electionline Weekly – August 17, 2006
I. In Focus This Week
News Analysis: The Coming Paper-Trail Debacle?
Ohio report finds challenges abound in evaluating voter-verified paper audit trails
By Dan Seligson
A 240-page report on failures and foibles during Cuyahoga County’s May primary raised more questions about the accuracy and reliability of touch-screen voting machines which researchers say failed to match up electronic ballots to paper versions of the vote.
Perhaps equally significant – and noteworthy – are the details of the considerable woes that plagued the voter-verified paper audit trail (VVPAT) system through careless election administration, printer failures or both.
Buried some 93 pages into the report, which was commissioned by county leaders and produced by the San Francisco-based Election Science Institute, are details of errors that included poll workers loading thermal paper into VVPAT printers backwards, blank audit trails, “accordion-style” crumpling of ballots, long blank spaces between ballots that could have represented missing or unprinted VVPATs, torn and taped-together VVPATs and missing ballot text.
ESI researchers found that nearly 10 percent of VVPAT ballots sampled were in some way compromised, damaged or otherwise uncountable, an alarmingly high proportion for a state that requires that paper be used as the ballot of record in the event of a recount.
That led ESI to the ominous conclusion that “in the event of a recount or election contest, the risk of legal challenges is exceptionally high if no significant modifications are made to the current election system.”
“The VVPAT is only as reliable as the administration of the system that produces the paper trail,” said Tracy Warren, the ESI researcher who led the manual VVPAT recount.
Warren said she hoped the ESI findings would be “immensely valuable” in helping jurisdictions – and particularly Cuyahoga County – avoid future mishaps in administrating votes using VVPAT systems.
Diebold Election Systems, the company that produces the TSx voting system used in Cuyahoga and widely throughout the rest of Ohio, saw the report differently.
In a letter to the Cuyahoga County Commissioners, Michael Lindroos, Diebold’s vice president and counsel, wrote that the report was “inaccurate and the result of an erroneous and misleading investigation that is clearly false.”
Kenneth Blackwell, Ohio’s secretary of state and Republican nominee for governor, said in press reports that it would similarly be “irresponsible to summarily dismiss any of their findings, and it would be equally irresponsible to sound alarms about the accuracy and effectiveness of this equipment.”
Diebold spokesman David Bear said most of the problems with Cuyahoga’s paper trails were caused by poll worker mistakes, with poor training as the primary culprit.
“Obviously it reflects poorly on the company and the county,” Bear said. “But the main concern is that you lessen the likelihood of that occurring. The things that we can do are to lessen the likelihood of problems with design. The other issue is that you have to beef up training. We work extremely hard with jurisdictions to help them make sure their training is at as high a level as possible.”
Bear said successful elections using the same equipment in other Ohio counties and outside of the state suggest “they’re not too difficult. It’s an issue of familiarity.”
Increasing training will be critical as more states adopt rules requiring paper trails. The latest count by electionline.org found 23 states require the use of VVPATs. Four other states require paper-based systems.
Michael Vu, Cuyahoga County’s embattled election director, said while problems certainly occurred, there are safeguards in place. For example, misprinted or unprinted VVPATs could be reproduced from a machine’s memory. However, he acknowledged a paper version of the electronic record would not qualify necessarily as an independent, voter-verified ballot.
“Certainly there’s the redundancy issue,” he said. “But at least no votes are being lost.”
Vu said he was moving ahead with more training for election-day technicians as well as poll workers in advance of the November vote. And he sounded at least cautiously optimistic that lost, damaged or otherwise uncountable VVPATs would not cause a Florida-esque meltdown with blank thermal paper spools becoming the new hanging chad.
“I think there are a number of ways problems can be resolved,” he said. “One is training, two there is a mechanism confirmed by the secretary of state in place that resolves any situation that we come across, like when a paper trail is put in backwards… It’s not a perfect system, but neither was the punch card or optical scan system.”
To critics, the high percentage of damaged or uncountable VVPATs damaged rated as significantly worse than ‘imperfect.”
“Ten percent is a complete disaster and totally defeats the purpose of a VVPAT,” said David Dill, a Stanford University computer science professor and founder of Verified Voting. “You can blame it on poll worker training, but there are ways to design equipment that makes user error less likely. There are indications that Diebold has done a less than adequate job in design. The company has adopted a generally reluctant and unenthusiastic stance to paper trails and it shows in the design.”
The answer to VVPAT problems, Dill said, would be precinct-counted optical-scan units.
“There are fewer questions about it,” he said. “We know with appropriate care and poll worker training, we can run a good election on optical scan.”
electionline Weekly and electionline.org ALERTS are produced by the staff of electionline.org, a nonpartisan, non-advocacy research effort supported by The Pew Charitable Trusts and administered by the University of Richmond.