Sometimes lo-tech is better

When we first started researching the problems attached to electronic voting back in 2004, we were tempted to advocate for the good old-fashioned lever systems that New York—and at that time, Wayne and Pike counties in Pennsylvania—already had. Some studies have shown them to be the most accurate voting method—other than occasionally jamming so that the wheels don’t turn over, there is no way for them to make a mistake—and their transparent mechanical method of counting votes is virtually impossible to hack.

But after a glance at section 301 (a) of the Help America Vote Act (HAVA), we decided that that idea just wouldn’t fly. In particular, the provisions that require that voters be able to verify in the voting booth that their vote has been recorded correctly, and that the machine produce a paper record that can be used for a manual audit, seemed to preclude lever machines.

But the recent advocacy of the machines by Sullivan County’s Senior Legislative Action Committee prompted us to renew our research, and we think the idea deserves another look.

In 2004, Douglas A. Kellner, then the elections commissioner of New York County, gave testimony that we find particularly interesting. He said, “Our lever machines satisfy all but one of those [HAVA] standards, that there be at least one machine at each poll site that is accessible for individuals with disabilities, including non-visual accessibility for the blind and visually impaired.” Since 2004, all polling places in New York State have acquired at least one disability-friendly machine. That means that in the opinion of at least one highly qualified expert, New York has now fully satisfied HAVA standards.

So let’s leave the haggling over legality for a moment and instead ask: what is the most reliable, fraud-free voting system?

The big advantage of lever machines is that they are impossible to hack on a wholesale basis. Someone with access to an individual machine might be able to tinker with the mechanism to alter the gears. But hacking one machine at a time is nothing compared to electronic machines, in which a single malicious program can be installed in thousands at a time, or in the central tabulator for dozens of precincts, or implanted wirelessly or by phone modem. And one factor that is widely overlooked in analyses of election security is that even with an optical scan system—the type scheduled to be adopted by most New York counties—in which the cards are individually marked by pencil, the scanners that tally the cards can be hacked.

To keep lever machines honest, all you have to do is open the back of the machines and look at the numbers on the internal counters, with observers from all parties present—in fact, you can observe the entire mechanism and verify visually whether it has been tampered with. Write the numbers down, put them on a piece of paper with a clear train of custody, and you know the results are reliable.

It is this procedure that lever proponents say counts as a paper trail for lever machines. They argue that it is no less rigorous than the supposed “paper trails” commonly accepted by champions of electronic machines. In the case of Direct-Recording Electronic (DRE) machines like those in Pennsylvania, which do not print out individual paper ballots, a paper trail is just a machine printout of the vote total. It has been demonstrated that such printouts can easily be programmed to differ from the internal electronic count used to produce the total for the entire election. And, without individual ballots, there is no way to compare them to the actual votes. This bogus “trail” is, if anything, less rigorous than that for lever machines.

In an optical scan system, the optical-scan cards do provide a paper trail. But the only way a discrepancy would be discovered between the cards and the electronic scanner total would be if there were a hand recount. Since any malicious program would obviously avoid giving results so close as to require a recount, we would have to rely on random audits to detect fraud.

We will have to let the lawyers dot the i’s and cross the t’s as to whether section 301 (a) of HAVA excludes lever machines. But it seems to us that the reasons opponents are citing for ditching them adhere to the letter rather than the spirit of the act. And proponents argue that even the expense of returning the Federal money received to implement HAVA, if they are not in fact compliant, would be far less than the costs of purchasing and maintaining the new systems.

The optical scanners do appear to be the best of the electronic options (though if they are adopted we believe that it is important to make sure a high percentage of random audits is performed). But we think our existing system is probably better, and encourage state officials to give serious consideration to keeping it.

Also in this issue:

• ‘I came upon a child of God’: musings on Woodstock

Dr. Punnybone

Slight of Hand