This article has been updated.

The first concrete claim President Trump made about how the election was theoretically being stolen from him came during remarks he made in the middle of the night after polls closed on Nov. 3.

“This is a major fraud on our nation,” Trump claimed, after stating that he had won his reelection bid. “We want the law to be used in a proper manner. So we’ll be going to the U.S. Supreme Court. We want all voting to stop. We don’t want them to find any ballots at 4 o’clock in the morning and add them to the list, okay?”

The next morning, he made a similar claim, saying “surprise ballot dumps” were shifting the results of the contest. These “surprise” vote totals were anything but, of course. They were simply states slowly counting a deluge of mail-in ballots spurred by the coronavirus pandemic, generally in states that had been prevented from beginning to count ballots before Election Day. They were, in other words, regular old votes, slathered with a nefarious veneer by a president eager to deny the reality of public opinion.

Over and over and over, we’ve seen Trump allege fraudulent activity in the 2020 election that is quickly revealed to be unexceptional. He claimed that ballots were missing that weren’t. He hyped that a batch of votes in a small Michigan county were wrongly given to Joe Biden, an error caught and rectified. He assured his supporters that his team was being prevented from observing vote-counting, though his lawyers later had to admit that they hadn’t been — though a lack of observers doesn’t necessarily mean illegal activity occurred anyway. He repeatedly insisted that electronic voting machines made by Dominion Voting were somehow used to flip votes against him, though there’s no credible evidence that this occurred and both recounts and paper-record audits in Georgia, where the machines were used, showed no such errors. He dropped all sorts of putative evidence into his Twitter feed — surveillance videos, sketchy reports from dubious cable news channels, intentionally complicated documents — trying to create an impression of malfeasance even if he couldn’t prove any.

There was no allegation too minute or too old for Trump to elevate it. And as the days passed, nothing stuck. There was no actual evidence to prove his insistence that the election was stolen. His supporters were buffeted from one claim to the next, accepting the premise that even if each of these claims was quickly dispatched, the pile of bones left behind implied that something had gone terribly wrong. The same pattern kept playing out: A new claim or new bit of “evidence” emerged, and all of the world’s Mark Levins and Eric Trumps threw up their hands at the sheer audacity of the theft — and then, from that elevated platform of agitation, stepped up to whatever claim Trump made next. Higher and higher and standing on nothing.

On Monday, Trump threw out the latest claim.

It was not big news.

The “judge” to whom Trump refers is the one who authorized the release of the report claiming to serve as robust independent analysis of Dominion machines in Antrim County, Mich. He did so because county officials had publicly rebuked the report’s findings, not out of any “bravery” or “patriotism.”

The rebukes stemmed from the immediately apparent bias of the report itself. It was published by Russ Ramsland, an “expert” whose last effort to prove rampant fraud in the election involved his mistaking precincts in Minnesota for ones in Michigan. After reviewing the results of the voting in a county where 0.3 percent of the state’s votes were tallied, he presented his stark conclusion: “We conclude that the Dominion Voting System is intentionally and purposefully designed with inherent errors to create systemic fraud and influence election results.”

It is almost certainly fair to assume that Ramsland arrived at that conclusion well in advance of his actual analysis. The document that resulted from his research asserts that 68.05 percent of the “events” logged by the county’s system were “recorded errors,” a vague term later clarified as being “critical errors/warnings,” most of which were “related to configuration errors” that could result in “overall tabulation errors.” In other words, this top-line number, elevated in conservative media, doesn’t mean that 68.05 percent of votes were changed or otherwise suspect. It’s not clear what it means, if anything, though he asserts it is somehow related to ranked-choice voting options.

Dominion read through Ramsland’s report, too. It points out that a number of assumptions Ramsland made are simply wrong. Antrim didn’t “adjudicate” any votes (meaning determine a voter’s intent after the fact), Dominion says, because the county didn’t have that tool. Contrary to what Ramsland claims, the county’s machines don’t have modems. Oh, and the county doesn’t use ranked-choice voting.

Most of Dominion’s response, though, focuses on something Ramsland would probably agree with — that county administrators made errors in their administration of the vote. After all, Antrim is the aforementioned small Michigan county where thousands of votes for Trump were incorrectly recorded as being for Biden. County officials updated their ballots without updating how the scanners read the ballots, meaning that Trump votes appeared where the machines expected Biden ones. This is why Ramsland was looking at Antrim in the first place — because it had already messed up.

Once those votes were fixed, though, there’s nothing particularly weird about its results. The county saw a bigger increase in the number of votes for Biden than most counties did, but Trump’s margin of victory this year was only 200 votes less than his margin in 2016. Compared with Michigan’s counties overall, Antrim’s increase in votes cast and its shift in the overall margin is unremarkable. Also unremarkable is how those shifts looked in counties that did or didn’t use Dominion voting systems.

Oh, also: Antrim County backed Trump by a 24-point margin. If elections administrators there were trying to throw the state for Biden by futzing with the county’s 16,000-odd voters, they did a very, very bad job.

Ramsland’s report attempts to obfuscate his claims using a flurry of complicated-sounding analyses. That he reports the error rate using two digits to the right of the decimal point is fully in keeping with the rest of the document, which, for example, goes into great detail about how a server’s “Windows updates are 3.86 years out of date,” which is certainly a normal way to refer to time periods. It is in keeping with many of the post-election scramblings offered by the president, documents and claims that often rely on the murkiness of technological terminology to obscure the shakiness of what’s being asserted.

What Trump ignores is that even if Ramsland proved that something nefarious happened in Antrim County, which he did not, it doesn’t mean anything. Even if elections administrators in that one county were actively skewing the vote against the president, that means nothing for any of the state’s 82 other counties, particularly those counties where 45 percent of the state lives and that don’t use Dominion machines. It means nothing for the thousands of other counties that voted more heavily against Trump than they did four years ago.

To put a fine point on it: Many of the people who are alleging that this proves any fraud, much less rampant fraud, are lying. Trump may at times elevate things that he thinks might make his case, but it’s clear that, at other times, he’s lying about fraud, too.

Having a report in hand that makes a sweeping claim about the possibility of fraud no more proves fraud than you holding a knife proves that you committed murder.

Update: Antrim County finished a hand-recount of its ballots, comparing the electronically tallied results with the paper records produced when voters voted.

The recount matched the original tally, save for a swing of 12 votes in Trump’s favor.