Wrong Time for an E-Vote Glitch

[JackSpratts]

Kim Zetter

It was simultaneously an uh-oh moment and an ah-ha moment.

When Sequoia Voting Systems demonstrated its new paper-trail electronic voting system for state Senate staffers in California last week, the company representative got a surprise when the paper trail failed to record votes that testers cast on the machine.

That was bad news for the voting company, whose paper-trail, touch-screen machine will be used for the first time next month in Nevada's state primary. The company advertises that its touch-screen machines provide "nothing less than 100 percent accuracy."

It was good news, however, for computer scientists and voting activists, who have long held that touch-screen machines are unreliable and vulnerable to tampering, and therefore must provide a physical paper-based audit trail of votes.

"It goes to our point that a paper trail is very much needed to (ensure) that the machine accurately reports what people press," said Susie Swatt, chief of staff for state Sen. Ross Johnson (R-Irvine), who witnessed the glitch in the Sequoia machine.

With a paper-trail system, the voting machines would print out a record when voters cast ballots on a touch- screen machine. Voters could examine, but not touch, the record before casting their ballot. The paper would then drop into a secure ballot box for use in a recount.

For nearly a year, voting companies and many election officials have resisted the call for a paper record. Election officials say that putting printers on voting machines would create problems for poll workers if the printers break down or run out of paper, and the paper records will cause long poll lines with voters taking more time to check the record.

Voting activists maintain, however, that election officials don't want the paper trail because it opens the way for recounts and lawsuits if paper records don't match digital vote tallies. And they say that paper records would provide proof the machines are not as accurate as companies claim.

Acting on public pressure for a paper trail, Sequoia became the first of the four largest voting companies to add printers to their voting machines earlier this year. Two smaller voting companies have had paper-trail machines for longer, but have had trouble selling the machines to election officials.

During the demonstration of the Sequoia machine last week, the machine worked fine when the company tested votes using an English-language ballot. But when the testers switched to a Spanish-language ballot, the paper trail showed no votes cast for two propositions.

"We did it again and the same thing happened," said Darren Chesin, a consultant to the state Senate elections and reapportionment committee. "The problem was not with the paper trail. The paper trail worked flawlessly, but it caught a mistake in the programming of the touch-screen machine itself. For some reason it would not record or display the votes on the Spanish ballot for these two ballot measures. The only reason we even caught it was because we were looking at the paper trail to verify it."

Sequoia spokesman Alfie Charles said the problem was not a programming error but a ballot-design error.

"It was our fault for not proofing the Spanish language ballot before demonstrating it," Charles said. "We had a demo ballot that we designed in a hurry that didn't include all of the files that we needed to have the machine present all of the voter's selections on the screen and the printed ballots. That would never happen in an election environment because of all the proofing that election officials do."

Charles said the machine did record the votes accurately in its memory, but failed to record them on the paper trail and on the review screen that voters examine before casting their ballot. Swatt and Chesin could not confirm this, however, because the company did not show them evidence of the digital votes stored on the machine's internal memory.

"We've been saying all along that these things are subject to glitches," Chesin said. "The bottom line is that the paper trail caught the mistake. Ergo, paper trails are a good idea."

Charles agreed the paper trail worked exactly as it was supposed to work. "If this happened in an election, the first voter would see it and could call a pollworker. They would take the machine out of service if they saw a problem," he said.

Ironically, just one week after the demonstration occurred, California took one step back from making sure voters in the state will have the reassurance that a paper trail provides.

On Thursday, a Senate bill that would require a voter-verified paper trail on all electronic voting machines in the state by January 2006 suffered a setback when the Assembly Appropriations Committee, where the bill resided, decided not to push the bill forward during this legislative session, which ends Aug. 31. This means legislators will have to reintroduce a new bill next January when they reconvene.

The bill (PDF), introduced by Johnson and state Senator Don Perata (D-Oakland), had bipartisan support and the backing of Secretary of State Kevin Shelley.

"I'm a little mystified why the committee has stalled the bill," Swatt said. "E-voting machines, like them or not, are here to stay in California. It is clear that if we are going to be living with e-voting machines the only way to protect voters and to ensure that their votes are counted accurately is to have a paper trail."

Swatt said she hoped the public would pressure the legislature to push the bill forward before the session ends.

http://www.wired.com/news/evote/0,2645,64569,00.html

[tambourine-man]

Originally Posted by JackSpratts

During the demonstration of the Sequoia machine last week, the machine worked fine when the company tested votes using an English-language ballot. But when the testers switched to a Spanish-language ballot, the paper trail showed no votes cast for two propositions.

Sounds like it worked perfectly.

And the question on every politician's lips: "could it be modified for Ebonics?"

[jcat]

haha ditto, sounds like it works perfectly,

[Nicobie]

Man, I wish I could repeat other peoples words like they do here.

But I guess not, as I don't have a liberal arts degree paid for by the gov.......

shehe



First class required = cut&paste the professor's pc correct drivel.

[tambourine-man]

Originally Posted by Nicobie

Man, I wish I could repeat other peoples words like they do here.
But I guess not, as I don't have a liberal arts degree paid for by the gov.......
shehe
First class required = cut&paste the professor's pc correct drivel.

Maybe you could use your education to post three word replies and insults. Or maybe you could do that nicobie classic of quoting yourself and carrying on your own conversation? Or maybe you could put your considerable intellect into action by cluttering up a forum with inane posts and use clever words like 'ghey' and 'ass'.

Yes. Do that... Do that - and put us to shame with your remarkable originality.

[thinker]

I don't like huge c&p's either, but you need a subscription to read this, and I don't have a communal one.

Lost Votes in N.M. a Cautionary Tale
As Election Day Nears, a Look at Problems in 2000 Shows Fallibility of Machines

By Dan Keating
Washington Post Staff Writer
Sunday, August 22, 2004; Page A05

ESPAÑOLA, N.M. -- Four years ago, about 2,300 voters traveled the winding roads through this remote county to cast their ballots before Election Day on state-of-the-art, push-button electronic voting machines. For 678 of them, their votes were never recorded.

Vice President Al Gore won this state by 366 votes. Even if the missing votes had gone for George W. Bush and given him New Mexico's five electoral votes, it would not have changed the outcome of the presidential race.
But the missing votes in Rio Arriba County show that even in the finest electronic voting systems -- New Mexico holds itself out as a leader after a decade of experience -- serious miscounts that could sway elections can occur if the computerized machines are not correctly programmed.

But the missing votes in Rio Arriba County show that even in the finest electronic voting systems -- New Mexico holds itself out as a leader after a decade of experience -- serious miscounts that could sway elections can occur if the computerized machines are not correctly programmed.

With many states making moves to electronic voting machines this year, critics of the new technology say it is fraught with the potential for fraud. But what happened in Rio Arriba County shows what some computer experts say is a far more pressing concern: mistakes in computer programming by inexperienced local election staffs.

The Washington Post examined the voting results here because New Mexico had the narrowest winning margin in the presidential contest, and Rio Arriba County had the largest percentage of voters who had no presidential vote. The review discovered that 203 voters turned out in one of Rio Arriba's voting districts, but the state's certified results show "0" votes were recorded for Gore or Bush. The same was true for the U.S. Senate and House candidates. In another district, two-thirds of those who voted in the month before Election Day -- early voting is allowed in New Mexico -- had no votes recorded in any races. Steve Fresquez, a state computer technician who oversaw vote counts for Rio Arriba County, said the electronic machines had been programmed incorrectly for early voters, but it was not discovered until days after the election.

"It was such a mess, but there was nothing we could do about it because it was over. It was too late. The election had already gone through," Fresquez said. When it came to reporting the results, "we allowed the county to do the best they could and, as you can see, it wasn't too good."

In the months after the disputed 2000 presidential vote in Florida, which was marred by "hanging chads" and other problems with paper ballots, advocates of electronic voting machines said computerized systems would end concerns about the accuracy of ballots.

A number of states, including Maryland and Georgia, have moved to such systems, spending tens of millions of dollars.

Critics have said the machines are not perfect and are subject to deliberate tampering, but the experience in Rio Arriba County shows that simple, benign mistakes in programming can lead to results being wildly off.

Mistakes are likely to arise when thousands of small counties nationwide program ballots for multiple districts with dozens of races in each election, said Steve Ansolabehere, a political scientist at the Massachusetts Institute of Technology who is participating in the CalTech-MIT Voting Technology Project. "That is the Number 1 problem with electronic voting: the programming for each election," he said. "These offices in rural areas do not have the staff with the kind of technical expertise necessary to do electronic voting."

The need for better training of local workers and volunteers is one point on which supporters and opponents of electronic voting agree. Several states, the federal government and think tanks all say that undertrained workers are the weakest link.

"You can spend all the money you want to spend on technology and you're still not going to get better elections," said William F. Welsh, board member and former chairman of Election Systems & Software, one of the industry's biggest companies.

About 90 percent of New Mexico's voters cast their ballots electronically. Rio Arriba County sits on the state's northern border and features a mile-high valley between two plateaus, with purple mountains in the distance. It has a population of just more than 40,000 spread across an area almost half the size of Maryland. It takes 2 1/2 hours to drive from one end of the county to another -- from some pueblos, driving to a polling place requires following a road into Colorado and back.

Because of those distances, County Clerk Fred Vigil encourages voting during the month before the election on the push-button electronic voting equipment used here for a decade.

Neither Vigil nor state elections director Denise Lamb remembered problems in Rio Arriba when asked about them for The Post's review. They referred questions to Fresquez, who said he remembered the problem well.

Rio Arriba County has three voting districts -- the candidates for state legislature in each are different -- but for early voters the county used just one ballot listing the names of all the candidates.

"There was no way we could get the correct votes because that was how they programmed the machine," Fresquez said.

Fresquez said the county had only two early-voting locations. Rather than programming separate machines at each location for the county's different voting districts, Rio Arriba tried to program one machine to cover all the districts. "They were trying to use less machines," he said. "They thought they could put it all on one ballot. They were not aware of" any problem.

Still, he and Lamb said they thought the error did not mean votes were really lost. Rather, they said it was likely the votes in one or two districts were credited to the totals of another district.

That outcome does not appear to square with tallies from the county's three election districts. In one district, none of the 203 ballots cast were recorded for Bush or Gore. In another, 188 of the 569 voters cast a presidential vote. The third district had a more typical pattern, with 1,500 of the 1,594 voters recording a presidential choice.

New Mexico is the only state to have an elaborate, three-step audit process of voting results. Precinct results are checked by the county and state and then by a certified public accounting firm. The federal Election Assistance Commission, established after the 2000 Florida recount to help states establish new voting systems, has cited the audit as a "best practice" to be used elsewhere.

Lamb testified to the commission that the "triple audit" would alert the state to problems with the electronic voting machines. Fresquez's work on Rio Arriba's results did uncover the programming error. But it was never publicized.

In fact, the audit could show only that the programming error occurred. There was no way to recount the missed votes. They were simply gone.

Mistakes with new computer technology leave election officials with no recourse, said electronic voting critic Avi Ruben of Johns Hopkins University.

The outcome of a close presidential election could hinge on votes that cannot be reconstructed. "What are we going to do?" he asked. "Do we throw our hands up on a national scale and say 'We messed up'?"