Activists who still want the county to use a paper ballot ... or at least to have a paper record of every vote ... say the few pages spit out by the county's newly purchased electronic voting machines during last month's primary show why we can't trust this new voting technique.
As required by federal law, Allegheny County switched for the May 16 vote from lever machines to a system that allows more people with disabilities to vote for themselves. The county Board of Elections chose the iVotronic, a touch-screen computer voting system, from among several machines certified by the state.
While the county hails the primary as relatively problem free, members of VotePA and other watchdog groups point to evidence that components of the machines weren't running on exactly the right software ... and that there's no way for the average voter to tell whether each machine is really clear of votes when it starts up each Election Day morning.
Collin Lynch, a University of Pittsburgh graduate student and VotePA member, holds up the long, skinny printout that state law requires from each machine before the polls open ... the zero count, showing that no vote has been recorded so far that day. Yet as City Paper discovered in visiting polling places throughout the county, poll workers were sometimes unable to print zero counts before voting started. They were able to print zero counts as late as mid-afternoon ... after many voters had used the machines.
Lynch also displays a zero count tape in which the machine claims to know that no votes have been cast or counted on no particular machine.
"You're standing in a dark room, shouting 'Is anybody here?' and somebody shouts back, 'No.'," Lynch says. "We just know that some machine printed this and was happily ... incorrect."
The machines "should bar voting until a zero tape is produced," says Richard King of Squirrel Hill, another advocate for paper voting records.
David Eckhardt, a lecturer in computer science at Carnegie Mellon University and a judge of elections in Mt. Lebanon, points to other tapes generated by the iVotronic after the polls close. They show two different pieces of firmware were in use on May 16 in the county.
The small handheld computing device that allows poll workers to load the proper ballot into each machine, and which also takes the machine's vote tally at the end of the day, is called a Personal Electronic Ballot, or PEB. The PEB in use at one machine in the 14th ward, 40th district polling site ... the Jewish Community Center in Squirrel Hill ... was running firmware 1.05 during the primary election, its tape shows. A machine a few feet away, for district 35, was using firmware version 1.07.
The federal government has OK'd only versions 1.07 and 1.08 for use.
As was also obvious at some polling sites, two different machines were in use side by side on May 16. One allowed those with disabilities to vote. The other was another iVotronic model.
"They were the same technology-wise," assures county spokesperson Kevin Evanto, adding that all 4,700 machines to be used in the fall general election will be the same state-certified model.
Ken Fields, spokesperson for ES&S, manufacturer of the iVotronic, said in a phone message that he could explain whether there was a significant difference between the two machines but was unreachable by press time.
Concludes Eckhardt: "We didn't get the machines we were supposed to and we didn't get the software we were supposed to."
But did we still get the election we were supposed to?
"I know the allegations and I've seen the evidence," says Michael Shamos, the CMU professor in charge of testing and certifying electronic voting machines for the state. "The short answer is that we should put procedures in place to avoid these slip-ups in the future, but they had no effect whatsoever on this election. Some of these allegations are ... minor violations of protocol. We don't want to repeat them. But the fact that they occurred doesn't diminish the integrity of the election."
He isn't concerned about zero counts being produced by machines late in the day. "When you start up these machines and open the polls, they ... produce a zero report, indicating that there are zero votes on all these counters," he says. "The zero report is in the machine's memory, waiting to be printed out. You could print it out five weeks later."
As for the incorrect firmware on the PEBs ... "Did those PEBs work properly or not?" he asks, pointing to the lack of complaints about inappropriate ballots being used anywhere in the county, for one.
Lynch is unconvinced by Shamos' dismissal of the problems as mere "protocol."
"The protocol is actually what is supposed to provide the security of the election," he says. And those zero counts held in the memory of every voting machine: "If you can run one off at any time, there is never any reason to believe it."
Adds activist King: "You have a choice between faith-based elections and auditable elections."
"It's a matter of trust," Lynch concludes. The county, he says, is "simply trusting the machines and by extension anyone who has touched the machines ..."