http://www

http://www.post-gazette.com/pg/06344/745015-85.stm

Pittsburgh Post-Gazette

Are you nuts if you think the machine flipped your vote on Nov. 7?

Though solid evidence is hard to pin down, complaints abound about voting machines

Sunday, December 10, 2006

By Bill Toland, Pittsburgh Post-Gazette

Darrell Sapp, Post-Gazette

The screen on one of Allegheny

County's new voting machines.

It was her first time using Allegheny County's new touch-screen voting machines, but, even so, Carlana Rhoten is positive she voted for Democrat Bob Casey on Nov. 7. That's why it was so surprising -- no, infuriating -- that the voting machine told her she'd voted for his Republican opponent incumbent Sen. Rick Santorum.

For those who chose not to participate in this most recent episode of America's great democratic experiment, here's how it worked: The ES&S iVotronic machine pages through the ballot, asking voters to push the area on the screen corresponding to the candidate of their choice. Once the voter has completed the ballot, the machine offers a summary page, which reviews the selections, asking for confirmation. If all looks right, the user hits the blinking "vote" button.

But the review page told Ms. Rhoten, 63, she'd voted for the senator, not the treasurer, she said.

"There's no way I would accidentally touch the screen for Santorum," said Ms. Rhoten, who has been a registered Democrat for 25 years. She summoned a poll worker, who suggested Ms. Rhoten delete her votes and start over. The second time, she said, the machine recorded her vote correctly.

Ms. Rhoten's afternoon experience at Garfield's Rogers Middle School is one shared by untold numbers of voters across the country. The phenomenon is called vote-flipping, or vote-hopping. You think you've voted for one candidate, but the machine swears you've voted for another. And as electronic voting machines become more common because of the Help America Vote Act, reports of vote-flipping have begun to accumulate.

Heber Springs, Ark., mayoral candidate Jackie McPherson reportedly tried to vote for herself on a test ballot, but the voting machine told her she'd voted for her opponent.

A losing candidate for Pinellas Park, Fla., council said he tried to vote for himself, but the machine registered a vote for his opponent. The Pinellas County Supervisor of Elections Office confirmed to St. Petersburg Times that the machine in question was malfunctioning.

In Texas, voters said they tried to cast straight Democratic ballots, but the touch-screen machines told them they'd voted the straight Republican ticket.

On Election Day, Pennsylvania's GOP complained that touch-screen voting machines in several counties were flipping would-be Republican votes to Democratic candidates. Pennsylvania Secretary of State Pedro Cortes, an appointee of Democratic Gov. Ed Rendell, said the GOP complaints were unfounded, but Marjorie Jones thinks otherwise.

Ms. Jones, 79, who voted at an Ohio Township fire hall, said that, when she tried to vote for Rick Santorum, the machine signaled that she'd voted for Mr. Casey. Like Ms. Rhoten, she summoned a poll worker, who instructed her to delete her votes. She voted again. "I thought everything was all right," but then the review screen told her that she'd voted for the sitting governor, when she'd intended to vote for Republican challenger Lynn Swann.

"In my mind, it was a computer mix-up," said Ms. Jones, who turns 80 this month. "I couldn't have voted for the wrong person twice. ... Knowing my age, people will say, 'She's just a mixed-up old lady.' Well, it's not true. I know who I voted for, and I know I did it right."

So what's at play here?

Is it voter error? Is there something wrong with the hardware? Is it an occasional bug in the software programming? Or is it something more sinister?

"I know about the history of dead people voting," Ms. Rhoten said, recounting historical instances of Election Day funny business. Her first thought, after seeing Rick Santorum's name in place of Bob Casey's, was that "the machine has been programmed to push the vote in a different direction." In other words, the politicians were up to no good.

You can't blame voters for being suspicious, given all we've heard about the possibility of voting machines being hacked. Princeton University professors made the rounds in September, telling anyone who would listen that "there's really no limit to the amount of mischief that can be done" by an able hacker armed with a computer virus, illicit political motivations and free access to a voting machine.

But there's no evidence to support that this ever has taken place, said Michael Shamos, a Carnegie Mellon University computer science professor who tests voting machines for Pennsylvania and other states. The vote-flipping phenomenon, to the degree that it exists outside the voter's imagination, is caused by faulty hardware, he said.

"There's no other credible explanation," he said. "This idea that machines are suddenly going to start swapping votes, it's a pipe dream." The problem, he said, is caused by calibration errors; you're touching the screen in one spot, but the hardware believes your finger is poking it elsewhere. It reads the coordinates of your fingers incorrectly and compounds the error by matching those mistaken coordinates with the wrong candidate.

The voting machine touch-screen is a complicated thing. There's a layer of see-through polyester film, a resistive circuit layer, a tin-oxide conductive layer, another circuit layer, an insulating layer, then the glass monitor surface. Pushing on the top film forces the upper circuit layer to touch the lower circuit layer, producing a current switch in the conductive membrane.

The machine measures the voltage and, in so doing, pinpoints the coordinates where the circuit layers met. Those coordinates must be matched with spots on the monitor where candidates' names are displayed.

If just a small percentage of the thousands of voting machines that are deployed on Election Day are miscalibrated, vote-flipping can result. Mr. Shamos blames the relatively low quality of the touch-screen hardware used in voting machines. Better quality touch-screens, with low performance degradation over time, is obviously more expensive. (Typically, "resistive" technology performs worse than "inductive" sensory technology, the kind that's used when your stylus pen makes an impression on your laptop, but for now, inductive technology is too expensive for voting machines.)

"There truly is nothing sinister about this. It's bad. It's a scandal. But it's not deliberate," Mr. Shamos said, adding: "It sure could look that way to a voter."

As positive as Mr. Shamos is that the hardware is to blame, Daniel Sleator is just as uncertain. Mr. Sleator, also a computer science professor at Carnegie Mellon, said touch-screen voting machines, like most computers, can behave unpredictably at times. He said it was "unlikely, but not impossible," that a software bug could cause a machine to tabulate votes incorrectly.

In Texas, for instance, where some voters reported that their Democratic ticket vote was flipped to a straight Republican ticket, the county clerk said her office checked the calibration of the touch-screen machines and found no problems. That means the vote-flip, if it happened, was caused by something else.

Mr. Sleator challenged Mr. Shamos' steadfastness on the issue of hardware vs. software.

"Michael Shamos is a very persuasive guy, ... but he's really one of a very, very small minority of computer professionals who are of the opinion that we should be happy to get away from paper [ballots]. Lots of us in the computer field are kind of appalled that this is being considered."

No matter the cause, it's not the reported instances of vote-flipping that are the most worrisome, but the unreported instances. When a voter, such as Ms. Rhoten, notices the mistake, the mistake can be corrected. But if a voter doesn't carefully inspect the summary screen at the end of the computerized ballot, incorrect votes can go unnoticed.

For their part, the makers of the voting machines believe instances of vote-flipping are usually instigated by user error; the voter thinks he voted for one candidate, but simply made a mistake. Or, perhaps, he was voting with one hand, but the other hand inadvertently grazed the screen. A spokesman for Diebold Election Systems told Computer Word magazine that vote-flipping was "not a problem. ... It doesn't exist. This, again, falls into the 'what if" scenario." Sequoia Voting Systems said concern about vote-flipping is "conspiracy theory from activists and bloggers."

The opinion is more or less reinforced by the Voting Technology Project being conducted by the California Institute of Technology and MIT. Their experiments have found that voters incorrectly chose a candidate on their ballots one in 30 times.

But Heather Heidelbaugh, a GOP attorney, isn't buying it. Statewide, Republicans fielded dozens of calls from Republicans who said the machines tried to switch their votes to Democratic candidates. And while she received about 70 complaints on the matter, she suspects there were thousands more."I'm convinced that it has nothing to do with voter error," she said. "And we have no way of determining who didn't catch it."

(Bill Toland can be reached at btoland@post-gazette.com or 1-412-263-1889. )