No, Michigan didn't use an algorithm to manipulate 2020 election results

Antrim County, Michigan, the site of a human error that briefly produced inaccurate unofficial results on election night in November, remains the target of a misinformation effort that aims to sow doubt about the integrity of the 2020 election.

A resident’s lawsuit against the county, filed in late November, helped fuel a debunked conspiracy theory that tabulators made by Dominion Voting Systems switched votes on behalf of Joe Biden. Donald Trump seized upon an analysis filed in the suit that was rife with inaccurate information to advance his claims of a stolen election. A hand recount of every ballot cast in Antrim County affirmed the results and Trump’s victory there. 

Now, a new filing in the case claims a more far-reaching conspiracy, one in which the state government purportedly deployed a predetermined "key" to mathematically "convert" records of registered voters into counted votes.

The filing cites an analysis by Douglas Frank, who chairs the math and science department at a school in Cincinnati, that claims to have discovered that Michigan and other battleground states each used a unique key.

The filing defines the alleged "key" as a "sixth degree polynomial" that "unlocks the door and uncovers the ability to manipulate data and results." 

Frank’s analysis was filed as an exhibit in the case, and he explains his findings in a YouTube video that has received over 26,000 views. Frank says his work shows that "elections are being decided at the state level."

How? "Ballots are being harvested at the precinct level, regulated at the county level, and determined at the state level," he claims.

In an interview with MyPillow CEO Mike Lindell, who has peddled disinformation about the election, Frank says he found algorithms were used in every state he’s analyzed to regulate voter registrations and ballots and control the election.

The filings were submitted by Matthew DePerno, the lawyer for the Antrim County resident, who is accusing the county of mishandling the results on a marijuana-related proposal that was also on the November ballot.

They were shared widely by The Gateway Pundit, a conservative website that has published falsehoods about the election and former Michigan state Sen. Patrick Colbeck, who was recently asked by Dominion Voting Systems to retract claims of voter fraud.

While the analysis is new, it boils down to a familiar falsehood: that Michigan’s 2020 election was tainted by widespread misconduct and fraud.

There is no evidence for that and a comprehensive series of audits validated election processes and affirmed the results.

In the filing, DePerno argues based on Frank’s finding that the number of ballots counted from Antrim and other counties tracks so closely with the vote totals predicted by the purported key, that they could not have occurred naturally without manipulation by the state.

They claim the correlation proves that an algorithm had access to voting databases to determine the election result. 

Neither DePerno nor Frank provides any evidence for their claim of a state-level conspiracy to manipulate local results, other than the statistical correlation between ballots they claim were cast in the election and what the key — whose derivation they don’t explain — predicts.

Gerrid Uzarski, elections director for Kent County, one of the counties examined by Frank, called the allegation "very far-fetched."

And he questioned the analysis used to support the assertion.

"I’m not sure where he got the numbers from," Uzarski said. "I’ve looked for them myself. I have not been able to find the source."

Featured Fact-check

X posts

stated on October 19, 2024 in a post

Michigan has “more registered voters than eligible citizens. This creates a huge risk of election fraud.”

Frank didn’t respond to emailed questions about how the key was generated, but he said he stood by his findings, which he said are based on information "directly extracted from the databases provided by the state."

PolitiFact Michigan looked at data from the Secretary of State’s website and official county results and found problems with numbers Frank used and the conclusions he drew.

For his analysis, Frank said in the video, he compiled voter demographics, registrations and ballots in nine Michigan counties — Antrim, Barry, Charlevoix, Grand Traverse, Kent, Livingston, Macomb, Oakland and Wayne — and predicted the number of ballots cast using his key. In charts for each county, he breaks down the ballots cast by voters’ age. Trump won six of the counties analyzed by Frank.

Frank lists the Census Bureau’s July 2019 population estimates as his source for the number of residents in the county. But the numbers listed in the court filing differ slightly from the census estimates. In a phone call, Frank said the numbers differed because he performed an "interpolation" of the census data.

For each county, Frank provides a breakdown of the voting age population and compares it to the number of registered voters in the county and concludes that the number of registered voters in each county is near or exceeds the size of the voting age population. Ahead of the election, many counties did have more registered voters on their rolls than voting age residents. That can happen if the rolls aren’t updated to reflect people who have moved out of the county or died.

But in order to cast a ballot, voters have to show up to their polling location or fill out and sign an application to request an absentee ballot. Election officials verify that the ballots are given to individuals legitimately registered in Michigan. Cases of voter impersonation are exceptionally rare.

Frank then looks at what he claims is his database of registered voters from October 2020. His numbers track closely with voter registration numbers listed on the Secretary of State’s website on Oct. 2, 2020.

That suggests that his analysis disregards registered voters who were added after that date — which was one month before Election Day — and the ballots they cast. Indeed, Frank acknowledges that his analysis of turnout and ballots cast counts only those who he says were registered by October. He refers to the other ballots as "unregistered ballots," and alleges that these were "injected" at the last minute.

But voters in Michigan can register to vote right up through and on Election Day, and thousands took advantage of this right in the November 2020 election. Election officials check to make sure that every ballot cast belongs to a registered voter and that every voter only votes once.

The exclusion of thousands of ballots cast in the November election could explain why the turnout rates he lists in the court filing are also inconsistent with those reported in each county’s official election results.

Uzarski said the type of fraud Frank alleged would be an insurmountable task given Michigan’s highly decentralized election system.

In Michigan, 1,520 city and township clerks process voter registrations, verify ballots and oversee the count on Election Day. Results are compiled by 83 county clerks and certified by county boards of canvassers who review ballots, poll books and the results. Only after the Board of State Canvassers certifies the results is the outcome considered final.

It would be impossible to cook up thousands of ballots that were invalidly cast, Uzarski said. "There is not room for any kind of injection of ballots or injection of ghost voters or people who didn’t vote but people are voting for them."

David Becker, founder and executive director of the Center for Election Innovation & Research, called the allegations in the pleading "fantastical."

In an email to PolitiFact Michigan, Becker explained that such a conspiracy would have had to be carried out by thousands of people and leave behind a mountain of evidence: meddling with the voter file, extra ballots that couldn’t be explained, voters who tried to vote but couldn’t because someone else had cast their ballots and audits confirming tabulators were tampered.

"There is literally zero evidence of any conspiracy, involving thousands of people, in any state," including Michigan, Becker wrote.

In a YouTube video, Frank claims his analysis of Michigan’s 2020 election results shows that the election outcome was decided at the state-level ahead of time using a mathematical "key" to manipulate vote totals.

Frank offers no evidence for his claim other than a statistical correlation to this key, whose derivation he doesn’t explain. The figures he cites as the basis of his analysis are different from those in official records.

Experts said the vote-manipulation effort Frank theorizes would involve a wide-ranging conspiracy that would not be possible under Michigan’s decentralized election system.

We rate his claim Pants On Fire.