that system, but it might be moved back a year and is in "danger of going away completely, to be replaced by voter verification hardware," he says. It's too bad, adds Gideon. All that political noise about delivering complete vote-count accuracy might simply be "more smoke and mirrors."