To steal elections is human nature. In 1948 a U.S. Senate seat was stolen, by simple ballot-box stuffing, for Lyndon Johnson; without it he would never have reached the presidency. John Kennedy only won the presidency in 1960 with the support of dead voters in Chicago. But now we approach a national election that is susceptible to theft in its very machinery—not just in Florida but almost anywhere.
Roughly three out of every ten of the ballots cast by voters on November 2 will vanish into direct-recording-electronic (DRE) computers the moment they are cast. These votes cannot be recounted independently of the computers because there will be no voter-marked ballots to recount. The exception is Nevada, where the obvious solution, a printer, will produce verified paper ballots. In at least twenty-five other states, the votes cast on DRE computers will be unrecountable, meaning that programmers will know their work cannot be checked against any evidence created by voters.
Although complex safeguards against error and fraud in computer-voting systems do exist, they are, most computer scientists agree, overwhelmingly inadequate. Computers can be programmed to steal elections and then erase all evidence of the theft. The reliability of the totals spat out by the DRE systems is dependent on the validity of their internal audit trails and “ballot images,” which can be rigged. Some computerized voting systems, such as those that scan punchcard or mark-sense ballots, at least have voter-marked paper ballots that can be recounted. But to “recount” the 35 million or so votes in DRE jurisdictions, officials will simply command the computers to regurgitate second printouts of the same results, which, in point of fact, are not recounts at all.
Especially troubling is the fact that the vote-counting process is now dominated by private corporations, whose machines (which include not just DREs but paper-ballot scanners) will count five out of six of all votes. Of the top five companies in the field—Election Systems and Software (ES&S) of Nebraska, Diebold Election Systems of Texas, Sequoia Voting Systems of California, Hart InterCivic of Texas, and Danaher Corporation of Illinois—at least two have strong partisan ties. ES&S, which will oversee the counting of more than half of all the votes, is part-owned by an investment firm whose CEO, Michael McCarthy, is a frequent contributor to Republican causes. Another 8 to 9 million votes will be tabulated in computers provided by Diebold, whose CEO, Walden O’Dell, caused a scandal by declaring he would help deliver his home state, Ohio, to George W. Bush.
Of the 110 million votes that will likely be cast this year in the most important election on the globe, 35 million will be unrecountable. Far from preventing another debacle like 2000, the DRE machines make one all the more likely. If either side asks for a recount in decisive counties but gets, instead, only second computer printouts, the 2004 election, too, may end in rancor, recriminations, and another four years of dubious legitimacy.
* * *
Where are the weakest points? Thirteen cities of more than half a million residents will cast their votes in ballotless DREs: Houston, Philadelphia, San Diego, San Antonio, San Jose, Columbus, Austin, Baltimore, Memphis, the District of Columbia, Nashville, Charlotte, N.C., and El Paso. So will four large urbanized counties in California and another four in Florida: Hillsborough on the Gulf and the tier of three counties on the Atlantic—Palm Beach, Broward, and Miami-Dade—whose crabwise gyrations mesmerized the country during the 2000 recount. Ninety-two U.S. counties with more than 100,000 registered voters will use DRE systems, accounting for 22 million votes in all. This includes counties in ten “swing” states—Colorado, Delaware, Florida, New Mexico, North Carolina, Ohio, Pennsylvania, Tennessee, Virginia, and Louisiana—that will record a total of 9 million votes on DREs, from 200,000 in New Mexico up to 3.5 million in Florida. In a close race any of these states could undermine the winner’s legitimacy; a few in particular are worth mentioning.
Five large counties in Pennsylvania (Philadelphia, Montgomery, Dauphin, Berks, and Beaver) will use unrecountable ballots, representing a quarter of the state’s votes overall. I asked Ed Schulgen—the deputy in the Philadelphia city commissioner’s office, which oversees elections—what he would do in the event of a recount. “We’d reinsert all the computer chips back into the main reader,” he said, “and print out the results.” What result would he expect? “The same one that we got before.”
In a recount, according to Steven G. Chiavetta, the director of elections in Dauphin, another DRE county in this crucial state, “What’s in those [computer] cartridges is never gonna change. You can run ’em over, and over, and over, and over, and it’ll be the same.” And how does Chiavetta know there are no election-stealing algorithms in his voting machines? “I know ’cause I program ’em, I do it myself!” he exclaimed.
Complicating matters in Pennsylvania is the fact that statewide recounts are exceedingly difficult to obtain there. To get one, a candidate must obtain notarized statements from three registered voters in each of the roughly 9,400 precincts, all in the span of five days. Moreover, the candidate must deposit at least $50 for every precinct; i.e., nearly half a million dollars, which is forfeited if neither fraud nor “substantial error” is found.
Almost a third of the voters of North Carolina live in the state’s four large DRE counties. Guilford County (Greensboro) was, according to George Gilbert, its director of elections, the first jurisdiction in the nation to use a touch-screen voting system, in 1988, and he claims that since then the county has “lost only four votes.” Although he does the local programming of the machines himself, he says that he does not even have a copy of the source code, the central set of instructions that controls the vote counting, which is supplied by ES&S. “You don’t test a system by looking at the source code,” he told me. “The program itself can be examined without ever looking at the source code with test-voting, simulating election conditions.”
Mecklenburg County, which includes the city of Charlotte, uses a DRE system from MicroVote of Indiana. “Hey, I don’t know any of that,” the county’s director of elections, Michael G. Dickerson, told me when I asked him about his computer security. “I watch it, how it works.” He dismissed concerns about vote stealing. “I’m worried whether my car starts—lots of things you worry about.”
In Ohio five counties with more than 100,000 voters each have been using DREs for years. This year the secretary of state, Ken Blackwell, a Republican, allowed more than thirty counties to consider new systems, but only if their proposed manufacturers met Ohio’s new security standards. According to Blackwell’s spokesman, Diebold, ES&S, and Hart InterCivic all failed to bring their DREs up to the standards. Consequently, of the eighty-eight counties in Ohio, sixty-eight, or about 70 percent of the state’s voters, will still be using the unreliable punchcard ballots on November 2. Furthermore, in a close election—and Ohio is expected to be close indeed—any of the DRE counties could swing the state one way or another.
Sequoia provides the software for the AVC Advantage DRE machines used for elections in Ohio’s Lake County. Linda Hlebak, the deputy director of elections, says that to do a recount without having paper ballots, “We will reset the tabulator to zero and reread all those cartridges. At the warehouse we reproduce those election reports that appeared election night.”
So it goes state to state. Ray Barrett, the administrator of elections in Nashville, told me a story about an old gentleman who, ten years ago, lost an election by 30,000 votes. The man couldn’t believe it, so they ran the cartridges through again to satisfy him. That was a second printout? “Right,” Barrett said. Was that a recount? “I don’t know if it’s a recount or not,” he said. “Anyhow we didn’t get sued over it.”
If, after this election, though, a demand for a recount in any decisive state is thwarted by its unrecountable votes, the nation could face a calamity far worse than a civil suit. Should voters come to understand that no record independent of the computers exists for nearly a third of the nation’s votes, the damage to the declared winner—and to the political system as a whole, perhaps to the democracy itself—would be staggering.
I asked the elections administrator in Knoxville, a fellow named Greg Mackay, what he would do if there were a close election this year. “Shit, I’d go home and get drunk!” he said. “No, we could run the cartridges through again.”
“’Course,” he added, “you’d get the same result. . . . It’d be the same damn thing.”
About the Author
Ronnie Dugger is the founding editor of the Texas Observer and the author of biographies of Lyndon Johnson and Ronald Reagan, among other books.
This is No Appeal, originally from November 2004, publishedwww.harpers.org/NoAppeal.htmlE-mail this article