How E-Voting Threatens Democracy
March 29, 2004
By Kim Zetter
In January 2003, voting activist Bev Harris was holed up in the basement of her three-story house in Renton, Washington, searching the Internet for an electronic voting machine manual, when she made a startling discovery.
Clicking on a link for a file transfer protocol site belonging to voting machine maker Diebold Election Systems, Harris found about 40,000 unprotected computer files. They included source code for Diebold's AccuVote touch-screen voting machine, program files for its Global Election Management System tabulation software, a Texas voter-registration list with voters' names and addresses, and what appeared to be live vote data from 57 precincts in a 2002 California primary election.
"We weren't concerned about being refuted," Stubblefield said. "We knew the technical accuracy of what we discovered. (Critics) could try to spin things against us, but in the end truth prevails."
It wasn't the first time someone had found problems with Diebold's system. Doug Jones, a computer scientist at the University of Iowa and a member of Iowa's voting system board of examiners, found the same problems in 1997 when his state was considering buying the systems. Jones was particularly disturbed by the same problem that Kohno and Stubblefield found regarding the encryption key that was coded into the system and was the same for every voting machine. He told Diebold about his finding, but a non-disclosure agreement prevented him from going public.
[Editorial gripe: This is not entirely correct, the nondisclosure agreement didn't prevent my public discussion of this flaw, and I certainly did go public in 2001 when I used the Global example before the House Science Committee!]
"I was disappointed to see that the company had done nothing to fix the problems in all of these years," Jones said after reading the Johns Hopkins report. Diebold spokesman Bear said the company fixed the encryption key problem after a second research report came out last September that raised the same concerns raised by Doug Jones and Rubin's group.
"If any of the multitudes of reviewers of our system find any issues we immediately investigate the issues and where appropriate modify the system to address the issues," Bear wrote in an e-mail.
To many e-voting critics, the Rubin report highlighted serious problems with federal certification processes and standards, which they say addressed the functionality of voting systems but not their security.
"If the Diebold system made it through the certification process, then the certification process is really broken," Rubin said. There was no reason to believe that systems made by other vendors were any more secure, he said.
In fact, in a certification report for the Diebold system that Doug Jones read in 1997, an unnamed certifier for Wyle Laboratories called the Diebold system, which was then called the I-Mark Electronic Ballot Station, the best of the lot. "This is the best voting system software we've ever seen," the certifier wrote.
Today, Rubin and other scientists, including Doug Jones, are writing a proposal for a $10 million, five-year National Science Foundation grant to study e-voting issues and design a system. If the funding comes through, the group could have a system built by 2006, Rubin said.
This week, the Open Voting Consortium, an international group of researchers, plans to demonstrate a free, open-source voting software that runs on inexpensive PCs. The group recently posted a demo of its ballot software, which took four years of planning, online.