University of Iowa News Release
Jan. 18, 2005
Jones Finds Voting Machine Defects In Arizona
Douglas Jones, associate professor of computer science in the University of Iowa College of Liberal Arts and Sciences, has released a report detailing the results of his recent investigation of voting machine problems in Maricopa County (Phoenix) Arizona. The report raises questions about the quality of voting technology oversight -- not just in Arizona -- but nationwide.
In particular, Jones found that the marks recorded on a ballot can be ambiguous, as interpreted by a machine. "This work brings into question the quality of state and federal oversight of our voting technology. The current system of Voluntary Federal Voting System Standards does not address these issues, and the current federal recommendations for best practices in pre-election testing ignore them," he said.
Jones suggested that only by examining how actual voters mark their ballots can we determine how the thresholds on the voting machines ought to be set. He also noted that when a recount differs significantly from the first count, only direct human examination of the ballots can determine whether the machinery was out of adjustment or whether ballots might have been altered.
The Arizona study was initiated by an invitation from Arizona State Sen. Jack Harper. On Dec. 20, 2005, Jones performed extensive testing on the county's eight Optech 4C vote-tabulating machines. The machines are used to scan absentee ballots, which make up approximately half of all ballots cast in a typical Maricopa County election.
Jones' examination was prompted by the unexpected appearance of a number of new votes during a recount of a race in the 2004 Republican primary. While his work does not conclusively explain the results of that recount, it does point to serious defects in the regulation of optical mark-sense scanners currently in use.
The problems Jones identified hinge on the question of what marks count as votes. Under current voting system standards, optical mark-sense scanners are tested only with ballots marked in the way the vendor prescribes. In contrast, real voters make a number of marks that vary from the exact prescription, both in the shape of their marks and in the pen or pencil used.
Unfortunately, federal, state and county testing of optical mark-sense scanners doesn't adequately test the exact setting of the mark-sensing threshold. These problems are not confined to Arizona or to the Optech line of vote tabulators used in Maricopa County.
The instructions on the Maricopa County ballot said, "Connect the arrow with a single line", and county election officials had publicly recommended that voters use Bic ballpoint pens. Unfortunately, when the machines were tested, two of them had their mark-sensing threshold set to a level that could not reliably sense a single stroke of a Bic pen.
"Had the marking instructions demanded a 'dark mark' instead of 'a single line,' this might not have been a problem," Jones said. "Had the county officials recommended using blue or black rolling ball pens, this might not have been a problem. Had the county tested the voting machines with typical pens that real voters use to mark their absentee ballots, they might have taken the two problem machines out of service."
Jones's complete report is available on the Web at http://www.cs.uiowa.edu/~jones/voting/ArizonaDist20.pdf
Jones, the former chair of the Iowa Board of Examiners for Voting Machines and Electronic Voting Systems on which he served for a decade, is widely known for his work on electronic voting security. In November 2005, he served as an election observer in Kazakhstan at the invitation of the Office for Democratic Institutions and Human Rights of the Organization for Security and Cooperation in Europe. He was one of three technical experts to assess Kazakhstan's locally developed electronic voting system.
In August 2005, he was awarded a five-year, $800,000 National Science Foundation (NSF) grant to investigate the use of electronic voting systems in U.S. elections. The grant is part of a $7.5 million NSF project called ACCURATE (A Center for Correct, Usable, Reliable, Auditable, and Transparent Elections) that includes researchers from the UI, Stanford University, the University of California at Berkeley, SRI International, Rice University and Johns Hopkins University. The center, under the leadership of Aviel Rubin of Johns Hopkins University, will coordinate the work of 10 of the nation's leading experts in electronic voting, computer security, public policy issues relating to the use of computers and human-computer interaction.
STORY SOURCE: University of Iowa News Services, 300 Plaza Centre One, Suite 371, Iowa City, Iowa 52242-2500.