Eight years have passed since the autumn of the hanging chads. Most punch-card voting machines have been retired, and several billion dollars have been spent on new election hardware as well as the preparation of updated, computerized, statewide registration lists. A major federal elections law was passed in 2002 (the Help America Vote Act); a new agency (the Elections Assistance Commission) was created; and state officials across the country have been working hard to insure that "Florida" will not happen again.
Yet no one—or almost no one—is anticipating a smooth, uneventful election, free of technological problems or flawed registration lists. Already, numerous issues have surfaced in early voting states. In four, including West Virginia and Missouri, touch screen machines have been "flipping" votes: voters who touched the name of one candidate were greeted by screens indicating a different selection. Elsewhere, machines have simply stopped functioning for various reasons, including unreliable Internet connections. Voters who were certain that they were registered have arrived at the polls to be told that their names were not listed, obliging them to cast provisional ballots that may or may not ever be counted. In the key state of Ohio, the registration lists have already been the subject of multiple lawsuits. And with turnout expected to be extraordinarily high, even routine technical glitches like replacing printer cartridges and rebooting computers could cause problems for polling place staffers who are rarely trained to make on-the-fly mechanical adjustments or repairs.
But the most serious problems could come from the counting of ballots. All tabulation systems (including human ones) make counting errors—which is why recounts are mandated in close elections. But on Tuesday, some or all of the voters in nineteen states, including Georgia and New Jersey, will be utilizing paperless Direct Recording Electronic (DRE) machines that do not permit recounts—because there is nothing to be recounted. If anomalies in the totals for any particular machine or precinct should appear, there is precious little that can be done, except to pray that the size of the anomaly is smaller than any candidate's margin of victory. A congressional election in Florida 2006 (to replace former Secretary of State Katherine Harris, appropriately enough) was mired in confusion for months because the voting machines appeared to have produced an 18,000 vote undercount—while only 800 votes separated the two candidates.
How did we end up here, with fingers crossed on the eve of an election, eight years after the debacle in Florida? The answer resides, at least in part, in our turning all too rapidly to technology as a quick fix for the array of problems that surfaced during the 2000 electoral crisis. The meltdown of our electoral system that year did have technological sources but it had other roots as well, including the Electoral College, the supervision of elections by partisan officials (such as Katherine Harris), and the decentralization of election administration not only across but within states.
Instead of tackling the whole array of problems that bedeviled our elections (some of which had a heavy political valence), we opted—in characteristically American fashion—for a soothing, apolitical solution: buying new machines. To be sure (and to be fair to some dedicated public servants), some attention was paid to other issues, but it was new technology that was heralded as the magic bullet that would cure our electoral maladies. Most of the money that Congress appropriated for HAVA was expressly earmarked (pardon the expression) for the purchase of new voting machines. (So little money was budgeted for the EAC that it could barely carry out its quasi-supervisory and information clearinghouse missions.). And if states and counties were going to buy new machines, why not the newest and shiniest electronic ones? At a time when personal computers sat on every office desk and most of the country banked through ATMS, the choice of DRE voting machines was almost irresistible. Fast, crisp, accurate, with no messy paper ballots that were a nuisance to count and store. Any worries about the technology were quickly dispelled by the expert representatives of the handful of companies that manufactured DRE machines and who knew that their ship had come in on the wake of Bush v. Gore.
The rush to invest in new voting technology (not simply new machines but a new type of machine) was, as we now know, a mistake. Between 2002 and early 2008, DRE machines deployed during elections sometimes malfunctioned—not with great frequency but in ways that were uncorrectable and thus cast doubt on the recorded vote totals. At the same time, computer-security experts demonstrated that the machines could easily be hacked: in just a few minutes, adept experts could unlock a voting machine (often with a key available at any hardware store) and re-program it to guarantee the election of a preferred candidate. The first response to these problems was to promote the development of voter-verified paper trails generated by the DRE systems. More recently, many—if not most—experts have concluded that these printed paper trails are themselves inadequate: they don't guarantee the security and accuracy of the totals, and they're extremely clumsy to utilize.
The upshot of this multi-year experiment has been an increasingly broad consensus that paper ballots (read by optical scan machines) constitute the best available technology for voting. Several states that had purchased DRE machines, including Florida and New Mexico, have junked the modern technology in favor of paper ballots, and many others are in the process of doing so (leaving them in a state of improvisational technological limbo for this election). Happy are the secretaries of state in places like Minnesota that stuck with optical scan systems and never chased the electronic grail in the first place.
What this means for Tuesday is that votes will again be cast and counted, as they were in 2000, on a wide array of technologies, varying not only by state but often by county. Voters in some locales, moreover, will be given the option of utilizing old-fashioned paper ballots if they distrust whatever new technology is in place. We have a national voting system that is very much in transition—not only forward from the clearly flawed punch card and lever machines but also backwards from touch screens to paper and optical scans. In transition too is the effort (mandated by HAVA) to develop comprehensive statewide databases of registered voters; the shortcomings of these databases may prompt significant challenges to individual voters and to the tallies.
On balance, we have probably made progress in the last eight years: our voting systems are likely more reliable, and certainly better scrutinized, than they were in 2000. But we are not close to a worry-free election, a day when all citizens (and even voting experts) can calmly go to the polls with confidence that the will of the people will be smoothly and accurately translated into the vote tallies. The response of our political parties to the threat of election-day problems has been as quintessentially American as the rush to buy new technology: they are sending in the lawyers. A report last week indicated that 5,000 lawyers were being dispatched to Florida, and something similar seems to be unfolding in Virginia. Throughout the nation, many thousands of lawyers will be stationed at polling places or manning hotlines, ready to go to court at the first sign of trouble. Thousands of lawyers, flawed voter lists, and machines we don't trust: is this any way to run a democracy?
Alexander Keyssar is Stirling Professor of History and Social Policy at Harvard's Kennedy School. He is the author of The Right to Vote: the Contested History of Democracy in the United States.