23 April 2003—The State of California is taking an unprecedented second look at the security and reliability of touch-screen voting machines. On 19 February, California Secretary of State Kevin Shelley appointed a special task force on touch-screen voting to investigate if and how paper ballots should be produced in electronic voting machines for recounting votes. The committee’s recommendations are likely to lead to a new statewide policy on electronic voting, one that could, ultimately, affect elections across the United States.

Such a policy could be a breakthrough for the many computer scientists who have been warning that touch-screen voting machines are not secure. "California can make a big impact on the rest of the nation, if we can get California to tighten things up," says Douglas Jones, a professor of computer science at the University of Iowa and the former chair of the Iowa Board of Examiners for Voting Machines and Electronic voting systems.

But time is against the new task force. "A lot of decisions are going to need to be made by counties in California shortly," says IEEE Fellow David Dill, who is a task force member. "We need to come up with something quickly; otherwise it’s not going to matter very much."

This sense of urgency is shared with many county election officials who have been plunging headlong into electronic voting. Prodded by a federal court order that mandates that certain types of punch-card systems be replaced by March 2004 and tantalized by the promise of federal and state election reform dollars to upgrade, many county election boards are gearing up to buy flashy new touch-screen voting machines. If Dill has any influence on the task force, they will have printers in them.

Secretary of State Shelley was prompted to take action in large part because of Dill’s recent activism. A professor of computer science at Stanford University, Dill wrote a resolution opposing electronic voting, and in less than three months, he has garnered over 940 signatures, some 600 of which are from computer scientists. He first took the resolution to Santa Clara County to persuade the election board to reconsider its decision to buy touch-screen machines that don’t print out a verifiable paper ballot.

"David Dill is…scaring the crap out of the people in Santa Clara," VoteHere president Jim Adler told IEEE Spectrum. VoteHere (Bellevue, Wash.) builds voter verification software, and is, unsurprisingly, "in deep support of what California is doing."

Dill’s resolution scared more than just county officials. "The level of concern in the state government has increased greatly because of our involvement here," he says.

The problem, Dill says in his resolution, is that electronic voting machines not letting voters verify that their vote is being cast correctly put the integrity of elections at great risk. Many electronics experts fear that what is recorded in the machine might not be what the voter intended. A malicious bug or an innocent error in the software could throw an entire election, and worse, since there is no verifiable audit trail, there would be no way to prove that the votes were recorded incorrectly. "We’re talking about basically subverting democracy in this country," says Barbara Simons, past president of the Association for Computing Machinery (ACM) and co-chair of the U.S. Public Policy Committee of ACM.

The solution, according to Simons and most other experts studying electronic voting, is to print out the ballot as the voter votes. The voter can then verify that the vote is correct, and the ballot is automatically deposited in a ballot box while the electronic vote is cast. The paper ballot, not the electronic vote, serves as the final legal ballot in a recount. [See "A Better Ballot Box?" by Rebecca Mercuri, IEEE Spectrum, October 2002, p. 46-50].

Counties may be eager to buy touch-screen voting machines because they are unaware of the technical issues. When Dill brought his concerns to the election board in Santa Clara, the heart of Silicon Valley, board members were surprised that there might be some danger in adopting electronic voting equipment. "I don’t think anyone had really raised the issue before," says Simons who accompanied Dill to meetings with the board. "The official who was presenting the material referred to it as being similar to buying city busses. You go out, you get the money, and you make the purchase."

The county board was not concerned about security because the machines were certified for state elections, says Simons. During the certification process, outside testers evaluate voting systems to ensure that they meet the standards set by the U.S. Federal Election Commission. Additionally, each state follows its own procedures to certify its own voting equipment. The process catches some technical errors, but is not perfect. "We can’t really expect the certification authority to find all of the bugs in these programs," Dill explains.

"In the decade or so that I've been on the Iowa Board of Examiners for Voting Machines and Electronic voting systems, I have seen several certified machines that contained major flaws," says the University of Iowa’s Jones. He has seen machines that "obviously aren’t ready for use" that have been certified by other states. But too often election boards don’t realize that bad voting machines can be certified.

Problems in the certification process have also been highlighted recently by a whistleblower suit filed in Seattle, Washington. [See "Whistleblower Lawsuit Points to Weaknesses in Electronic Voting Technology,"] Former senior test engineer Daniel Spillane filed a lawsuit on 25 February against his onetime employer, VoteHere. He claims that he was fired to keep him from talking to auditors from the Independent Test Authority (ITA) and the United States General Accounting Office about problems he found in voting software. That software was certified despite its weaknesses, says Spillane.

Dill hopes that his resolution will bring these issues to light beyond the State of California. Since electronic voting machines are being adopted across the globe, he intends the resolution to be a statement that scientists worldwide can endorse, showing international scientific consensus. "If you get a whole bunch of computer scientists saying the same thing," he says, "I think it has to be taken more seriously."

David Dill’s resolution on electronic voting can be found at http://verify.stanford.edu/evote.html.