paywall

paywall

This is an edition of The Atlantic Daily, a newsletter that guides you through the biggest stories of the day, helps you discover new ideas, and recommends the best in culture. Sign up for it here.

Stop me if you’ve heard this story before: Partisan claims of fraud in the presidential election. Elaborate statistical analyses. Reports of shadowy, closed-door doings. All of this, they say, points to one conclusion: The results were compromised, and the real winner was kept out of the White House.

That sounds like the aftermath of the 2020 election, but it’s also what’s happening right now. Kamala Harris’s loss in last November’s presidential election produced few prominent claims of fraud, and nothing like the concerted effort, using both lawsuits and force, to keep President Donald Trump in office that followed his defeat nearly five years ago. In the past few months, however, spurious allegations that fraud helped Trump win back the White House have been flourishing more online, elections experts told me, though why they’re so popular right now—other than the left’s compounding anger with the Trump administration—is not clear.

The parallel to fraud theories about the 2020 presidential election is more than superficial, Justin Grimmer, a political scientist at Stanford who has studied election-conspiracy theories, told me. “The most remarkable thing is the similarity in the analysis that we’re seeing from the bad claims made after 2020 and these similarly bad, really poorly set up claims from 2024,” he said.

One popular example alleges that an NSA audit of the 2024 election found that Harris, not Trump, had actually won, according to a former CIA officer who allegedly participated in the audit. On July 31, an anonymous Substack newsletter called This Will Hold, which claims to offer “the truth they’re not telling you,” published a post stating, “In an exclusive interview, former CIA operative Adam Zarnowski laid out pieces of an intricate network of bad actors and covert operations behind transnational organized crime and the stolen 2024 election.” It adds that “none of his revelations are classified” and that Zarnowski “is prepared to testify under oath.” The implication of this bombshell is clear to the author: “We have the authority and the obligation to remove this entire unelected, illegitimate regime.”

The theory has many problems. No evidence exists for Zarnowski’s claims about his background other than his own word. Elsewhere, a LinkedIn profile calls him a “former CIA paramilitary operations officer” and an expert in the subject of human trafficking, but nothing suggests his statistical or elections expertise; a self-published book is full of oddball claims. I attempted to reach Zarnowski using a couple of different methods but received no response. (Snopes, which was able to contact Zarnowski, reported that he did not provide definitive proof of his professional background or the alleged audit.)

Moreover, nothing in the Substack post actually supports Zarnowski’s claims; instead, it offers innuendo about voting-machine failures and the companies that sell elections equipment. Neither the NSA nor any other federal agency conducts elections audits, nor is there any plausible explanation for why they would do so. The absence of an actual audit here or anywhere else is notable: As with the claims offered by Trump and his allies in 2020 and 2021, the theory relies on implication, with hard evidence seemingly always just out of reach.

But there are more fundamental issues of logic in the theory. States actually do conduct audits of their votes, and unlike the supposed NSA audit, the process and results of those reviews are public. The theory appears to suppose that Democratic officials in key swing states conspired to help Trump and hurt Harris, for whatever unstated reason. These claims “ring as hollow and grifting as nearly identical claims made by those who profited off the Big Lie that Trump didn’t lose the 2020 election,” David Becker, the executive director of the Center for Election Innovation & Research, a nonpartisan nonprofit, wrote in an email.

The post from This Will Hold mentions another group peddling similarly bogus claims of fraud. The Election Truth Alliance, which describes itself as nonpartisan, offers a different spin on fraud claims—less cloak-and-dagger, more regression-analysis-and-spreadsheet—but ultimately not one that is any more convincing. The ETA argues that “patterns consistent with vote manipulation are present in 2024 U.S. election results,” a conclusion “based on analysis of publicly-available state and local election data using multiple evidence-based methodologies.” This is more promising than the Zarnowski chimera, but only on the surface. One of the ETA’s methods involves analyzing how Harris fared compared with down-ballot candidates. For example, both the ETA and another group called SMART Elections have zeroed in on Rockland County, New York, noting that Harris got many fewer votes there than did Senator Kirsten Gillibrand, a fellow Democrat.

Like so much other purported evidence for fraud, the disparity can be explained in mundane terms: Many people don’t vote in every contest, and although presidential candidates tend to receive more votes than down-ballot contenders, the rule isn’t firm. Former Senator Joe Manchin received more votes than did Democratic presidential candidates in West Virginia, cycle after cycle. The MIT political scientist Charles Stewart III demonstrates that Harris’s underperformance in Rockland County relative to Gillibrand appears to stem from her unpopularity with ultra-Orthodox Jews in the county.

The ETA employs two similar metrics that compare reported results with projections, based on the number of votes cast. The problem is that voter behavior is messy and unpredictable. That’s the point. If elections fully conformed to models and expectations, there would be no need to hold them. But the patterns are also sometimes predictable. “This is one of the big errors that was made in the post-2020 analyses,” Grimmer told me. “A whole group of amateur statisticians were shocked to find out that in a small number of heavily populated counties, the Democratic candidate does quite well, and that in a large number of sparsely populated counties, the Republican candidate tends to do better.”

One reason that claims of fraud seem to grab hold of some people is that they are conveyed via elaborate-seeming statistical analyses, which may or may not be valid uses of the data but are enough to impress casual viewers (or at least to make their eyes glaze over). The ETA also posted a “working paper” by Walter Mebane, a respected political scientist at the University of Michigan, that statistically examined 2024-presidential-election results in Pennsylvania. When I reached out to Mebane recently, he told me that he had not closely examined claims of misconduct in Pennsylvania but believed colleagues who had deemed them unfounded. He added that the ETA had provided him with useful data but that he didn’t endorse its claims. “They have a lot of things they say I don’t agree with, but I’m not taking the time to fight with them in public,” he said. (In an email, the ETA agreed that “a noticeable down-ballot difference could result from a more popular candidate at either the presidential or lower-ballot level,” but stood by their methodologies and findings. They added that they intend to move forward with litigation in two states in which the organization claims to have discovered “evidence consistent with vote manipulation.”)

The problem with any claims of election fraud on a scale that could change results, setting aside the statistical flaws, is that they ask audiences to accept abstract interpretations of numerical data while ignoring real-world information. For example, almost every state allows election observers and has poll workers from multiple parties. To change votes would require that multiple people across parties conspire to flip votes and then stay quiet about it—and also that no voters or observers notice. “That seems pretty far-fetched to me,” Tammy Patrick, the chief program officer at the Election Center and a veteran elections official, told me.

Knocking down false claims is frustrating work, especially when the same ideas that were debunked four years ago pop up again from new culprits. Grimmer has spent countless hours chasing down the truth, explaining it to reporters, and even debating election deniers. And so I was struck by the compassion he showed for people who fall for the theories. “The people who believe them, they’re not crazy people,” he told me.

“It’s hard to believe that a majority of the country disagrees with your choice when you’re so passionate and certain about your choice,” Grimmer said. “They’re smart people, and they think, I must be able to discover what’s going on here.” Sometimes, though, reality just doesn’t work the way we expect.

Related:

Election officials are under siege.Anne Applebaum: Trump wants you to accept all of this as normal.

Here are three new stories from The Atlantic:

Trump tells pregnant women to “fight like hell” not to take Tylenol.Daniel Byman and Riley McCabe: Left-wing terrorism is on the rise.Alexandra Petri: Sure, let’s try bribes!

Today’s News

At the United Nations, President Donald Trump gave a speech that criticized the organization and its global efforts on climate change. He also warned that without action on immigration and stronger borders, “your countries are going to hell.”Trump posted on social media that Ukraine, with NATO’s financial support, could feasibly win back all of its territory lost to Russia during the war.According to a New York Times investigation, Errol Musk, the father of Elon Musk, has been accused of sexually abusing five of his children and stepchildren since 1993; Errol denies the claims, calling them “false and nonsense in the extreme.”

Evening Read

photo of a mother and her two children at home Evelyn Dragan / Connected Archives

AI Is Coming for Parents

By Miranda Rake

A few weeks before my daughter’s fourth birthday, I stumbled across an AI party planner called CelebrateAlly. “Looking to plan a themed party, a surprise bash, or just a relaxed get-together?” read a banner on its website, which promised that the app would take care of “all the details—themes, activities, and decorations.” It also offered to write birthday cards, “capturing your heartfelt sentiments beautifully!”

The offer had a certain appeal. I was overwhelmed, entering the phase of planning where I actually had to execute on my daughter’s vision for her bash. We’d been talking about the party for months, and her requests were specific yet constantly changing. (She wanted a unicorn cake—no, a unicorn piñata; to invite only her cousins—then a few of her friends too, and then all of the kids on our block.) But I was genuinely curious to hear them. Each question I asked her was a way to draw closer to her: I learned about who she is right now while, I hope, showing her that I really want to know. After all those conversations, using AI would have felt like a betrayal.

Read the full article.

The People Who Are Still Convinced Kamala Won

The People Who Are Still Convinced Kamala Won