Research and Reports

Counting Votes

November 7, 1988

By Ronnie Dugger

During the past quarter of a century, with hardly anyone noticing, the inner workings of democracy have been computerized. All our elections, from mayor to President, are counted locally, in about ten thousand five hundred political jurisdictions, and gradually, since 1964, different kinds of computer-based voting systems have been installed in town after town, city after city, county after county. This year, fifty-five per cent of all votes—seventy-five per cent in the largest jurisdictions—will be counted electronically. If ninety-five million Americans vote on Tuesday, November 8th, the decisions expressed by about fifty-two million of them will be tabulated according to rules that programmers and operators unknown to the public have fed into computers.

In many respects, this electronic conversion has seemed natural, even inevitable. Both of the old ways—hand-counting paper ballots and relying on interlocked rotary counters to tabulate votes that are cast by pulling down levers on mechanical machines—have been shown to be susceptible to error and fraud. On Election Night, computers can usually produce the final results faster than any other method of tabulation, and so enable local officials to please reporters on deadlines and to avoid the suspicions of fraud which long delays in counting can stimulate.

Recently, however, computerized vote-counting has engendered controversy. Do the quick-as-a-wink, computerized systems count accurately? Are they vulnerable to fraud, as well—even fraud of a much more dangerous, centralized kind? Is the most widely used computerized system, the Votomatic, which relies on computer punch-card ballots, disenfranchising hundreds of thousands of voters?

It appears that since 1980 errors and accidents have proliferated in computer-counted elections. Since 1984, the State of Illinois has tested local computerized systems by running many thousands of machine-punched mock ballots through them, rather than the few tens of test ballots that local election officials customarily use. As of the most recent tests this year, errors in the basic counting instructions in the computer programs had been found in almost a fifth of the examinations. These "tabulation-program errors" probably would not have been caught in the local jurisdictions. "I don't understand why nobody cares," Michael L. Harty, who was until recently the director of voting systems and standards for Illinois, told me last December in Springfield. "At one point, we had tabulation errors in twenty-eight per cent of the systems tested, and nobody cared."

Robert J. Naegele, who is the State of California's chief expert on certifying voting systems and is also the president of his own computer consulting firm, has been hired by the Federal Election Commission (F.E.C.) to write new voluntary national standards for computerized vote-counting equipment and programs. Last spring, in San Francisco, at a national conference of local-election officials, I asked Naegele whether computerized voting as it is now practiced in the United States is secure against fraud.

He pointed a thumb at the floor. "When we first started looking at this issue, back in the middle seventies, we found there were a lot of these systems that were vulnerable to fraud and out-and-out error," he said.

I asked him whether he regarded as adequate the typical fifty-five-ballot "logic-and-accuracy public test" that is conducted locally on the Votomatic computerized punch-card vote-counting system—which about four in ten voters will use on November 8th—and he said, "No."

Would such a test discover, for example, a "time bomb" set to start transferring a certain proportion of votes from one candidate to another at a certain time, or any other programmers' tricks?

"Of course not," Naegele said. "It's not a test of the system. It's not security!"

The old mechanical machines prevent citizens from "overvoting"—voting for more candidates in a race than they are entitled to vote for—but the Votomatic systems do not. Not only can people using these systems overvote but election workers, if they are dishonest, can punch extra holes in ballots to invalidate votes that have been correctly cast or to cast votes themselves in races the voter has skipped. In the 1984 general election, about a hundred and thirty-seven thousand out of a total of 4.7 million voters in Ohio did not cast valid ballots for President—mostly, according to Ohio's secretary of state, because of overvoting. The computerized punch-card voting system is "a barrier to exercise of the franchise," and causes "technological disenfranchisement," Neil Heighberger, the dean of the College of Social Sciences at Xavier University, in Cincinnati, concluded in a recent study he made of the subject.

A federal judge, William L. Hungate, ruling last December on a lawsuit in St. Louis, declared that the computerized punch-card voting system as it has been used in that city denies blacks an equal opportunity with whites to participate in the political process. The suit was filed by Michael V. Roberts, a black candidate for president of the Board of Aldermen who in March of last year had lost to a white by a fourth of one per cent in a city election in which voting positions on ballots in the black wards were more than three times as likely not to be counted as those in white wards. Roberts, who was joined in the suit by the St. Louis branch of the National Association for the Advancement of Colored People, contended that computerized voting is such a relatively complex process that it is tantamount to a literacy test, and literacy tests have been prohibited by federal law as an unconstitutional burden on the right to vote. Judge Hungate found that in four local elections since 1981 voting positions had not been counted by the computerized system on anywhere from four to eight of every hundred ballots in black wards, compared with about two of every hundred in white wards (and also found that in the March, 1987, election the computerized returns from six per cent of the precincts had "irreconcilable discrepancies"). The evidence indicated that the computer had passed over the uncounted positions because of either overvoting or undervoting, which is failing to cast a vote in a race. The Judge ordered officials to count by hand all ballots that contained overvotes or undervotes and to intensify voter education in the black wards, but the city appealed, arguing that the racial differential does not always hold true in the city's elections. The Missouri secretary of state, Roy Blunt, called the order to recount the ballots by hand unfair and said that it could "make punch-card voting unworkable."

Link to article in the New Yorker

Rock, Paper, Scissors

October 13, 2008

By Jill Lepore

On the morning of November 2, 1859—Election Day—George Kyle, a merchant with the Baltimore firm of Dinsmore & Kyle, left his house with a bundle of ballots tucked under his arm. Kyle was a Democrat. As he neared the polls in the city’s Fifteenth Ward, which was heavily dominated by the American Party, a ruffian tried to snatch his ballots. Kyle dodged and wheeled, and heard a cry: his brother, just behind him, had been struck. Next, someone clobbered Kyle, who drew a knife, but didn’t have a chance to use it. “I felt a pistol put to my head,” he said. Grazed by a bullet, he fell. When he rose, he drew his own pistol, hidden in his pocket. He spied his brother lying in the street. Someone else fired a shot, hitting Kyle in the arm. A man carrying a musket rushed at him. Another threw a brick, knocking him off his feet. George Kyle picked himself up and ran. He never did cast his vote. Nor did his brother, who died of his wounds. The Democratic candidate for Congress, William Harrison, lost to the American Party’s Henry Winter Davis. Three months later, when the House of Representatives convened hearings into the election, whose result Harrison contested, Davis’s victory was upheld on the ground that any “man of ordinary courage” could have made his way to the polls.

Voting in America, it’s fair to say, used to be different. “Are you not a man in the full vigor of manhood and strength?” a member of the House Committee on Elections asked another Harrison supporter who, like Kyle, went to the polls but turned back without voting (and who happened to stand six feet and weigh more than two hundred pounds). The hearings established a precedent. “To vacate an election,” an election-law textbook subsequently advised, “it must clearly appear that there was such a display of force as ought to have intimidated men of ordinary firmness.”

What was at stake, in the House Committee on Elections, was whether men like George Kyle were too easily daunted. What wasn’t at stake was that Kyle, an ordinary voter, carried with him to the polls a bundle of ballots. Nowhere in the United States in 1859 did election officials provide ballots. Kyle, like everyone else, brought his own. The ballots he carried, preprinted “party tickets,” endorsed the slate of Democratic candidates, headed by Harrison. Voters got their ballots either from a partisan, at the polls, or at home, by cutting them out of the newspaper. Then they had to cross through the throngs to climb a platform placed against the wall of a building (voters weren’t allowed inside) and pass their ballots through a window and into the hands of an election judge. This was no mean feat, and not only in Baltimore. In the middle decades of the nineteenth century, eighty-nine Americans were killed at the polls during Election Day riots.

The reform that ended this unsettling state of affairs was imported from Australia, and was not achieved in the United States until the eighteen-nineties. The American adoption of the “Australian ballot”—and the radical idea that governments should provide ballots—was hard fought. It lies, if long forgotten, behind every argument about how we ought to vote now, from the 2002 Help America Vote Act’s promotion of paperless voting to the more recent backlash, favoring a paper trail. And it is also, like every other American election reform, a patch upon a patch.

The United States was founded as an experiment in eighteenth-century republicanism, in which it was understood that only men with property would vote, and publicly, since they were the only people who could be trusted to vote with the commonweal, and not private gain, in mind. What went on in 1859 was something altogether different: voting was still public, but all white men could vote, and nearly seventy per cent of them managed to do so in the congressional elections that year, pistols and fisticuffs notwithstanding. From an eighteenth-century point of view, how we vote now looks even stranger. Casting a ballot remains the defining act of American citizenship. But, especially since the election of 2000, with its precariously hanging chad, many people worry that voting in America is a shambles and even a sham, that the machinery of our democracy is broken, crippled by confusing, illegible, and deceptive ballots; vote-counting devices either rickety and outdated or new, gimmicky, and untested but, in any case, unreliable and by no means tamperproof; and a near total absence of national standards and federal oversight. In this fall’s Presidential election, every citizen who is eighteen or older—except, in some states, prisoners and felons—will be eligible to vote. Somewhat more than half of us will turn up. We won’t be clobbered, stabbed, or shot. We will not have to bring our own ballots. We will insist that how we vote be secret. The founders didn’t plan for this. No one planned for it. There is no plan. It’s patches all the way down.

Americans used to vote with their voices—viva voce—or with their hands or with their feet. Yea or nay. Raise your hand. All in favor of Jones, stand on this side of the town common; if you support Smith, line up over there. In the colonies, as in the mother country, casting a vote rarely required paper and pen. The word “ballot” comes from the Italian ballotta, or little ball, and a ballot often was a ball, or at least something ballish, like a pea or a pebble, or, not uncommonly, a bullet. Colonial Pennsylvanians commonly voted by tossing beans into a hat. Paper voting wasn’t meant to conceal anyone’s vote; it was just easier than counting beans. Our forebears considered casting a “secret ballot” cowardly, underhanded, and despicable; as one South Carolinian put it, voting secretly would “destroy that noble generous openness that is characteristick of an Englishman.”

Link to article in the New Yorker