January 7, 2008
Glaring omissions in otherwise excellent NYT article on voting machines
“Can You Count on Voting Machines?” is the question posed by Clive Thompson in his cover article in this week’s New York Times Magazine. The answer, of course, is a resounding “No” due to flawed design, buggy software and poor quality control in the manufacture of these machines, as Thompson ably demonstrates. http://www.nytimes.com/2008/01/06/magazine/06Vote-t.html?_r=1&hp=&pagewanted=all&oref=slogin
Thompson’s solid article is a timely reminder on the eve of the New Hampshire Primary that the nation’s election system is still broken. The curtain was pulled back in Florida 2000 to reveal the sorry state of U.S. elections and, despite the efforts of activists and politicians, there is not as much progress as one would hope. Some states like Florida are making great strides toward transparent paper-based systems, while others like Maryland, Georgia, Pennsylvania and Virginia are still struggling to ditch paperless voting machines.
However, there are a few areas of omission that need to be filled in and a few bits of mis-information that need to be corrected.
One thing that Thompson got right was the necessity of having paper as the record of the voter’s intent. However, he appears to have missed the point as to why paper ballots are needed. In short, what can you do with paper ballots that you can’t do with the electronic paperless voting machines? The answer is simple — audits and recounts! Thompson does not even mention the power of statistically sound randomly sampled audits of paper ballots as a check on the integrity and validity of the machine count. What is standard operating procedure for business, namely routine audits, should be applied to elections. There is growing consensus that mandatory routine post-election audits conducted with statistically sound sampling provide a needed confidence-builder for all stakeholders — citizens, election officials, candidates — in the accuracy of the election results.
The so-called VVPAT (Voter Verified Paper Audit Trail) on flimsy cash-register-tape-paper that is attached to many touchscreen machines makes it difficult if not impossible to conduct meaningful audits and recounts.
However, the optical scan ballot marked by the voter makes these necessary checks on the integrity of the election results practical. This is why so many election integrity advocates are calling for the use of paper ballots counted on optical scan machines. Not because optical scan machines do not make mistakes but because it is possible to discover the errors through hand-counted scientific audits of the paper ballots. Then in closely contested races where possible errors are discovered that might flip election results, the entire race can be recounted. While many states have already passed laws requiring post-election audits (Minnesota, New Jersey, Connecticut, Oregon, North Carolina, to name just a few), these audits are by no means universal. Until there is paper to audit and recount, meaningful audits cannot be conducted in the numerous jurisdictions that have paperless voting machines — and they will be very challenging in states with VVPATs because of the flimsy nature of the thermal paper used. Plus, as Thompson points out, many printouts on these machines are lost due to printer jams and related problems with the cheap printers attached to these voting machines.
Speaking of optical scan systems, Thompson relies on nameless “experts” as the source for time needed to implement them:
There are also serious logistical problems for the states that are switching to optical scan machines this election cycle. Experts estimate that it takes at least two years to retrain poll workers and employees on a new system.
There are serious problems with the above statement beyond use of an anonymous source — states like North Carolina have successfully made the switch to optical scan in months, not years. With good planning Ohio should be able to accomplish the changeover within the time allotted.
Since optical scan uses familiar paper and pencil for marking the ballots the entire process is more intuitive for both poll workers and voters making the transition as painless as possible. Take a look at what David Orr, County Clerk in Cook County, IL, has to say about the reasons why he likes optical scan systems. http://www.voterinfonet.com/sub/news_view.asp?NEWS_ID=125
1. Optical-scan voting is more intuitive
• Voters make choices by shading in oval-shaped bubbles next to a candidate’s name, similar to a standardized test. Because each candidate’s name is printed directly on the ballot, optical scan provides an intuitive method of choosing candidates that is most familiar to voters, particularly those who have never voted before.
2. Fall-off rate/disparities reduced with optical scan
• Optical-scan systems register low percentage rates of unrecorded votes. Across the board, optical scan systems not only record low fall-off rates, but also the disparity between poor/minority and rich/white precincts is significantly reduced, compared to punch cards (see attachment).
3. Mistakes easy to recognize
• Because voters interact directly with an optical scan ballot, they can easily view their work, reducing the potential for errors and eliminating any guesswork in terms of a voter’s selection.
All of the statements about the interaction of voters with optical scan systems applies equally to poll workers, making the training time needed minimal. Some have complained that Cuyahoga county in Ohio (Cleveland) might not be able to easily make the switch to paper: however, it is often overlooked that punch cards were used in that county as recently as the 2004 election. Therefore poll workers would be familiar with paper-ballot procedures and would need only a refresher course not a complete re-training in totally new technology as is the case when touchscreen machines are deployed for the first time.
Another area where Thompson presents only a partial picture is about the issue of Open Source software, an area for discussion that is deserving of its own post. So look for a later analysis of the issues surrounding the use of Open Source software in elections. For now, let me just note that Thompson did not reference any of the major resources on Open Source Software nor provide any clear definition of what he meant by Open Source. He appears to have conflated publicly disclosed software with Open Source software — and the two are not the same thing. But I will stop right now before I get too deeply involved in the finer points of attempting to answer the question “What is Open Source software?”
One final brief comment on the issue of whether voting machines have been hacked in past elections. Thompson tends to dismiss suggestions that hacking might have taken place. However, this is an area where no one can say with any certainty that the machines have not been hacked. Without paper ballots and robust mandatory audits there is no way prove it one way or the other. Certainly many of the anomalies in recent elections can be explained by pointing to software bugs. But insider access remains one of the factors which must be taken into consideration in any threat assessment of voting machines. That is another reason for the clamor for paper ballots, post-election audits and good chain-of-custody procedures for U.S. elections. Until those are in place across the nation, hacking remains a distinct possibility. American voters deserve better.