California and Ohio were among the first states to enact legislation requiring that electronic voting machines generate a contemporaneous paper record, or voter verified paper audit trail (VVPAT). Recent developments in California show that these laws, though well-intentioned, were a bad idea from the beginning and should be repealed. In reality, the challenges presented by electronic voting are too complicated to be resolved by such simplistic solutions like the VVPAT, which is likely to cause more problems than it solves.
Secretary of State Bruce McPherson recently tested Diebold's VVPAT-equipped voting system, finding printer jams on 10% of the machines tested. Diebold, for its part, counters that all the votes were accurately captured and that problems occurred only with a small percentage of attempted votes. But printer jams are only the beginning of the problems with VVPAT laws, and the problems go well beyond Diebold.
In 2004, California took a big risk by requiring electronic voting machines to have a VVPAT, before this device had been proven workable or effective. The widely criticized Carter-Baker Commission Report also embraced this device -- even though the scant evidence it cites for the proposition that the VVPAT "appear[s] to have worked well" actually supports precisely the opposite conclusion. Now, there's new legislation on California Governor Arnold Schwarzenegger's desk that would make the paper record the official ballot for purposes of recounts.
As an initial matter, it's worth asking why we should assume that paper records are safer than electronic ones. After all, there's a long history of ballot-stuffing and other types of fraud with paper ballots. By contrast, there's no documentation of votes having been altered or otherwise "stolen" with electronic voting. That doesn't mean that electronic voting is foolproof. It does mean that we should be skeptical about assuming that paper is the answer. The only state to have experimented with the VVPAT on any significant scale is Nevada . What little serious study has been done on the Nevada experiment provides reason to question the efficacy of paper printouts, given that very few voters even bothered to check them.
Now, there are new reasons for questioning whether the VVPAT is a good idea. California Secretary of State Bruce McPherson has opposed SB 370, a bill recently passed by the state legislature to require that the VVPAT be used in manual recounts of one percent of precincts. The bill is now on the Governor's desk, and must be signed or vetoed by October 9. The problem, according to McPherson, is that those records don't meet the legal definition of a ballot. Also, blind voters can't see and therefore can't verify it. McPherson's right to question the wisdom of this proposed legislation, and Governor Schwarzenegger would be well-advised to veto it. The four leading congressional co-sponsors of the Help America Vote Act ("HAVA") have expressed a similar concern, stating in a letter to colleagues, " Not only are such proposals [for VVPAT] premature, but they would undermine essential HAVA provisions, such as the disability and language minority access requirements, and could result in more, rather than less, voter disenfranchisement and error. "
Moreover, no one has yet attempted to recount all those lengthy paper strips in an effort to achieve an unambiguous, clear election result. Even if the paper strips from one percent of precincts were recounted, it won't serve as an effective check on election results, particularly if voters don't check the paper printouts. VVPAT advocates argue that, if the paper records aren't recounted at all, they'll be useless. On this point, they're quite right. Unfortunately, they're not likely to do much good even if the paper strips from one percent of precincts are recounted. The problem is that some advocates latched on to the VVPAT as the one and only solution, without thinking through whether it was a workable or effective solution to the asserted problem.
The reality is that paperless electronic voting systems have been successfully deployed elsewhere, giving voters with disabilities equal access to a secret ballot and contributing to the estimated million votes saved in the 2004 elections. To be sure, there are genuine concerns regarding the security and transparency of electronic voting. But there are other mechanisms by which to pursue these values are much more likely to be effective. One example is parallel monitoring, already employed in California , which in 2004 showed that electronic voting systems performed with 100% accuracy. Another possibility is to develop technology that relies on "open source" software.
Rather than demanding simplistic solutions like the VVPAT, we should be pursuing alternatives that will genuinely enhance the security and transparency of electronic voting. That's essentially the approach recommended by an excellent report entitled Asking the Right Questions of Electronic Voting, recently published by the National Academies' Committee on a Framework for Understanding Electronic Voting. In marked contrast to the Carter-Baker Commission, the National Academies report takes a research-driven approach to the problems presented by electronic voting. The committee finds that "electronic voting systems offer potential for voting and election management that is an improvement over what has thus far been available," but that realization of this potential will require:
ongoing effort that includes support for a new national research process, with research laboratories at the national, regional, or state levels; the implementation of research and development efforts to resolve the security and usability issues associated with existing and new election technologies; a lasting commitment to open and dynamic standards, testing, and certification efforts for election technologies; and ongoing efforts to educate election officials, poll workers, voters, and the general public about these new election technologies....
In other words, improving existing voting systems isn't nearly as simple as hooking up a printer to them. The report wisely avoids getting sidetracked on the issue of whether to require a VVPAT, instead recognizing that the present debate "has been carried out in the absence of substantial empirical data about how a VVPAT would actually work in the context of direct recording electronic systems."
Although the VVPAT may serve as a placebo for some voters, the consequence of this requirement may well be that states like California won't be in compliance with HAVA's 2006 deadlines, and face the prospect of giving back millions of dollars to the federal government. Worse still, it may lock in technology that will likely prove ineffective and obsolete in a few years. States that rushed to embrace the VVPAT would be well advised to abandon the false solution, and instead pursue the more promising approach advocated by the National Academies.
Unhappy trails are sometimes paved with good intentions.