Keep the Cat Memes Flowing: Interpretive Polarization and the Strategy Behind One of 2024’s Most Effective Political Narratives

On September 15, 2024, JD Vance sat across from Dana Bash on CNN and defended the claim that Haitian immigrants were eating pets in Springfield, Ohio.

The story was false. It had begun as a fourth-hand rumor in a local Facebook group and had already been rejected by Springfield’s Republican mayor and Ohio’s Republican governor. Yet Vance did not back down. Instead he said something revealing: “If I have to create stories so that the American media actually pays attention to the suffering of the American people then that’s what I’m going to do.”

This was widely treated as an admission that he had lied. More importantly, it was a blunt account of how political reality gets made.

Dan Williams recently argued in Conspicuous Cognition that our deepest political divisions are often not disputes over facts but disputes over interpretation. Rival factions may accept much of the same underlying evidence while placing it within radically different narratives and moral frameworks. His central examples are the deaths of George Floyd and Henry Nowak, two deaths captured on video and elevated by opposing political camps. The disagreement in each case was less about what happened than about whether the incident was representative, who counted as the victim or villain, and what political conclusions should follow. Williams calls this interpretive polarization.

I think he’s largely right, with one asterisk: on some major issues the disagreement really is about facts. Millions of people sincerely believe that the 2020 election was rigged or that vaccines cause autism. But Williams’s larger point stands — that much political conflict turns not only on which facts people believe, but on which they consider representative and what they take those facts to mean. I’ve made a version of this argument myself when talking about bespoke realities.

Williams is also clear that interpretations do not arise solely inside individual minds; they are socially supplied and politically constructed. He nods at a professional class of politicians, pundits, and activists who “construct and update the details of the pseudo-environment in real time,” along with the status incentives that keep them doing it.

It’s those figures and incentives that I want to stay with in this post: the suppliers, their motivations, and their power. What turns an obscure event into a representative symbol? Who benefits when it catches on? And what happens when the people who profit from an interpretation also work to destroy the legitimacy of anyone capable of challenging it?

Put another way: When does interpretive polarization emerge through ordinary public sense-making, and when is that process deliberately engineered?

Lippmann’s Problem, Bernays’ Opportunity

The idea that people inhabit rival interpretations of the same world is not new. Williams reaches back to Walter Lippmann’s 1922 book Public Opinion and its idea of the pseudo-environment: the simplified “pictures in our heads” through which we encounter a world too large to experience directly. We don’t first observe a neutral set of facts and then decide what they mean. We inherit systems of classification that decide what we notice and what conclusions seem available. “We define first,” Lippmann wrote, “and then see.”

Lippmann’s citizen is a sympathetic figure. His world is big and fast. He has a job, a family, and a finite amount of attention, and he cannot become an expert on monetary policy, immigration law, public health, and every other subject on which democracy asks him to hold an opinion. He must rely on maps drawn by other people.

Democratic government therefore depended, Lippmann argued, on interpreters — experts, institutions, officials, and journalists — who could investigate the world, make it intelligible, and communicate what the public needed to know. That work necessarily involved selecting and interpreting information. Lippmann understood that this could be manipulated. He called persuading people into agreement, rather than waiting for it to form, the “manufacture of consent,” and warned that advances in psychology and mass communication had made the old art of persuasion far more powerful.

Edward Bernays looked at the same problem and saw a profession.

Bernays was Lippmann’s contemporary, Sigmund Freud’s nephew, and more or less the father of public relations. He reasoned that if people lived by pictures in their heads, professionals could decide which pictures acquired authority and appeal. He advised working to reach people through their group identities — their symbols, aspirations, and fears — and recruiting trusted opinion leaders to carry a message. In his 1928 book Propaganda, Bernays described the people who performed this work as an “invisible government” and “the true ruling power of our country.” “There are invisible rulers who control the destinies of millions,” he wrote. He meant it as a compliment. He later called the process the “engineering of consent.”

“Manufacture” describes the production of the picture; “engineering” directs our attention to the agent behind it.

A Century Later

A century later, the engineering is no longer controlled by a small class of publicists working through a handful of mass-media gatekeepers. Political interpretations now emerge and evolve through the interaction of politicians, influencers, partisan media, platforms, and participating crowds.

Bernays worked through a comparatively top-down mass-media system, cultivating editors, organizations, and opinion leaders. The system that replaced it is a wide-open market for influence, with algorithms as its gatekeepers. Millions of people make content, and thousands have built significant audiences. As platforms push more “unconnected content” from accounts users do not follow, creators make content for recommendation systems as much as for their own followers. A story that performs well and signals the right loyalties is worth more than one that merely happens to be accurate.

The system rewards participation at every level. Platforms tweak their features to tap monetizable eyeballs. Influencers receive attention, money, and access. The algorithm rewards the person who spots the resonant incident, welds it to an existing frame, and riles up the base. Politicians receive loyalty, votes, and an issue placed on the agenda. Ordinary participants receive belonging, identity, and the status that comes from fighting for their side loudly and well. This happens on the right as much as on the left.

Williams describes these political groups as tribes. I prefer factions. Tribe suggests an inherited identity or an innate alignment. Faction emphasizes participation – what they do. Members don’t simply hold an interpretation; they circulate it, perform it, defend it, and enforce it against defectors. A faction has authorities, aspirants, and foot soldiers, and its account of reality is continually rebuilt through their interaction.

Returning to Springfield

Let’s go back to the cat—or rather, the cats.

Springfield started as a local rumor, until foot soldiers supplied the memes, aspirants amplified them, and the faction’s authorities—in this case, Vance and the influencers—supplied the broader meaning.

One of the first viral posts came from the large right-wing account End Wokeness. It placed the screenshot from the Springfield Facebook group alongside an unrelated photograph of a man carrying a dead goose and claimed that pets and ducks were disappearing after 20,000 Haitians had been “shipped” to Springfield. The post received millions of views. Within days, major right-wing accounts and Republican politicians—including Ted Cruz, Jim Jordan, Marjorie Taylor Greene, and Kari Lake—had amplified the broader story.

On September 9, Vance posted that pets had been “abducted and eaten by people who shouldn’t be in this country.” His own staff had already been told by Springfield’s city manager that the claims were baseless. He left the post up. The next day he wrote that it was “possible, of course, that all of these rumors will turn out to be false”—a slight hedge—then told his followers to “keep the cat memes flowing.” That night Donald Trump repeated the claim from the debate stage: “They’re eating the dogs. They’re eating the cats.”

Then people began trying to prove it. Christopher Rufo offered a $5,000 bounty for “hard, verifiable evidence”. YouTubers posted videos mixing interviews in Springfield with AI-generated images and unrelated footage. The Vance campaign kept looking for supporting evidence; a spokesman gave The Wall Street Journal a police report from another Springfield woman who had suspected her Haitian neighbors when her cat, Miss Sassy, disappeared. When a reporter visited her, she said the cat had been found safe in her basement. She apologized to her neighbors.

Springfield, meanwhile, endured more than thirty bomb threats against schools, government buildings, hospitals, and officials’ homes.

Vance and Trump were not speaking to a confused public that lacked access to the facts. Corrections had come from law enforcement, Springfield’s Republican mayor, and Ohio’s Republican governor. They were repeated by national news organizations. Miss Sassy had been found alive! A Washington Post poll conducted several weeks later found that most Ohio voters rejected Trump’s claim. Republicans were more divided: 42% said it was true, 29% said it was false, and 26% were unsure.

But “eating the pets” had already made Springfield a national symbol of what Vance and Trump wanted to say about immigration. Its political value did not depend on every voter continuing to believe the literal cat claim.

This is an example of why I’ve long argued that “misinformation” is the wrong frame for what is happening. It focuses the attention on whether a claim is true or false. Sometimes that matters—there are such things as facts, and the pets were not being eaten. The 29% of Republicans who rejected it matter: factual correction was perhaps not futile. But Vance was using the story to advance a larger interpretation: immigration was destroying American communities, the press was ignoring the damage, and he was bravely willing to force the issue into public view.

Corrections can change what people believe without undoing the larger story or conclusions they drew from the false claim—a possibility academics continue to investigate as the continued influence effect and what Emily Thorson calls “belief echoes.”

But Springfield was not simply a case in which a false claim left a residue. Vance had already told his followers to keep circulating the story even if the rumors turned out to be false. He thwn gave them a reason to dismiss the correction without defending the underlying facts. The press, he argued, cared more about proving him wrong than about the suffering immigration had caused. “The media didn’t care about the carnage wrought by these policies until we turned it into a meme about cats,” he said. His followers could give up the cat without giving up Springfield as proof that immigration was ruining American communities.

The Referee Problem

Williams suggests that interpretive polarization might be reduced by improving the reliability and impartiality of the institutions that produce knowledge. He is not naive about this. Institutions can become insular or captured, and even a reliable institution cannot settle disputes if the public considers it corrupt. I agree. But reliability and better communication alone will not restore trust, because distrust is also something political actors actively produce.

When Tulsi Gabbard released declassified documents last year with the headline ‘Fauci Funded Wuhan Lab Research That Sparked COVID,’ the documents themselves showed internal debate and uncertainty—not the conspiracy the headline implied. But the goal wasn’t to win an argument with evidence; it was to undermine the legitimacy of both the institutions that served the public as well as those that could challenge the preferred narrative. The method was straightforward: put the desired conclusion in the headline, release a large collection of documents, and let the framing do the work. Few people would read hundreds of pages closely enough to see the distance between the documents and the interpretation being sold.

This does not mean institutional mistrust is baseless. Institutions sometimes subordinate accuracy to advocacy and hand their critics real material. In June 2020, more than a thousand public-health professionals signed a letter arguing that protests they supported should be treated differently from other large gatherings during the pandemic. Newsrooms and universities have had episodes of ideological signalling and bad judgment.

But those failures did not create the campaign against institutional authority. Organized efforts to discredit inconvenient science and researchers date at least to the tobacco industry. Recent institutional failures did not create that strategy; they gave it new material. Each lapse becomes another reason to argue that experts are political and official investigations cannot be trusted.

Actors who profit from rival realities have an incentive to attack anyone who could authoritatively contradict them. There is working the refs—and then there is the project to eliminate the idea of legitimate referees.

Springfield shows the payoff of that long-term effort. Vance disputed some of the debunkings, but his success did not depend on proving the fact-checkers wrong. He reframed their insistence on verification as evidence of misplaced priorities: they were pedants counting cats while Americans suffered.

Lippmann explained how people can share facts and still live in different worlds. Bernays makes us ask who shaped the picture, how, and for whose benefit. In Springfield, political actors took a rumor they had reason to know was false and made it carry a larger argument about immigration. When officials and journalists challenged it, Vance recast their concern with accuracy as part of the argument.

Institutions cannot answer such a system with a fact-check and a press release. They need their own networks of trusted messengers, relationships built before the next crisis, and people prepared to explain not only what is true but why an event should be understood differently. They—and the people who depend on them—must also recognize certain attacks on institutions as attacks on the possibility of legitimate authority itself, and respond collectively rather than leaving each target to defend itself alone. Rival realities are produced through networked participation. Preserving a shared one requires the same.