No, there is not a 68 percent error rate in Dominion voting machines.
No, there is not a 68 percent error rate in Dominion voting machines.
A ntrim County in Michigan was the kernel of the “rigged election” narrative: It was there that the county clerk, an earnest Republican woman, copped to a mistake that, for a short time, caused 6,000 Trump votes to be counted as Biden votes.
It was a good-faith error. It was, inevitably, corrected. And there are paper receipts for all votes. Even if the machine tabulation is thrown off by a formatting problem, the paper doesn’t lie, and there are multiple layers of bipartisan review. When Biden appeared to be ahead in this solidly Republican county, that raised eyebrows, the ballots were checked, and the problem was found and quickly solved. Ironically, the rigging narrative arises out of a county President Trump actually won by about 3,700 votes, in a state he lost to President-elect Biden by 40 times that amount.
No matter. The Antrim vignette became the foundation for fantastical claims by the president and his never-concede supporters that the software in Dominion voting machines, if they could just get a look at it, would be revealed as the key to a massive . . . well, it’s here that the story gets confusing.
Is it a massive, highly sophisticated election-rigging conspiracy orchestrated by Democrats, abetted by overseas Communists? Or is it an error-ridden testament to sheer incompetence that predictably got the result plain wrong? The theory varies by the day. But either way, Antrim remains the Big Bang of the stolen election theory. That, along with the happenstance that Dominion voting systems are used in much of Michigan, as well as several other states — i.e., used in many precincts and states that Trump won . . . which, evidently, is how the cagey commies can maintain plausible deniability.
There is symmetry, then, in that Antrim is central to the latest fraud claim: namely, that a long sought “forensic imaging” investigation has shown that the software used in Dominion voting machines has a 68 percent error rate!
The claim is false. President Trump retweeted it all the same, notwithstanding that, for the source of the claim, Russell Ramsland Jr., a personal 68 percent error rate might be something to strive for at this point.
The error rate claim was debunked by Christopher Krebs, the former head of CISA (the Cybersecurity and Infrastructure Security Agency at DHS), in Senate testimony this week. Krebs, of course, is the official the president fired on November 17, this despite his previously touting CISA’s success under Krebs in hardening our election systems against foreign interference. Krebs had publicly rejected allegations by Trump supporters that votes for the president had been deleted, lost, or altered — including the loopy “Hammer and Scorecard” cyber-fraud theory that some Trump diehards briefly latched on to before embracing the Dominion narrative.
Not surprisingly in light of his performance to date, Ramsland’s report regarding Dominion contains various errors. Putting all of that aside for a moment, Krebs explained to the Senate panel that even Ramsland does not claim that Dominion voting machine software had a 68 percent error rate. This figure, in an apples-to-oranges manner, refers not to an error rate but to an alert rate, and not to ballots but to the election-management system’s tabulation logs.
The high number of alerts (i.e., things that ought to be checked, not necessarily things that are wrong) appears to have stemmed from the single, afore-described formatting error. As County Clerk Sheryl Guy has repeatedly explained, this was a matter of human error, not an inherent problem in the software. Specifically, some candidates were left off the ballot in some local races. To make the vote-tabulation process operate properly, the county clerk’s office needed to update the ballot configuration. Guy’s staff made these changes in some of the relevant precincts, but not all of them. This led, initially, to the inaccurate tabulation that was detected and corrected.
As Guy told the Detroit Free Press, “The equipment is great. . . . It’s just that we didn’t know what we needed to do” to update the ballot information properly.
To the contrary, Ramsland’s report claims that “the Dominion Voting System is intentionally and purposefully designed with inherent errors to create systemic fraud and influence election results.”
Ramsland is a former (unsuccessful) Republican congressional candidate who resides in Dallas. He advertises himself as a Harvard-educated executive manager of a security-services outfit called Allied Security Operations Group, LLC. The company “has a particular emphasis on cyber security,” for which “we employ a wide variety of cyber and cyber forensic analysts.” His educational background is not in elections or cyber: The Harvard degree is an MBA from the business school in 1979, after he graduated from Duke as a political-science major.
If you have been following the Trump campaign challenges in various states, you may remember him as the “expert” who submitted an affidavit alleging that 19 precincts in Michigan recorded more votes — in one instance, he insisted, as many as 350 percent more votes — than the number of actual voters living in the precincts. There was just one problem with Ramsland’s findings: All 19 of the precincts he cited are actually in Minnesota, not Michigan.
In another section of the same affidavit, Ramsland claimed that records from 25 of 47 precincts in Wayne County (where Detroit is located) indicated, implausibly, that there had been 100 percent turnout. In fact, those precincts were also located in Minnesota, not Wayne County, Michigan. My friend John Hinderaker of Powerline, who lives and works in Minnesota, has observed that in making this “catastrophic error,” Ramsland (a Texan who appears to be unfamiliar with the two states in question) may have confused the postal codes: Minnesota is MN, but some mistakenly assume it to be MI, which is Michigan.
In a subsequent affidavit, Ramsland managed to name Michigan cities and towns correctly, but made wildly inaccurate allegations of inflated turnout figures that suggested massive fraudulent voting. He claimed, for example, that turnout in Detroit was nearly 140 percent. Of course, it would have been implausible to claim that 100 percent of eligible Detroit voters had turned out; but that’s too modest for Ramsland, who larded on another 40 percent for good measure; after all, we’re talking fraud here. In reality, the Detroit Free Press reports (based on government records) that turnout in the Motor City was about 51 percent, right in line with what it has been for many past presidential elections.
Compared to other claims he’s made, Ramsland’s Detroit allegation was pretty tame. To take another example, he claimed that turnout in North Muskegon (pop. 3,786) was a whopping 781.91 percent. This was a tenfold overstatement of the actual figure, 78.11 percent.
Ramsland’s latest report did not arise out of the 2020 presidential election. The dispute, instead, traces to a local referendum in the Antrim County village of Central Lake. It involved a proposal to permit a marijuana dispensary. The vote was a dead heat, 261 to 261, on the initial tabulation. The outcome shifted after a recount on November 6, with the proposal passing by one vote. A local resident, William Bailey, filed a lawsuit challenging the result and alleging that ballots were damaged during the recount, possibly causing his own vote to be discounted. Mr. Bailey retained Michigan lawyer Mathew DePerno, who asked the local circuit court for permission to conduct “forensic imaging” of Dominion’s tabulators and brought Ramsland’s firm aboard for that purpose.
State Circuit judge Kegin Elsenheimer (a former Republican state legislator who was appointed to the bench by Rick Snyder, the former Republican governor) permitted the testing under state and county supervision. Ramsland’s resulting report was submitted to the court under seal and there were whispers that it was explosive. Judge Elsenheimer made the report public Monday (the same day the Electoral College met in each state, and Michigan cast its 16 votes for Biden). The president responded by tweeting, “68% error rate in Michigan Voting Machines. Should be, by law, a tiny percentage of one percent. Did Michigan Secretary of State break the law? Stay tuned!”
Sigh.
As noted above, even the report does not claim a 68 percent error rate. It is an alert rate that refers to lines in Antrim’s tabulation logs. The logs record sundry technical activities related to the process of tabulation, not to individual ballots. The Free Press further reports that the lines in the tabulation logs are not the lines of software code that run the system — which is what gets reviewed in certifying election results.
In asserting that the “error rate . . . Should be, by law, a tiny percentage of one percent,” the president was referring to a section of the report in which Ramsland claimed: “The allowable election error rate established by the Federal Election Commission guidelines is of 1 in 250,000 ballots (.0008%).”
This is gibberish. The FEC does not regulate voting equipment; its bailiwick is campaign finance. As the Free Press details, the Election Assistance Commission, which has certified Dominion equipment, is the federal agency that deals with voting equipment. It is not clear from where Ramsland pulled his .0008 percent standard, but it is neither a law nor an FEC guideline. And as he imagines it, it relates to ballots (“1 in 250,000 ballots”); yet his report is not claiming a ballot error — the 68 percent rate he touts relates to “15,676 individual events” in tabulator logs.
Michigan’s state government points out that if President Trump truly believed there had been wrongdoing, he “could have requested a hand recount of every ballot in the state.” Having lost the state by 155,000 votes, the Trump campaign did not ask for a hand recount. To allay concerns, however, Michigan conducted one in Antrim County anyway. On Thursday, this audit resulted in a net gain of exactly 12 votes for the president, out of 15,962, meaning the original count was off by 0.07 percent. That is, Trump won the county by 9,759 to 5,959 — not by the original tally of 9,748 to 5,960.
Another rabbit hole.