Project LEAP Analyst Observes PBC 2100 |
Peter Zelchenko (pete@chinet.com)
for the Chicago Tribune March 21, 2002 |
I visited nearly 40 polling places March 19 and, at several of these, contributed to Project LEAP’s detailed observation of the new instant tallying systems. The PBC 2100 machines generally performed dutifully. Too many categories capable of an undervote caused most ballots to be rejected. Correction efforts were exacerbated by there still being no reasonable way for voters to visually explore the ballots.
The key problem, aside from those already inherent in the punchcard method, was in judge behavior. I found one judge repeatedly grabbing rejected ballots and holding them up to the light. At the other end of the spectrum, I witnessed one very honest judge, knowing he may not touch the ballot, contort himself into countless positions in order to pantomime to the voter exactly how to insert her ballot. Other observers reported seeing similar behavior.
Largely, it was not the judges’ hands, but their mouths, that got them into the most trouble. We discovered that what exactly the judge says to the voter, in the moment after the ballot is rejected, greatly influences what the voter will do next.
LEAP President Robert Hodge likens this moment to a sales pitch at Sears: The shopper approaches the salesperson with a product, who offers an improved model at a slighly higher price -- in this case, the time it takes to vote properly. "The shopper ponders this," says Hodge, "and concludes it’s not worth the time and effort." This triggers a minor sense of guilt in the shopper for "not taking the chance of seeing what’s behind Door Number 3." The better the sales clerk, the better the response. In this case, of course, the incentive for both judge and voter is simply to get it over with.
At the moment the ballot is kicked out, the precise words stated by the judge are essential. One judge was telling a voter, "You have to go back and vote some more." That voter took four round trips until out of frustration he finally begged the judge for some alternative. "Sure, I can override it with this button. Why didn’t you say so?"
In contrast, another judge was saying in a hurried tone, "You undervoted. Do you want the machine to accept the ballot?" This, highly likely to elicit an affirmative response, is subtly different from alternatives which encourage greater reflection. "Are you satisfied with your ballot the way it is?" at least posits the question more critically and in a more proactive, open-ended way. Stated this way, voters were more likely to return to voting.
Then we have the highly explicit but semantically unequivocal: "You may now return to the voting booth to continue voting or have me instruct the machine to accept your ballot the way it is." Several conscientious judges struggled to utter some semantic equivalent of this sentence and were rewarded by those voters who were willing to apply themselves in an otherwise somewhat hurried, sometimes bewildering atmosphere. LEAP’s Bob Hodge suggested perhaps the best shorthand: "Have you voted for all the candidates you wanted to vote for?"
There was no legal formula for expressing this option in English or any other language, and there is significant opportunity for abuse here. This judge station has now joined the ballot-handler’s position among the most highly abusable in the polling place.
These linguistic problems were exacerbated by the fact that neither Bob Wallace, in his largely ignored videotape, nor anyone else has clarified to any voter what it means to undervote or overvote. Judges, when asked, had terrible difficulties explaining. Try it yourself. Now try it in Spanish.
In the end, the PBC cannot show voters whether they voted in the way they had intended. The problem lies with the punchcard system, not with the PBCs.
Responding to criticism of the punchcard system, Cook County Clerk David Orr recently asserted that groups including the ACLU are "starting from the faulty assumption that voters don’t know what they're doing." More appropriately, one should say that the equipment doesn’t always know what each voter might want them to do, even when it is functioning perfectly. Those in today’s workforce who are more adept at operating machines and at making the minute fine-motor adjustments necessary to manage the machines’ idiosyncracies are less likely to have problems. It is not a question of human intelligence, but of who in the machine age is machine-savvy and whose exposure to them is more limited.
In addition, says LEAP’s Alice Rubio, "many, many voters are not paying attention to what they are doing." Thus, some of the blame still rests on the voter. However, proper system design, in theory, should greatly limit the effects of any voter and judge idiosyncracies.
At about $5,000, the PBC 2100 is one of several reasonable aids, assuming we are stuck with punchcards. It does everything it is supposed to do very well: It offers the voter physical control over the ballot to the end. It is very accurate about tabulating ballots. It has the potential to keep ballots securely and will limit the effect of a corrupt judge or voter if used properly. On the downside, because people are justifiably confused about how to use the ballot sleeve securely, the ballot is, for the first time in history, viewable by all in the polling place. Rejected ballots were regularly handled and scrutinized by the judges. Still, the problem does not lie in the PBC 2100, it lies with the punchcards.
The voting punchcard system, though of pure intentions because of its relatively simple engineering principles, is flawed at its core. Most importantly, each individual ballot has nearly 2,000 tiny notches cut into it in order to facilitate chad removal with a cheap stylus rather than a costly punch. This type of cutting job has no precedent in printing science that I know of, in my 25 years of printing experience. During each voting session, dozens of these minuscule squares must successfully be forced out of the paper using the tiny stylus, with visual access to the holes obscured by the mask. The mask must travel a precise distance, using the card as a kind of long pushrod, in order to allow clean punching. The styli are tiny and pointy, while the punchouts are larger and rectangular, demanding we put a tiny round peg into a giant square hole. Finally, chads can jam the Votomatic devices in much the same way as with any paper punch.
Add to this the fact that the ballot cannot carry the names of the candidates. With 456 positions, there is scarcely enough room for the numbers.
Taken all together, the punchcard system presses too hard against the limits of where human dexterity intersects with electromechanical tolerances and machine fallibility. The result is the failure of the franchise which we are witnessing, primarily impacting people who are not knowledge workers.
This is not to say that technology-rich systems will be an improvement, nor that we should return to pure paper-and-pencil voting. One must not throw high technology at a problem, nor blindly retreat to lower technology. Rather, technology should find its perfect place in a system. Whether this should be optical technology or computer technology should be reviewed under a magnifying glass. But until punchcard systems improve specifically in the voting booth and in the ballot design, they will not survive a fair test of reliability. The PBC 2100 works, and it has value. It does not solve certain important problems, and it introduces certain new problems.
Peter Zelchenko
Board member, Project LEAP
Co-Founder, AIGA Voting Experience Redesign Initiative (now Design for Democracy)