If you were familiar with Mike Lindell a year or two ago, it was probably because you watch Fox News and had seen the ubiquitous ads for his company, MyPillow. Lindell appears in those ads to hype his pillows with Billy Mays levels of gruff enthusiasm.

Over the past six months, though, Lindell’s become better known as a salesperson for something far less comforting: former president Donald Trump’s unfounded claims that the 2020 presidential election was stolen.

Lindell’s wealth has made him a particularly loud voice among those clamoring about the election. He has the resources to hire various dubious “investigators” and to produce shakily constructed videos detailing what they’ve found. He also has the resources to respond to a 10-figure defamation lawsuit from Dominion Voting Systems, not by acquiescing to having spread unverifiable claims but, instead, with a countersuit of his own in which he repeats and elevates those claims.

That countersuit, filed this week, is the written version of Lindell’s “documentaries,” melodramatic, glitchy, sweeping and deeply flawed in both obvious and non-obvious ways. Central to the effort are those claims that the election was stolen, a claim that the suit reiterates explicitly as an exculpatory point for Lindell’s assertions about Dominion’s voting machines.

“Fact,” the suit states at one point: “Direct and circumstantial evidence demonstrates that, during the 2020 General Election, electronic voting machines like those manufactured and sold by Dominion were manipulated and hacked in a manner that caused votes for one candidate to be tallied for the opposing candidate.”

This is, of course, not a fact, since it isn’t true. But this claim — that Lindell can prove or has proved that fraud occurred — is meant to bolster his public assertions about the company. After all, if rampant fraud occurred in places where Dominion’s machines were used, how could he be to blame for saying that they made that possible?

The catch here is that Lindell offers very little that’s actually intended to serve as direct evidence of malfeasance. There is a lot of hand-waving about questions that had been raised about Dominion’s machines and lots of ad hominem assertions about the company and its employees, but the suit introduces very little that might be considered actual, direct evidence that votes were manipulated.

Instead, there’s a lot of circumstantial stuff — like that Dominion wouldn’t turn over proprietary passwords to the team in Arizona that’s working on behalf of the Republican majority in the state Senate — to gin up questions about the election results. This was addressed in a scathing letter from county officials in Arizona explaining that the problem is the “auditors” chosen by the senators lacked the necessary credentials to do the research they wanted to do. The existence of that effort is itself presented by Lindell as evidence that something untoward happened, which is a bit like spending $250,000 on Bigfoot deterrents and then pointing to that investment as evidence that Bigfoot must exist.

More than six months after the 2020 presidential election, Arizona Senate Republicans are leading an audit of the 2.1 million ballots cast in Maricopa County. (Erin Patrick O'Connor/The Washington Post)

There were some minor questions raised. One was that reported vote totals in Bibb County, Ga., changed on election night — a common occurrence during elections, given that thousands of locations are inputting vote totals in a short time period. Another was that Dominion had unexpectedly updated software the night before the election, marring voting in one Georgia county; this was later shown not to be the case.

Another bit of evidence presented by Lindell cites a complaint filed by an attorney named Matthew DePerno on behalf of his client in a lawsuit targeting Antrim County, Mich., where a vote adjustment also occurred on election night. Lindell’s lawsuit quotes a determination about Dominion’s machines made by an organization called Allied Security Operations Group: “The system intentionally generates an enormously high number of ballot errors ... The intentional errors lead to bulk adjudication of ballots with no oversight, no transparency, and no audit trail.”

His lawsuit leaves out the part preceding that line: “We conclude that the Dominion Voting System is intentionally and purposefully designed with inherent errors to create systemic fraud and influence election results.”

It also leaves out that this analysis of Antrim County has been debunked repeatedly. It also leaves out that Allied Security Operations Group is so unreliable that it was viewed as too partisan even to participate in the Arizona “audit.”

At its heart, Lindell’s lawsuit narrows in on two pieces of purported evidence.

The algorithm

One claim made in the lawsuit is that the vote results in several swing states show evidence of manipulation.

From the lawsuit:

“In early 2021, a data scientist, Douglas G. Frank, PhD, uncovered an algorithm or ‘key’ — a sixth degree polynomial — that operates in the electronic voting machines in a number of states to determine the ballots cast. These algorithms are unique to each particular state. In other words, the algorithm used in Minnesota does not work next door in Wisconsin.”

Frank argues that he identified a pattern — that complicated sounding polynomial — that the results in nearly every county in a state that he looks at will almost precisely match. In a video of his own, Frank marvels at the fact that election results should so closely fit that pattern.

I actually looked at this claim last month. Unsurprisingly, it’s utter nonsense.

What Frank did to generate the key in Michigan, for example, is take the results in four counties in the state and average turnout by age in those four counties. Then he generated a line that was curvy enough to fit the average closely and fit it to the counties he was tasked with analyzing.

Nine counties in total, four of them being the ones used to generate the average.

This is like taking four Usain Bolt 100-meter races, averaging his time in those four and then comparing that average to those four races and then another five. Guess what? The average is going to be pretty close to the results across the board. Because the average was derived from the data it was evaluating.

Anyway, my prior article goes into this in more depth if you require more evidence. But I will point out one quote from Frank.

“There are a few little wiggles that don’t perfectly line up, but that’s not unusual because, after all, we’re dealing with human behavior,” he says in his video — despite his point being that what his key shows is not human behavior. “But for me to be able to predict that that well, you know there’s an algorithm function.”

Yeah, there is: the average that you yourself generated.

A fun aside: one of the people who’s being most actively hyping Frank’s research is a lawyer named Matthew DePerno.

The hacking

The other primary piece of evidence presented by Lindell in his lawsuit is that his analysts have found evidence that Chinese actors manipulated vote totals in a number of states.

From the lawsuit:

“Exhibit 12 shows a subset of 20 documented successful hacks through the election management system in the states of Michigan, Pennsylvania, Georgia, Wiscinsin, [sic] and Arizona resulting in a total 555,864 votes switched from President Trump to candidate Vice President Biden in the 2020 general election. These hacks came primarily from within China and are identified by the date, location, and the network from which the hack orginated [sic] and the location and network that was the target of the hack.”

For further explication of the findings, the lawsuit points readers to the most recent of Lindell’s documentaries, “Absolute 9-0.” It includes a lengthy list of complicated-looking data scrolling rapidly over the screen, the analyst himself — his face blurred — and a table of information about the points of origin of the purported hacks.

That scrolling text is supposed to represent PCAPs, packets of Internet traffic. What’s shown, though, is what appears to be voter database information rendered into hexadecimal digits.

If you don’t believe me, you can type in examples of the text yourself and convert it to readable text. Incidentally, the Pennsylvania voter file, from which these data appear to come, can be purchased from the state or a data vendor.

In other words, it’s not clear where this data came from or what it’s supposed to show. If it was Internet traffic, it’s being sent unencrypted and, for some reason, in hexadecimal instead of binary. What it doesn’t show is votes being flipped.

It’s not really clear what that would even mean, in fact, though Lindell’s analyst does explain that they had to do some translation from Chinese.

“We, so we, you know, because we’re dealing with other languages,” he says at one point, “we validated the validation that was validated.”

Okay, got it.

The “Exhibit 12” referenced in Lindell’s lawsuit includes a list of 20 connections from foreign countries that purportedly involved reducing Trump’s vote total. Again, how this supposedly happened isn’t explained. What that translated and triple-validated Chinese discussion included in the PCAPs (and maybe in the Pennsylvania voter file?) actually says isn’t mentioned. It’s just, here are 20 times Trump had votes stolen.

Similar data also appears in Lindell’s first documentary. In that video, the alleged vote-stealing data includes more information, including what are called MAC addresses, unique identifiers for Internet-connected hardware. It appears to be the same data (as Twitter user “zedster,” cited above, notes) that was cited in a book written by former Overstock CEO Patrick Byrne. The full spreadsheet of purported connections is on Byrne’s website, though the list of places where votes were changed is different in the two documents.

It’s also the case that the MAC addresses presented in the Byrne document and the first Lindell video are obviously made up. A MAC address is composed of six numbers, from 0 to 255, encoded in hexadecimal from 00 to FF. Some of the digits have particular meaning; the second character of the first number, for example, can be used to identify how it is administered and how it receives information. Yet, despite those differences in functionality, the second characters in each address in that file are evenly distributed from 0 to F (that is, 0 to 16). None is significantly more common than any of the others.

I know this is confusing, so let me use an analogy. Imagine that you had a list purporting to show a random collection of license plates. In your state, cars are given plates that start with the letter “A,” trucks get the letter “B” and tractors get a letter “C.” You analyze your list and find that it includes 1,000 “A” plates, 998 “B” plates and 1,003 “C” plates. You might suspect that something was up.

In fact, the same holds true for nearly every number included in the MAC addresses in that file. They all appear about as frequently as the others, within a narrow range. All except 255, “FF” — as though someone was generating random numbers less than 255, forgetting that 255 should be a permissible value.

Maybe Lindell’s “Absolute 9-0” video is using some other set of data that just happens to comport entirely with individual parts of Byrne’s spreadsheet. And maybe Lindell’s analyst has some way of verifying that votes were flipped besides literally doing nothing more than claiming that they were. Then the question becomes how: how did this happen without anyone noticing, how did it route through the Internet in a process that’s localized in states and why did it happen in some cases days after the election (purportedly) and from eight different countries?

Oh, and how did someone figure out how to do all this but not figure out how to actually encrypt what they were doing?

The short answer is that there’s simply no reason to assume that this is true. Literally none. It is much more likely that the conspiracy here is that some tech-savvy guys figured out how to bill a credulous billionaire for weeks of work than it is that this unnamed guy found actual evidence of vote-flipping and, instead of going to the police and becoming a celebrity, he went to a pillow salesman.

In other words, one assumes that Dominion’s lawyers aren’t that worried about this lawsuit. Particularly given this new development on Friday afternoon: