By Lowell Finley, Deputy Secretary of State of California, Voting Systems Technology and Policy
February 09, 2007
In January 2007, I was appointed by newly elected California Secretary of State Debra Bowen to the position of Deputy Secretary of State, Voting Systems Technology and Policy. One of Secretary Bowen’s first orders of business is a top to bottom review of the electronic voting systems currently certified for use in California. Secretary Bowen, like many concerned voters, voting system security experts and a growing number of elections officials nationwide, has serious concerns about the security, accuracy and reliability of our electronic voting systems and the method by which those systems are reviewed, tested and certified at the national and state levels.
Today, I am here to testify primarily about my knowledge of electronic voting problems in the Florida 13th congressional district election on November 7, 2006. Before addressing that issue, it may be useful to review briefly the history of my involvement with electronic voting issues.
Before my appointment, I spent over 21 years in private law practice in California, specializing in election law. In 2003, I represented election integrity advocates in a California False Claims Act lawsuit against Diebold Election Systems, Inc., filed on behalf of the State of California and the County of Alameda, California. The complaint in that case alleged that Diebold made false representations concerning its use of “state of the art” security features in its AccuVote TS touch screen electronic voting system to secure a $12 million contract with Alameda County, paid for primarily with state bond funds. After negotiations with the California Attorney General, Diebold settled the case for $2.6 million in 2004.
In 2005, I became Co-Director of Voter Action, a nonprofit, nonpartisan election integrity organization. I represented New Mexico voters in a state constitutional and statutory challenge to the use of electronic voting machines, based on evidence of lost votes and security vulnerabilities in past elections. Shortly after plaintiffs secured a stipulation from the Secretary of State not to purchase new electronic voting machines, New Mexico’s governor and state legislature banned electronic voting machines and converted the entire state to optically scanned paper ballots, with ballot marking assistive devices for voters with disabilities.
In 2006, Voter Action, working with local pro bono attorneys from prominent firms, filed lawsuits on behalf of voters in Arizona, California, Colorado, Ohio and Pennsylvania, challenging the certification or use of electronic voting machines on state constitutional and statutory grounds. In one of those cases, a Colorado judge found that Colorado’s chief elections official failed to promulgate regulations establishing security standards for electronic voting systems as mandated by statute. The judge ordered adoption of security standards, followed by new certification testing of all of the four electronic voting systems used in the state.
The last case I was involved in before assuming my present position was in Florida, a case concerning the November 7, 2006, election for Congressional District 13. The election was extremely close, with a margin of just 369 votes out of 238,249 votes recorded, or 50.1 percent to 49.9 percent.
Sarasota County lies entirely within the 13th congressional district and accounted for 124,119 votes, a majority of all votes cast district-wide. Sarasota County employed two methods of balloting in the election. Touch screen electronic voting machines were used by all voters on Election Day and in early voting, which began 15 days earlier. Absentee voters used paper ballots that were tabulated by optical scanners at the county elections office.
Even before Election Day, a major irregularity emerged in the 13th district balloting that affected only Sarasota County and only ballots cast on the electronic voting machines. From the first day of early voting through Election Day, numerous voters notified elections officials and the press that they were having difficulty voting in the 13th district race. The tally of the electronic ballots cast in Sarasota County revealed 17,825 ”undervotes” in the 13th congressional district race, an extremely high, 14 percent of all ballots cast on the electronic voting machines used on Election Day and in early voting.
An “undervote” occurs when a ballot containing votes in some contests shows no vote in a particular contest. According to experts in the field, a normal undervote rate for a comparable, top of the ticket race would be about 2.5 percent, or in this case approximately 3,000 ballots showing no vote for congressional district race. According to the testimony of experts called by both sides in the lawsuit, the remaining 11.5 percent of ballots, representing 14,800 undervotes, were not the result of voters intentionally skipping over the race without voting. Stated differently, 11.5 percent, or more than one in ten of all ballots cast on the electronic voting machines, did not accurately reflect the will of the voters in the 13th congressional district race.
By way of comparison, the 13th congressional district undervote rate in Sarasota County absentee voting, in which optical scan paper ballots were used, was a low 2.5 percent. The undervote rates on paper ballots used in other counties in the district were also well under 2.5 percent. Finally, the undervote rates on the races that appeared just before and after the 13th congressional district race on the Sarasota County touch screen machines were even lower: 1.2 percent in the United States Senate race and 1.4 percent in the race for governor. Strikingly, fewer Sarasota County touch screen voters had a vote recorded in the 13th congressional district race than in the race for hospital board, the type of “down-ticket” race in which higher undervote rates are commonplace.
The extremely high rate of undervotes on the electronic voting machines corresponds with widespread accounts by voters of difficulty registering their choice of candidates for the 13th congressional district on the machines. Many reported that the touch screen initially indicated that their choice registered but then showed no vote in the race when they came to the review screen.
Based on these shocking figures and on the first-person accounts of numerous voters, I joined with attorneys from three other nonprofit organizations to file an election contest lawsuit in Florida state court on behalf of Republican, Democratic and unaffiliated voters, challenging the validity of the election. I am no longer serving as counsel in the case but follow its progress closely. This nonpartisan lawsuit seeks a re-vote, not a judgment that a different candidate won the election. In fact, the plaintiffs take no position on who won the election. The plaintiffs do contend, however, that the electronic voting machines failed one in ten Sarasota County voters, who intended to vote in the 13th congressional district race but had no vote recorded, and that all voters who voted on the machines should be allowed to vote again, on paper, to ensure against a repetition.
Several theories have been advanced concerning what caused the high undervote. One theory is that a large number of voters abstained from voting for any candidate in the race due to disgust with negative campaigning. This seems highly unlikely, in light of the normal, low undervote rates of voters in the same county who voted absentee on paper ballots and of voters in other counties in the district who did not use these machines.
A second theory to explain the high undervote is that bad ballot design and layout confused voters and caused them to overlook the race entirely. This theory fails to explain the experience of many voters who reported that they had no trouble locating the race on the screen but found it difficult to get the machines to register their choices in the congressional race. These voter complaints began during early voting, before anyone knew how close the election was or who would be the official winner.
The third and final theory as to why Sarasota County voters had such a high undervote rate only when they used the touch screen voting machines is that the software or hardware in the machines malfunctioned. The voting experiences described by many voters point in this direction. Whether machine malfunction explains some or most of the excessive undervotes cannot be determined, however, without independent examination of the source code for the software that runs the touch screen machines used in the election and without hands-on testing of the machines. This is a point that Professor Wallach, an expert in computer security and voting systems, will probably address in his testimony.
Thus far, the manufacturer of the voting machines, ES&S, has refused the plaintiffs’ attempts to gain access to the source code or the voting machines, and the state trial court judge has refused to order it. The manufacturer defends withholding this critical information based on two claims: First, that this information is a trade secret, the disclosure of which would violate the company’s intellectual property rights, and second, that secrecy is necessary to protecting the security of elections. In my opinion, neither of these arguments is valid. Keeping the inner workings of our voting machines secret is inimical to democracy and to the transparency that is necessary for voters to trust that our elections are fair and accurate. And computer security experts have long ago abandoned as bankrupt the concept that we can achieve “security through obscurity” when it comes to computers and computerized devices.
Before leaving the subject of the Sarasota County election, I would like to draw attention to some little noticed but important testimony in the 13th congressional district lawsuit by Dartmouth Professor, Michael Herron, the expert witness who testified for the touch screen manufacturer. Professor Herron acknowledged that overall the 13th congressional district votes recorded in Sarasota County ran 53-47 percent against the official winner. He also testified that it was reasonable to assume that the same split applied to the 14,000 voters who cast ballots and intended to vote in the congressional race but did not succeed in having a vote recorded. Therefore, Professor Herron testified, had it not been for the high number of false “undervotes,” a phenomenon that affected only the Sarasota County portion of the 13th congressional district, the outcome of the election, in which only 369 votes separated the candidates in the official tally, would likely have been different. Thus, whether the extremely high undervote rate in Sarasota County was caused by machine malfunction, as the nonpartisan voter plaintiffs contend, or by bad touch screen ballot layout, as the manufacturer’s expert contends, the use of these state-certified electronic voting machines produced false results that may well have changed the outcome of a congressional election.
It is appropriate to ask whether the serious problems that plagued this election could have been avoided while still using the electronic voting machines. One proposed solution, the addition of printers to the electronic voting machines that would produce “voter verifiable paper audit trails,” falls far short. It would not have prevented the high undervote rate, nor would it have provided evidence of the cause of the high undervote rate. By its very nature, a vote that a machine never records will not be reflected on a paper audit trail. In addition, studies show that a high percentage of voters never review the paper trail. The addition of voter verifiable paper audit trails would not have made a meaningful recount or audit any more feasible.
Getting it right when it comes to the way we conduct our elections is not and should not be a partisan issue. On February 1, 2007, Florida Governor Charlie Crist, a Republican, and Florida Congressman Robert Wexler, a Democrat, joined forces to advocate moving away from use of electronic voting machines in Florida elections in favor of readily recountable and auditable paper ballot optical scan voting systems. I am grateful for this opportunity to testify before the committee and will be happy to answer any questions. I believe I speak for Secretary of State Bowen and the voters of California in expressing deep appreciation for this committee’s interest in ensuring that when voters cast their ballots, they can do so with confidence that every vote will be recorded and counted as they intended.