In July 2003, Aviel Rubin, an untenured Johns Hopkins computer-science professor, was asked to analyze some software source code purportedly used in the Accuvote TSx voting machine, manufactured by Diebold. The Ohio-based company is one of several major vendors of Direct Recording Electronic (DRE) voting machines that almost 30 million voters, in almost 40 states, will use in the 2006 election.

Rubin and several graduate students quickly realized that Diebolds code was rife with mistakes and vulnerabilities, and laughably easy to hack. Worse, since the same code seemed to underlie every Diebold machine, a skilled hacker could conceivably corrupt every Diebold machine in America. These first-generation DREs leave no paper trail; voters have to trust their ballots are being properly recorded and counted.

Rubin has written an engaging memoir of his three years in the vortex of electronic voting controversy. In Brave New Ballot: The Battle to Safeguard Democracy in the Age of Electronic Voting, theres unmistakably a Mr. Rubin Goes to Washington quality to his narrative arc. Smart, idealistic computer scientist gets outraged about a perceived wrongand wades right into the shark-infested waters. There are billions of dollars at stake for the DRE vendors. Thousands of election officials have their reputations, even their careers, riding on the machines whose security and reliability Rubin is challenging.

Not that Rubin was a media naf. Before leaking their report to The New York Times, his team spent several hours at a whiteboard coming up with pithy quips. Discomfited vendors and election officials initially belittled Rubins freshman homework assignment. When Rubin proved adept at rebutting critics obfuscations at various public forums and legislative hearings, his integrity came under attack. Nothing like an Aha! Conflict of interest! angleespecially with a highly technical subject that few reporters truly understandto distract from the heart of an issue. (Rubin failed to disclose he was an unpaid advisor, with stock options, for another e-voting companythough the firm was more a hopeful vendor to Diebold than a competitor.)

For Rubin and other critics of DREs, the nightmare scenario is this: Some diabolical (read: corporate and/or Republican) hacker figures out how to steal thousands of votes, in key congressional districts. What could we do? Would we even know?Without a paper trail, could we ever know who had really won?

As Oregons secretary of state from 1991 until 1999, my job included oversight of elections, and I certainly agree with the importance of a paper trail. (During my tenure, we conducted three major election recounts for Congress, the U.S. Senate, and a contentious statewide ballot initiative.) Im thankful Oregon is one of the few states where DREs are virtually nonexistent (more on that later).

But is there any tangible evidence, after more than 100 million votes cast on DREs, that a deliberate fraud has actually occurred in any election to date? Rubin offers no specific example. In fairness, his critique doesnt require one. But by focusing on the specter of significant electronic voting fraud, Rubins critique of Americas election system suffers from an even bigger mistake than confusing the elephants tail with its trunk. In an unintended, but nonetheless real way, Rubinnot to mention the deep conspiracy theorists from whom he distances himselfhas fed a hyperactive (and bipartisan) obsession with the perfect election system that ultimately endangers our ability to revitalize the very essence of our democratic system, which is voter participation.

The biggest problem in American elections isnt deliberate fraud. Its garden-variety human error. Election officials keenly understand that mistakes and malfunctioning systems (technical and human) wreak far more havoc than actual fraud. (Al Gore literally lost thousands of Florida votes thanks to the infamous butterfly ballot designed by a Democratic county clerk.)

Virtually every DRE problem to dateand there are some dooziesfalls into this category. Red-faced election officials in Montgomery County, Md., (Rubins home state) cant blame Diebold for the debacle during last Septembers primary when election workers forgot to send voter access cards to the polls. Voters were turned away, or simply left, because they couldnt wait any longer. In desperation, some poll workers gave voters blank pieces of paper for ballots.

The exquisite dilemma, of course, is that DREs, with or without VVPTs (Voter-Verified Paper Trails), wont go away overnight. Billions have been invested in them. As Rubin notes, voters actually like DREs ATM-like convenience. And even as you read this, controversy and threatened litigation likely rage somewhere in America because one or more congressional candidates, short a few hundred or few thousand votes, will be crying foul.

May I see your Costco card?

No study reveals a strong Republican bias toward computer criminals who a) want to risk long prison sentences to commit election fraud and b) have the Oceans 11-like skill to pull it off. However, the legitimate, if overblown, concern about DRE-related fraud seems to be almost exclusively voiced (so far) by Democrats.

While Democrats fear the machinations of Diebold et al., Republicans have their own version of electoral horror: those vast hordes of non-citizens and illegal immigrants (read: Democrats) eager to take advantage of open voter registration laws. This specter is just as ephemeraland even less frightening.

The notion that widespread voter registration fraud exists is a long-standing obsession among many Republicans, who for the last few decades have fought virtually every innovation to help citizens overcome bureaucratic and logistical obstacles to exercise their fundamental right to vote.

Sure, registration fraud can happen. But how often? In my home state of Oregon, with two million registered voters, weve had all of a dozen complaints of illegal registration in an entire decade. Exactly two miscreants were found, and both were prosecuted. (One was a British con artist; the other a confused Vietnamese woman.)

False registration is a felony in virtually every state. And just how stupid do people like Rush Limbaugh think illegal immigrants are to avoid being detected by immigration servicesso they can then commit two deportable offenses by falsely registering and then voting?

A recent study for the U.S. Election Assistance Commission concluded there was little evidence of the type of polling-place fraud that so alarms the GOP. (This reports release has been delayed for months, reportedly by the EACs Republican members.)

Since absence of evidence isnt evidence of absence, to quote Donald Rumsfeld, GOP elected officials are undeterred in trying to impose new, restrictive regulations supposedly meant to combat this phantom threat. For example, Georgia and Arizona have passed laws to require photo identification to register and even vote. The U.S. House of Representatives has passed similar legislationalso mostly along party lines.

These bills have a surface appeal, but quickly raise a host of thorny questions. Must it be a drivers license or other state-issued ID? Can you show a Costco card? A utility bill (with a picture?) And since were now taking extra time at the polls, longer lines, anyone?

And what about the tens of millions of voters casting absentee ballots? Party operatives, including many Republicans are strongly encouraging people to vote absentee as part of their get-out-the-vote strategies. So now we put photo IDs into the ballot envelopesand, uh, anyone out there thinking about identity theft?

Fortunately, federal courts have moved quickly to end this nonsensea superior court judge struck down the Georgia law requiring voters to produce a state-issued voter ID at the polls.

Assume, for a moment, the purest of motives, and focus on the Republicans ostensible logic. Its not enough to have stiff penalties, or adopt refinements such as random spot checks of voter registration rolls against other government databases. We must prevent that last 0.0001 percent of possible fraud. The goal of a pure, fraud-free election system demands that we do whatever it takeshowever cumbersome and expensive.

This is not unlike the worldview that Rubin ultimately embraces, too. As a computer scientist, Rubin literally lives in a world of digital certainty, of ones and zeroes precisely arrayed to produce a desired result. Little wonder the idealist in him seeks a totally secure voting machine, a totally fraud-proof electoral process. But elections, by definition, are inherently incapable of being fraud-proof and 100-percent secure. And this is key: After a certain point, efforts to make them so do more harm than good.

A good analogy comes from the biological world. Healthy adults have small quantities of dangerous bacteria in our blood. Yet prescribing massive doses of antibiotics to kill them would ultimately compromise our immune systems natural abilities to counter these and other threats. What kills the bad stuff can also kill the good stuff.

So too with voting. The ability to cast a ballot is the essential democratic freedom. Trying to make the voting process 100-percent error proof inevitably diminishes that freedom. The key to dealing with election fraud is not its elimination, but its strict containment.

Mailing it in

So where does Rubins quest for the Perfect DRE lead him? On page 249, he describes the system he finds most promising. Its the brainchild of cryptographer David Chaum.

Voters receive two ballots with the candidates names in different order. Voters mark one of the ballots and keep the other one. Since the authorities running the election would not know which ballot each voter marked, the ballot the voter keeps can be used, through some mathematical manipulations invented by Chaum, to keep the election honest. Furthermore, the marked ballot can be stored in such a way that the voter is able to verify that the vote was counted in the final tally, and the tally can be performed by anyone.

Got that? At least Rubin has the grace to concede that the details of Chaums scheme are too complex and technical to list here, and therein lies the rub…. A fundamental problem would remain: the public would not understand how the mechanism works.

Bingo. If the public doesnt understand its own election system, whats to make citizens suddenly believe again in the integrity of that system? Would we then start issuing ballots in triplicate and concoct new, ever more complex algorithms? Rubins logic leads us to ever more sophisticated, technology-centric processes, to fix a problem thats created by an inordinate reliance on complex technology.

The biggest threat to the health of our democracy is low and declining voter participation. Even the atypically high turnout in the 2004 presidential election amounted to just over 50 percent of eligible citizens casting ballots. Midterm elections are now attracting just 35-40 percent participation. Far, far worse are primary elections, those largely overlooked first-round contests where the vast majority of candidates who win the dominant partys nomination are virtual shoo-ins for November. This is now 90 percent of Congress 435 seats, and perhaps an even higher proportion of our 7,382 state legislative seats. For these contests, turnout has been truly abysmalin many states 5 to 15 percent of the eligible population.

Various voter suppression tactics (legal and otherwise) contribute to this problem. But the most effective voter suppression tactic is something most Americans consider an untouchable icon: the polling place itself.
In 49 states, the basic premise of elections is that the citizen needs to come to the ballotand not the other way around. So reformers like Rubin focus on building a better (and more expensive) DRE. Others suggest early voting options with additional workers and polling stations set up for days or weeks before Election Day itself. Some even advocate a new national holidayall to get more voters to come to the ballot.

In other words, spend even more moneyand create more opportunities for error (and yes, even fraud) as you rely on ever-more complex machines, and even more temporary (often, ill-trained) workers needing to be vigilant at more locations for longer hours.

Far better than going even farther down this rathole (with or without DREs) is to look at an 18th-century innovation, the Post Office, and implement Vote by Mail (VBM). Reverse the dynamic: bring the ballot to the voter.

Since 2000, Oregon has conducted all elections by mail, sending a ballot to every registered voter about two weeks before election. Voters can return the ballot by mail or deliver it to an official election site. To prevent fraud, voters must sign the outer envelope. All signatures must be matched to actual voter registration cards before the ballots can be processed. In the 2004 election, Oregons voter turnout as a percentage of registered voters was 85 percentthe highest in the United States. Even more dramatic, the seemingly dismal 39-percent turnout of registered voters in our state primary also appears to be the nations highestcompared to states such as Illinois at 21 percent; Pennsylvania at 19 percent, Texas at 10 percent, and Virginia at 4 percent. (By the way, the median age of those who actually voted in Oregons primary was 60.)

Actually, citizens in all 50 states already vote by mail. More than 80 percent of Washington State voters will cast ballots this way in 2006: In California and several other states, well over 50 percent will do the same.
Rubins take on Vote by Mail? A casually dismissive half-sentence: Oregon, where citizens vote by postal mailanother terribly insecure system. Here again, Rubin, like so many other election reformers, falls into the trap of making the (unattainable) perfect the enemy of the good.
Could an individual voter theoretically scam Vote by Mail? Sureas you can any system. Evidence its happened in Oregon? One person convicted of forging his wifes signature; another, of voting twice. More to the point: In stark contrast to a DRE world, even if individual or small-scale fraud were to occur, by design a VBM system disperses the risk, rather than concentrates it.

Vote by Mail is good for participation. Its simple to understand. Recounts are easyby definition, theres a direct paper trail for every vote. And face it. If election night brings rain, snow, or sleetor sick kids and dinner to get on the table many (maybe most?) voters would be relieved to know their votes had already been cast by mail. (Question for pundits: Whats the message voters will be sending if Election 2008 is literally decided by unusually bad weather?)

Todays national debate about how we should vote is still such a sad mix of fossilized thinking laced with bipartisan paranoia. Lets brace ourselves for a flurry of post-election accusations and lawsuits, that might make Florida 2000 look tame by comparison. (Perhaps even pray for one side to win so big, it wont matter as much.) And then, the day after, start figuring this out for real.

Phil Keisling

Phil Keisling, a Washington Monthly contributing editor, served as Oregon secretary of state (1991–99) and is currently the director of the Center for Public Service at Portland State University.