by Bob Fitrakis and Harvey Wasserman
July 27, 2007
We’ve been long warned that foreign policy is like making sausage – you don’t want to know the gory details. The same is true of electronic vote counting on the so-called e-voting machines, or DREs.
Earlier this month, the 2006 election challenge court case Squire v. Geer gave us a rare tour inside the scam known as “touchscreen voting.” The “tour” comes with the backdrop of revelations that 56 of Ohio’s 88 counties have “lost” all or some of the election records from the stolen 2004 election. Though the records were protected by federal law, this means it may now be impossible to definitively tabulate who actually won the presidency in 2004. We will write more on this breaking news story in articles to come.
Carol Squire, a Franklin County Domestic Court Judge, refused to stand down after election observers and a post-election computer forensic analysis documented massive electronic voting machine irregularities in her race for re-election. She first paid for a recount of key precincts and used those results to file an election challenge. Squire’s election challenge sought a new election as the remedy.
Prior to last year’s election, the incumbent Democrat—the only African American female in the Domestic Relations Court her—was under fire from the Republican daily monopoly newspaper Columbus Dispatch as well as the Republican-controlled county court. Yet, most political observers expected her to win, particularly if the Democratic gubernatorial candidate Ted Strickland did well in Franklin County.
Strickland won with 58% of the vote in Franklin County. But Squire’s opponent Chris Geer purportedly won by 13,069 votes out of more than 262,000 cast. The outcome, to say the least, is highly implausible.
Magistrate Joel Sacco presided over the Squire v. Geer case, and while he declined to order a new election, his Findings of Fact in his decision of July 2, 2007 serves as a lasting testament to the illegitimacy of e-voting. The Finding of Facts encompassed the recount as well as evidence presented during the election challenge hearing.
Proving that they are either dyslexic or incompetent (or both), Franklin County Board of Elections (BOE) officials proved incapable of calculating 3% of the absentee vote as recount election law requires. The BOE officials “…counted 2072 absentee votes when they should have counted 2702 votes,” according to the Magistrate’s decision.
Compounding their inability to count, “Board of Elections did not check the public counters prior to this recount to verify that the numbers on these counters corresponded to the numbers on the VVPAT (Voter Verified Paper Audit Trail} and the poll books, poll lists or signature book records.”
This is a pattern repeated from the 2004 presidential election recount. Then-Ohio Secretary of State J. Kenneth Blackwell issued a directive that allowed counties to recount, not based on their certified election results, but simply by running a few voted ballots through a counter and then hand-counting them to see if the numbers match. It defies logic that the BOE officials would recount without checking the numbers on the machine counters.
Repeating another tactic made infamous from 2004 presidential election, BOE workers, contrary to Ohio Secretary of State’s directive, violated principles of transparency. The Magistrate wrote that: Franklin County BOE “…did not in the presence of two election officials of different political parties make the poll books, poll lists or signature poll book records available on December 11, 2006 for visual inspection by the recount witnesses.”
Franklin County BOE officials not only refused to provide the books for inspection as required by law, they then proceeded to tamper with evidence in direct violation of election law. The Magistrate found that BOE workers “reviewed the poll books during the period before the official canvas began. Certain corrections were made to correct errors, voter audits were made, and voters were given credit before the official canvas began. Notations were made in precinct workbooks and poll books by Board of Elections staff. No date was placed by the initials of the person making the notation to indicate when the notations were made.”
In any other democracy, this would be called illegal cooking of the books.
After the books were well-done and well-seasoned, the Magistrate found that:
“For the November 7, 2006 general election, in 721 out of 835 Franklin County precincts, the public count on DREs did not equal the number of signatures in the poll books plus provisional votes cast minus the number of cancelled votes.”
Do the math. Of all DRE precincts in Franklin County, 86.3% had vote totals that failed to match signatures in the poll books. This is one of the most stunning indictments of e-voting in United States history.
Even after illegally withholding and anonymously correcting the book, the Franklin County BOE still “…reported 2824 more votes cast than signatures scanned,” according to the Magistrate’s Findings.
What this suggests is that while election officials had time to come close to an overall match by secretly and illegally “correcting” the books, they didn’t have time to match up the precinct-by-precinct tallies.
In an all too familiar scenario with e-voting machines, massive unexplained and improbable undervotes marred the race. The Magistrate noted that “of the voters in the November 7, 2006 election who voted for Governor by paper ballots, 22.15% did not vote in the Squire/Geer contest. Of those who voted for Governor by DRE or electronic ballot, 33.81% did not vote in the Squire/Geer contest. This represents a difference of 11.66%. In the past three elections this ‘voter drop off’ was reversed in the sense that persons voting absentee dropped off at a higher rate than persons utilizing voting machines or voting in the precincts.”
As in U.S. Senate races in 2006 in Montgomery County, Ohio, and in Sarasota, Florida’s House race that year, statistically impossible high rates of undervotes have gone against the Democratic candidates.
A conservative estimate in the Squire race suggests that 30,649 votes went unrecorded due to machine malfunction. The number is no doubt higher, as the Magistrate points out. DRE machines, since they flash when a voter fails to vote, historically have a lower undervote rate than paper ballots.
A staggering 33.81% — more than a third of all voters — allegedly did not cast a vote for Squire or Geer—but only on the DRE machines. Computer scientists have long documented that undervotes can be easily programmed. They can occur either by accident or design.
In Montgomery County’s 2006 Senate race, 30,000 unexplained undervotes occurred in the U.S. Senate race between Sherrod Brown and incumbent Michael Dewine. The county voted Democratic, implying that the Democrat Brown was hurt most by the undervote. In fact, the last major poll before the 2006 election by the Columbus, Ohio Dispatch had Brown winning the state by 24%. That poll was conducted by participants filling out and mailing in an actual paper ballot. A few days later, on Election Day, Brown only received 12% of the vote in the 2006 election conducted almost entirely on DREs.
The reason why the DRE votes are so unreliable may well be explained by the Magistrate’s finding that, “the voting system certified by the Secretary of State and the USEAC and the voting system utilized by Franklin County on November 7, 2006 varied in the following component versions. … Election Data Manager, Hardware Programming Manager and M100 Optical Scanner.”
The use of uncertified hardware and software installed at the last second can create not only massive undervotes but unexplained vote shifts due to recalibration problems, which is a quaint term for vote flipping. The software changes involved, among other things, ballot definitions. And, with rotating candidates’ names by precinct, so that all have equal chance of appearing first, uncertified changes could cause a vote cast for Squire to show up as a vote cast for Geer.
Squire’s expert witness, Dr. Rebecca Mercuri, found and the Magistrate accepted as fact that “…The use of components that are mismatched — i.e., components that are configured in a way that was not certified by the USEAC — violates certification requirements and also runs the risk of exposure to programming errors or security vulnerabilities that could compromise the integrity of the election and could result in the loss of mistabulation of votes.”
Dr. Mercuri also testified and it was accepted as fact in the Magistrate’s decision that “…Franklin County had not completed a review of the 1.4 million lines of code contained collectively in the software which comprises the modules. Dr. Mercuri concluded that she had ‘very serious concerns about the security’ during the election.”
Computer election expert David Dill of Stanford publicly stated that, “It is practically impossible for someone to review software of any length at all – even 10,000 or even 1500 – lines of code to make sure that it’s 100% error-free.”
Professor Avi Rubin points out that it is easy to hide undetected codes in large code packages.
In 2004, Diebold, ES&S, and Triad technicians just prior to the election and recount, made numerous software and hardware changes to election voting machines, central tabulators and other computer technology without certification. In 2006, there were even more electronic voting machines in use, and the Squire challenge just happened to catch the use of uncertified software and hardware.
While the Magistrate found by clear and convincing evidence that there were election “irregularities,” he failed to order a new election in part because electronic voting creates a classic Catch-22. Because of the non-transparency of the computer black boxes we vote on, you can never really prove with “clear and convincing evidence” who won.
That now seems to be the case for Carol Squire—and for the majority of counties in Ohio’s stolen 2004 election.
Bob Fitrakis and Harvey Wasserman are co-authors of HOW THE GOP STOLE AMERICA’S 2004 ELECTION & IS RIGGING 2008, available at www.freepress.org, where this article was first published.
Original article at,