Preventing a repeat of the 2000 debacle
After the punch-card nightmare, more accurate machines are supposed to be in place, but experts still can't agree on the best system

BY BRYN NELSON
STAFF WRITER

March 2, 2004

Within the past month alone, there have been accusations of half-truths. Questions about credibility and experience. Angry retorts.

No, not between those candidates.

That election-year jostling seems downright demure compared with the fireworks over another issue: how to instill public confidence in a voting system buffeted by criticism after the 2000 presidential election.

The 2002 Help America Vote Act sought to phase out the much-maligned lever machines and punch-card systems in favor of better ballot designs, improved accessibility to voters with disabilities, and a restoration of the belief that every vote matters in an election.

What's emerged, however, is a thicket of finger-pointing, dueling statistics and Election Day horror stories. On one side are a group of researchers and advocates who say the rush to new technology - the ATM-like touch-screen machines in particular - is exposing the election system to fundamental security flaws that allow little redress if vote tampering or software glitches are suspected. But even among themselves, the experts have found little agreement on how best to address such concerns.

On the other side, touch- screen manufacturers like North Canton, Ohio-based Diebold Election Systems have portrayed the security experts criticizing their machines as fear- mongering academics who lack experience with practical Election Day issues.

Satisfying the law, voters

Often left in the middle are election officials keen to replace their aging systems with versions that conform to the new requirements of federal law and meet voter needs.

Diebold, among the largest touch-screen manufacturers in the country, has become a leading vendor in several states holding presidential primaries today, including Maryland and Georgia. In New York State, election officials received a federal waiver to delay the installation of new voting machines, meaning that the lever machines being used today in New York City and Long Island voting precincts likely won't be completely phased out until 2006.

The central question is whether any replacement will meet the demands of security, accessibility and simplicity while still maintaining the anonymity of voters.

The multimillion-dollar answer depends on whom you ask. Each expert and company has a Web site advocating a position: Touch-screen technology is accurate and easy to use, or it's flawed and untrustworthy. Paper ballots are the best current solution to a poor election system, or paper ballots are flawed and untrustworthy.

Questions also have been raised over the touted success of at least one touch-screen system in reducing what's known as the under-vote error, in which a voter's ballot is recorded as having been cast but no choice is ed.

Comparing accuracy

Diebold has boasted an under-vote error rate of less than 1 percent for its AccuVote-TS system, a number quoted in several media reports. A handout e-mailed to Newsday by the company states that the California recall election "proves" that its AccuVote-TS and another Diebold-manufactured optical scan system "beats the competition for accuracy." The claim cites its source an analysis by a Harvard University researcher.

But Harvard research fellow Rebecca Mercuri angrily accused Diebold of misrepresenting her research. Mercuri, who provided Newsday with a copy of the data she gathered with a co-researcher, said Diebold's release was based on only one part of her analysis of the California recall vote. In that part - the yes-no question of whether to recall then-Gov. Gray Davis - the AccuVote-TS system indeed scored the best of 10 optical scan, punch-card and touch-screen systems surveyed, she said, with an under-vote error rate of 0.73 percent.

But for the second question of the election, which asked voters to vote for one of 135 candidates, Diebold's touch-screen under-vote error rate jumped to 9.23 percent - eighth among the systems analyzed, she said.

The Diebold handout provided to Newsday did not include those results. The average error rate for both questions, in fact, left Diebold's system in fifth place overall, behind a system manufactured by touch-screen competitor Sequoia, two optical scan systems, and a punch-card ballot.

In a follow-up e-mail, Mercuri said she was "appalled" that Diebold is "perpetuating a false impression."

Diebold spokesman David Bear, reached as he was about to board a plane, said he believed the allegations were incorrect but referred Newsday to another spokesman. That spokesman did not return several calls seeking comment.

Accountability questions

Mercuri also echoed concerns raised during a symposium held last month in Seattle during the annual conference of the American Association for the Advancement of Science, where five computer scientists blasted current touch-screen systems for their perceived lack of security, accuracy in tallying and accountability in an audit or recount.

"The issue of accountability seems to be a no-brainer," said Peter Neumann, a principal scientist with SRI International's Computer Science Laboratory in Menlo Park, Calif. "Yet we're dealing with machines that have zero accountability."

Bear, in a phone interview last week, said many of the computer security specialists critical of touch-screen technology aren't necessarily familiar with how elections are run.

"Voting doesn't take place in a vacuum," he said, pointing out that checks and balances have been built into the voting process.

Among those concerns, election officials in New York City, Nassau and Suffolk counties and Maryland listed accessibility for those with disabilities, ease of use in multilingual voting districts, and wear and tear from a high volume of voters.

For today's primary, Maryland plans to use touch-screen voting for every jurisdiction except Baltimore County.

"The voters love it. They really do," said Linda Lamone, the state's administrator of elections.

In New York, election officials are waiting for the state to complete its vetting process of voting machines that meet the new federal and state standards. But touch-screen machines have been well received among New York City and Long Island officials.

Eleanor Sciglibaglio, the chief clerk of the Board of Elections for Nassau County, praised the user-friendliness and multilingual adaptability of touch-screen technology.

"It's like an ATM, but instead of money you get an elected official," she said.

Several of the computer scientists meeting in Seattle pointed out that ATMs provide receipts for all transactions to protect against fraud or mistakes, whereas no system is in place to do the same with anonymous voters.

In response, Bear said the technology for such a verification system already exists, but its availability would depend upon demand.

"I think everyone thinks there needs to be verification, but the question is to what degree," said Robert Garfinkle, the Republican commissioner of the Suffolk County Board of Elections. "In the voters' minds, they need to be satisfied that they have a machine that's reliable and will actually cast their votes."

In the absence of universal support for any one voting machine or verification system, however, that goal still faces plenty of obstacles.

Copyright © 2004, Newsday, Inc.