The American far right has discovered a convenient new justification for racism and xenophobia: pattern recognition. But peer-reviewed research, terrorism statistics, immigrant crime data, and cognitive psychology expose the intellectual trick. From Muslim terrorism and white mass shooters to American war crimes, the patterns some commentators recognize depend rather suspiciously on who’s being accused.
By Omar Afra
In a 1996 episode of The Simpsons, Homer becomes convinced that Springfield’s expensive new Bear Patrol is working because he hasn’t seen any bears. Lisa attempts to explain the elementary logical failure in his reasoning by picking up an ordinary rock and claiming it keeps tigers away. When Homer points out that there are no tigers around, Lisa agrees. That’s precisely her point. Homer considers the evidence and offers to purchase the rock. The 26-second scene is one of the better demonstrations of defective reasoning ever broadcast on American television.
What makes the episode particularly prophetic is that the story eventually turns into a political campaign against immigrants. Mayor Quimby, confronted with public anger over the cost of the Bear Patrol, redirects the town’s frustrations toward foreigners, whom he conveniently blames for the city’s problems. The writers managed to connect faulty inference, manufactured fear, political opportunism, and xenophobia in approximately twenty minutes of network television. Thirty years later, a substantial corner of the American right has elevated Homer’s understanding of statistical reasoning into something resembling an intellectual movement. They’ve even given it a respectable-sounding name: pattern recognition.
Thirty years later, a substantial corner of the American right has elevated Homer’s understanding of statistical reasoning into something resembling an intellectual movement. They’ve even given it a respectable-sounding name: pattern recognition.
The phrase has become a fashionable means of passing off racism, xenophobia, and religious bigotry as the product of superior observational intelligence. You’re not prejudiced, you see. You’re simply noticing things. And if someone challenges your conclusions, it’s because they’re too cowardly, politically correct, or intellectually compromised to acknowledge uncomfortable truths. It’s a wonderful arrangement. You get to make sweeping pronouncements about millions of people while dispensing with the tiresome business of representative samples, comparative statistics, confounding variables, causal mechanisms, and virtually everything else that distinguishes evidence-based reasoning from the neurological activity of a houseplant.
But there’s something far more revealing about the right’s newfound enthusiasm for pattern recognition than its elementary misunderstanding of how statistical inference works: the remarkable selectivity with which the alleged patterns are recognized. When a Muslim commits an act of terrorism, the self-appointed pattern-recognition brigade suddenly develops an insatiable curiosity about religious affiliation, cultural background, immigration history, and the supposedly inherent characteristics of entire civilizations. One perpetrator can apparently tell us something meaningful about nearly two billion Muslims. When a white American commits a mass shooting, however, the same intellectual pioneers frequently discover the virtues of individual responsibility, psychological complexity, social isolation, and the importance of not generalizing about entire populations. Suddenly, the appropriate unit of analysis is the individual. Statistical enthusiasm gives way to a touching appreciation for nuance.
The disparity becomes even more conspicuous when examining the conduct of Western governments. Why doesn’t the same obsessive pattern recognition activate when examining centuries of European colonial conquest, American military interventions, torture programs, civilian massacres, or the repeated destruction of civilian infrastructure during modern warfare? Why aren’t atrocities committed by Western militaries treated as revelations about the inherent characteristics of Christianity, whiteness, or Western civilization? Why does the analytical lens expand to encompass entire populations when the perpetrator is Muslim, yet contract to the individual, military command, or particular historical circumstances when the perpetrators are Western?
To be clear, collective guilt would be no more defensible in those cases. That’s precisely the point. The inconsistency exposes a methodology that is considerably more interested in confirming ideological preferences than discovering universal principles. It’s not that these people recognize patterns particularly well. It’s that they’ve developed an extraordinarily selective appetite for recognizing the ones that flatter their existing prejudices. And it isn’t merely that they’re mistaking correlation for causation. In many cases, they haven’t even established a meaningful correlation. They’ve selected examples, stripped away the denominator, ignored contradictory evidence, and announced that their conclusions are too obvious to require investigation. The resulting intellectual performance combines the analytical rigor of an eighth-grade science project with the absolute certainty of a medieval inquisitor.
When Racism Discovers a More Respectable Vocabulary
On August 25, 2026, far-right commentator Laura Loomer offered an unusually candid demonstration of the phenomenon during an episode of Loomer Unleashed. Discussing several Black public figures, including Jasmine Crockett, Ayanna Pressley, Cori Bush, Ilhan Omar, and Candace Owens, Loomer suggested their behavior represented a broader racial pattern. She then declared, “Some people call it racism. I call it pattern recognition.” Loomer continued by asserting that pattern recognition and profiling keep people safe. The remarks were documented by Media Matters, which published the relevant video and transcript on August 26.
There’s something magnificently stupid about the exercise. Loomer assembled a list of Black public figures whose politics, careers, constituencies, and ideological positions vary considerably. Some are Democrats. Candace Owens, for example, built much of her public identity as a conservative commentator. The supposed pattern appears to depend less on any consistently defined behavior than on Loomer’s ability to identify people she dislikes and locate a demographic characteristic they share. This is not statistical analysis. It’s a guest list assembled by a woman who has mistaken her personal grievances for a research methodology.
What makes the statement significant, however, is not Loomer’s particular intellectual limitations. It’s the rhetorical maneuver. By invoking pattern recognition, she reframes a racial generalization as an act of empirical observation. The burden of proof quietly changes hands. Instead of demonstrating that a meaningful relationship exists between racial identity and some clearly defined behavior, she implies that those who question her interpretation are refusing to acknowledge an obvious truth. A person who notices something is not necessarily a person who understands what they’ve noticed. But the modern racial-grievance economy has discovered that the distinction can be obscured by borrowing terminology from science.
The phrase itself is older than Loomer’s August declaration. Earlier online discussions contain similar defenses of stereotyping, while recent racist memes explicitly invite audiences to recognize supposedly forbidden patterns in racial behavior. The precise point at which the expression entered far-right vocabulary is difficult to establish, but its movement from online subcultures into high-profile political commentary is no longer particularly subtle. Researchers have now examined the machinery behind it, and the results are considerably more interesting than the amateur statistical theater performed by its loudest practitioners.
Cambridge Researchers Study the New Scientific Racism
In June 2026, Sarah Rodriguez-Louette of Sorbonne Nouvelle University published Memeing Scientific Racism: The Digital Reframing of Racialist Ideologies in Cambridge University Press’s peer-reviewed journal Evolutionary Human Sciences. The study examined 68 racialized memes collected from X between January and March 2025, alongside four detailed case studies. Its sample was selected to investigate how racist ideas were packaged and circulated, rather than to estimate the overall prevalence of racist content across the platform. That methodological distinction matters because, unlike the people it studies, the research actually explains what its evidence can and cannot establish.
Rodriguez-Louette identifies a recurring strategy in which older ideas about racial hierarchy are repackaged using scientific terminology, demographic charts, intelligence comparisons, and visual formats associated with objective analysis. The memes encourage viewers to move rapidly from observing apparent differences to accepting explanations about supposedly inherent group characteristics. One example is particularly revealing. The study describes a meme contrasting a crow being praised for recognizing patterns with a white man being condemned for recognizing patterns in racial crime statistics. The implied argument is that society celebrates pattern recognition in animals while irrationally punishing humans for making observations about race.
The joke depends on an enormous intellectual sleight of hand. It treats recognizing a possible association, interpreting that association, and establishing its cause as though they were interchangeable activities. It then portrays anyone who challenges the resulting conclusion as an enemy of observation itself. A legitimate statistical association may exist without supporting the explanation someone attaches to it. A photograph may accurately document a particular event without establishing how frequently comparable events occur. A demographic disparity may be measurable without revealing whether its causes are economic, environmental, institutional, historical, behavioral, or some combination of those factors.
The memes examined in the Cambridge research largely dispense with those complications. They use contrasts, inflammatory screenshots, mugshots, simplified charts, and repeated stereotypes to create the appearance of empirical legitimacy. They encourage the viewer to supply the conclusion rather than requiring the content creator to defend it. Rodriguez-Louette also identifies another important feature: the transformation of older racial theories into ostensibly neutral language about intelligence, competence, behavior, and social outcomes. Where nineteenth-century scientific racism relied heavily on physical measurements and racial classifications, contemporary racialist content frequently presents selected behavioral examples as evidence of stable group differences.
Where nineteenth-century scientific racism relied heavily on physical measurements and racial classifications, contemporary racialist content frequently presents selected behavioral examples as evidence of stable group differences.
The scientific vocabulary has been updated. The underlying attempt to turn human variation into racial destiny remains recognizable. This is worth remembering whenever modern political movements invoke the presumed intellectual superiority of Western civilization while resurrecting some of its most discredited ideas. As Testset recently explored in its examination of the Crusades and contemporary anti-Muslim historical revisionism, the Western supremacist imagination has an impressive capacity to turn historical ignorance into civilizational self-congratulation.
And the internet has made the operation cheaper. A century ago, producing pseudoscientific racial propaganda required elaborate publications, institutions, and at least the pretense of sustained research. Today, an aspiring racial theorist can assemble a few screenshots, slap a chart over them, add a caption about noticing patterns, and distribute the resulting intellectual turd to an audience of thousands. The Cambridge research helps explain why this works: its subjects often produce the appearance of evidence without actually presenting an adequate argument. Viewers encounter a conclusion in visual form before they have considered the information needed to evaluate it.
Psychology Identified This Problem Fifty Years Ago
The truly embarrassing aspect of this intellectual fashion is that psychologists were studying precisely these errors long before anyone could attach them to a racialized meme. In 1976, David Hamilton and Robert Gifford published Illusory Correlation in Interpersonal Perception: A Cognitive Basis of Stereotypic Judgments in the Journal of Experimental Social Psychology. Their experiments demonstrated that people can perceive relationships between group membership and negative behavior even when the information presented does not support the relationship they infer. One mechanism involves the salience of unusual combinations: membership in a smaller group and an infrequent negative behavior are both distinctive, making their co-occurrence particularly memorable.
The result is a tendency to overestimate the association between minority-group membership and undesirable behavior. Consider the implications. People can perceive a demographic pattern where no meaningful relationship exists because their attention and memory disproportionately emphasize particular combinations of events. You don’t even need a sophisticated political propaganda machine to manufacture the error. Ordinary human cognition is already vulnerable to it.
In 2002, researchers Caterina Primi and Franca Agnoli investigated the phenomenon among children in first, third, fifth, and seventh grades, as well as adults. Their experiments found that participants could infer an association between minority-group membership and negative behavior despite the experimental information containing no actual relationship between the two. The researchers also reproduced similar effects using abstract shapes and colors rather than human social groups. The underlying problem was not confined to explicitly racial thinking. It reflected a more general vulnerability in human judgment, documented in their paper Children Correlate Infrequent Behaviors with Minority Groups: A Case of Illusory Correlation.
There is something wonderfully appropriate about the fact that children were already being studied for this cognitive error decades before right-wing influencers began advertising it as a sophisticated political philosophy. The research also provides a useful correction to the notion that the problem is merely low intelligence. These errors can affect intelligent people. Recognizing their existence is precisely why disciplined methods matter. The relevant distinction isn’t between smart people who notice patterns and stupid people who don’t. It’s between people willing to examine their interpretations and people who treat those interpretations as self-validating.
Another foundational contribution came from psychologist Thomas Pettigrew’s 1979 paper, The Ultimate Attribution Error: Extending Allport’s Cognitive Analysis of Prejudice. Pettigrew described how prejudiced observers may explain negative behavior differently depending on whether the person responsible belongs to their own group or another group. Negative conduct among members of an out-group is more readily attributed to supposedly enduring group characteristics, while similar conduct among in-group members receives more situational or individualized explanations. Positive behavior can be treated asymmetrically as well. An admirable member of a disliked group becomes an exception, while comparable behavior among members of the favored group confirms the group’s virtues.
A Muslim commits violence, and Islam becomes the explanation. A white American commits violence, and we investigate the individual’s personal history.
Sound familiar? A Muslim commits violence, and Islam becomes the explanation. A white American commits violence, and we investigate the individual’s personal history. An immigrant commits a crime, and immigration becomes the explanation. A native-born American commits the same crime, and we discuss personal responsibility. The modern pattern-recognition racket is not merely reproducing a familiar cognitive error. It is turning that error into a badge of political sophistication.
Why Doesn’t Pattern Recognition Work on School Shooters?
Let’s apply the same methodology to American mass shootings. The Violence Prevention Project maintains a detailed database of public mass shootings in the United States dating to 1966. Its research reports that approximately 98% of the perpetrators in its database are male and 64% are white. The researchers also identify recurring characteristics involving personal crises, prior warning signs, trauma, and access to firearms. The database’s definition of a public mass shooting is narrower than definitions used by some other organizations, and these findings should not be confused with statistics covering every shooting on school grounds. Nevertheless, the demographic distribution is plainly measurable.
Now imagine taking these figures and applying the same undisciplined reasoning used in racist pattern-recognition memes. One could compile photographs of white male mass shooters, organize them into a visually striking montage, highlight their race and gender, and declare that the resulting image reveals something inherent about white men. One might even produce a dramatic graphic with a bald eagle, several assault rifles, and the words Notice the Pattern. But what, exactly, would that establish? Without properly comparing relevant population rates, exposure, social conditions, access to weapons, motivations, and other explanatory variables, the montage would not establish that whiteness causes mass violence. Even a statistically significant demographic difference would not automatically identify its cause.
And it certainly would not justify treating every white American as a potential mass murderer. That would be intellectually irresponsible, which is precisely why it is so revealing when people who enthusiastically make demographic generalizations about Muslims or immigrants suddenly recognize the dangers of collective attribution when the demographic category changes. School shooters are suddenly individuals with biographies, traumatic experiences, psychological problems, family circumstances, and social environments. Their behavior is not automatically interpreted as an indictment of every person who resembles them.
There is another complication. The familiar assertion that media coverage automatically describes white attackers as mentally ill while describing Muslim attackers as terrorists is itself too simplistic to survive careful scrutiny. Research has found a substantial difference in the application of the terrorism label, but one major study did not establish an equivalent pattern in references to mental illness based solely on whether the perpetrator was white. That distinction is important. An article criticizing bad statistical inference should not manufacture its own convenient statistical inference, particularly when the actual findings are damning enough without embellishment.
Muslim Violence Gets a Different Vocabulary
In 2019, researchers Erin Kearns, Allison Betus, and Anthony Lemieux published Why Do Some Terrorist Attacks Receive More Media Attention Than Others? in Justice Quarterly. The researchers examined coverage of 136 terrorist attacks in the United States between 2006 and 2015. After controlling for factors including fatalities, target type, and whether perpetrators were arrested, they found that attacks involving Muslim perpetrators received, on average, 357% more media coverage than attacks involving non-Muslim perpetrators.
The result is remarkable because it concerns the very information environment through which audiences develop impressions of how frequently different threats occur. If one category of violent incident receives substantially more coverage than another, the public is not being presented with a representative distribution of incidents. People are being exposed to a filtered account of reality in which certain kinds of perpetrators become disproportionately visible. A person consuming that coverage may sincerely believe they’re recognizing a pattern. What they’re recognizing, at least partly, is a disparity in what has been placed before them.
The same research team pursued a related question in How Perpetrator Identity (Sometimes) Influences Media Framing Attacks as “Terrorism” or “Mental Illness”, published in Communication Research. Examining print coverage of U.S. terrorist attacks between 2006 and 2015, the researchers found that, after accounting for relevant factors, the odds that an article invoked terrorism were approximately five times greater for a Muslim perpetrator than for a non-Muslim perpetrator. Importantly, the study did not find that white perpetrators were systematically more likely to receive mental-illness framing simply because they were white. The asymmetry in terrorism terminology was much clearer than the popular claim about mental illness.
These are related studies by overlapping researchers examining the same historical period, not two wholly independent confirmations. But together they offer substantial evidence of a measurable disparity in how different perpetrators become represented to the public. It is worth considering what happens when this information enters a political ecosystem already predisposed to associate Islam with violence. Muslim perpetrators receive disproportionate coverage and are more likely to be described through the vocabulary of terrorism. Audiences repeatedly encounter those associations. Some begin to perceive the resulting familiarity as evidence that the association is unusually prevalent.
The self-appointed pattern-recognition enthusiast then announces that the media is concealing the very truth the media’s uneven coverage has helped him believe. An entire political worldview can emerge from a feedback loop that nobody involved has bothered to examine. The distinction between observing an event and understanding its statistical significance has disappeared, replaced by the sensation of having discovered something everybody else is supposedly too frightened to acknowledge.
The New Orleans Attack: When the Pattern Is So Obvious It Doesn’t Have to Be True
The January 1, 2025, terrorist attack in New Orleans provides a particularly instructive example of how preconceived narratives can outrun the facts. Shamsud-Din Jabbar drove a truck into crowds celebrating the new year on Bourbon Street, killing 14 people. Investigators determined that he had expressed support for the Islamic State. The attack was an act of terrorism, and its victims deserved justice. But Jabbar was also a U.S.-born citizen, an Army veteran, and a Texas native.
Nevertheless, an erroneous early Fox News report helped fuel speculation that the attacker had recently crossed the southern border. Donald Trump and several Republican figures made statements linking the attack to immigration despite the absence of evidence supporting that connection. FactCheck.org documented how Republican politicians continued promoting or implying the immigration narrative even after the relevant information had been corrected.
Consider what happened intellectually. A terrorist attack occurred. The attacker had a Muslim name and expressed allegiance to ISIS. Before his biography had been adequately established, the incident was incorporated into an existing political narrative about immigrants and border security. The supposed pattern was so compelling to its proponents that it survived the inconvenient discovery that the perpetrator was born in the United States and had served in the American military. He was, in other words, an American terrorist. But the political machinery wanted an immigrant terrorist.
The difference matters enormously if one’s stated purpose is understanding why violence occurs. It matters considerably less if the purpose is attaching a frightening incident to a population that has already been designated as politically useful to fear. The observed event was real. The explanation attached to it was demonstrably wrong. And rather than being corrected by new evidence, the surrounding political narrative proved remarkably resistant to the basic biographical facts of the case.
When Immigrant Crime Statistics Refuse to Cooperate
The immigration example becomes even more interesting when we examine actual comparative crime data. In 2020, researchers Michael Light, Jingying He, and Jason Robey published Comparing Crime Rates Between Undocumented Immigrants, Legal Immigrants, and Native-Born US Citizens in Texas in the Proceedings of the National Academy of Sciences. Using Texas Department of Public Safety data covering 2012 through 2018, they found that undocumented immigrants had substantially lower felony arrest rates than native-born citizens across multiple offense categories.
During the study period, U.S.-born citizens were more than twice as likely as undocumented immigrants to be arrested for violent crimes, approximately 2.5 times as likely to be arrested for drug offenses, and more than four times as likely to be arrested for property crimes. The researchers also conducted additional checks using alternative population estimates and classifications. This is an actual statistical investigation of the relationship between immigration status and criminal behavior. It identifies its population, measures comparative rates, examines multiple categories, and tests whether its findings survive alternative assumptions.
Notice how little it resembles a montage of immigrant mugshots. The findings do not mean that immigrants never commit crimes, that every immigrant community has identical outcomes, or that arrest statistics perfectly measure underlying criminal behavior. They do mean that the broad assertion that undocumented immigrants represent an inherently greater criminal threat was not supported by this Texas evidence. And Texas is particularly useful here because its administrative records provide information about immigration status that is unavailable in many other states.
The broader historical evidence points in a similar direction. A 2024 study published in American Economic Review: Insights examined incarceration rates among immigrants and U.S.-born Americans from 1870 through 2020. The researchers found that immigrants, as a group, had lower incarceration rates than the native-born population throughout the 150-year period. In contemporary data, immigrants were approximately 60% less likely to be incarcerated than U.S.-born Americans.
Incarceration is not a perfect substitute for crime rates, and the researchers were careful about what their measures represented. But their nationally representative historical analysis provides a rather formidable challenge to sweeping narratives that associate immigration with criminality. A separate National Institute of Justice-funded research report examining immigration and crime across ten jurisdictions found no statistically significant association between estimated unauthorized immigrant populations and arrest rates after accounting for relevant factors and conducting additional statistical checks.
If the contemporary right’s pattern-recognition enthusiasts are genuinely committed to following empirical evidence wherever it leads, these findings should be central to their understanding of immigration and public safety. Instead, political discussion frequently revolves around particular incidents involving immigrant perpetrators, with those incidents presented as representative of a much larger population. The trick-fuck is not necessarily inventing the crimes. The crimes can be entirely real. The trick-fuck is presenting selected incidents as evidence of a population-level danger while disregarding the population-level data needed to establish that danger. A person can accurately describe every incident in a collection and still be profoundly wrong about what that collection means.
Texas: Where Ordinary Muslim Life Becomes a Suspicious Pattern
Texas has become an especially revealing laboratory for this selective form of reasoning. The state’s ongoing anti-Sharia political campaign regularly treats ordinary Muslim civic activity as evidence of something sinister. Mosques, schools, community developments, religious organizations, and civil-rights advocacy can become objects of suspicion because they are organized around Islam. As documented in Testset’s Texas Islamophobia and Sharia Panic dossier, the political narrative often depends on turning otherwise ordinary forms of religious participation into evidence of an alleged threat to American institutions.
A Christian congregation builds a large religious complex, and the project is understood as community development. A Muslim organization proposes a residential development with religious and educational facilities, and politicians demand investigations into whether the project represents a threat to the Constitution. The underlying conduct can be strikingly similar. The interpretation changes when the religion changes.
In November 2025, Governor Greg Abbott announced that he was designating the Council on American-Islamic Relations and the Muslim Brotherhood as foreign terrorist organizations under Texas authority. CAIR rejected the designation, while the action raised questions about the limits of state authority and the distinction between state-level actions and federal foreign-terrorist designations. Abbott’s official announcement asserted that the organizations sought to impose Sharia law and subvert American legal institutions.
By May 2026, the U.S. House Judiciary Subcommittee on the Constitution and Limited Government was holding a hearing titled Sharia-Free America: Why Political Islam & Sharia Law Are Incompatible with the U.S. Constitution: Part II. The committee’s announcement described the hearing as an examination of purported threats arising from Sharia-based institutions and proposed legislative responses. The hearing was part of an ongoing political dispute over the relationship between Islamic religious practices, political Islam, and American constitutional law.
The constitutional question is not whether American law should govern American institutions. Of course it should. The question is whether Muslim religious activity is being subjected to presumptions of disloyalty and danger that would be considered intolerable if routinely applied to other religious communities. Religious institutions may legitimately be investigated when there is evidence of wrongdoing. But religious identity is not itself evidence of wrongdoing, and observing multiple Muslims organizing around their faith is no more suspicious than observing multiple Baptists doing the same.
Yet the pattern-recognition enthusiast can transform almost any expression of Muslim civic normalcy into a warning sign. Muslims establish schools? They’re organizing. Muslims purchase property? They’re establishing enclaves. Muslims participate in elections? They’re seeking political influence. Muslims challenge government actions in court? They’re deploying institutional power. These are ordinary activities in a constitutional democracy. Their supposed menace depends on the identity of the people performing them.
What gets presented as pattern recognition is often simply differential treatment with a more elaborate vocabulary. It is the same theatrical suspicion that animates so much of the Texas Sharia panic documented by Testset, in which religious freedom remains a sacred American principle right up until the people exercising it start using Arabic words. Then some politician in an expensive cowboy hat begins behaving as though the Constitution is about to be stolen by a halal food truck.
Where Does Pattern Recognition Go When Western Governments Commit War Crimes?
Now we arrive at the most revealing asymmetry of all. The same political culture that frequently treats individual acts of violence by Muslims as evidence of broader civilizational defects can become remarkably sophisticated about context, national security, historical necessity, and the complexity of warfare when examining violence committed by Western states. Here the evidence isn’t limited to a collection of inflammatory social media clips.
The United States Senate Select Committee on Intelligence released a major investigation in December 2014 documenting the CIA’s post-9/11 detention and interrogation program. The committee found that coercive interrogation techniques had been far more brutal than publicly represented and had not produced the unique intelligence benefits claimed by the agency. The investigation examined extensive official records and identified failures involving detention, interrogation, oversight, and the accuracy of information provided to policymakers. The committee’s findings were contested by defenders of the program, but the documentary record established a substantial basis for institutional scrutiny.
If we were following the logic of demographic pattern recognition, why wouldn’t repeated instances of torture by American intelligence personnel become evidence of an inherent defect in American culture? The answer is that such an inference would be absurd. These were actions undertaken through particular institutions, under particular policies, by people operating within identifiable systems of authority. Understanding responsibility requires examining those structures, the decisions that produced them, and the individuals who participated. The fact that some perpetrators were American does not make every American responsible.
That analytical discipline is correct. It is also precisely the discipline that disappears when the discussion turns to Muslims.
The enormous human cost of American military interventions provides another example. Brown University’s Costs of War project estimates that the post-9/11 wars in Afghanistan, Iraq, Pakistan, Syria, Yemen, and related theaters have resulted in approximately 4.5 million to 4.7 million direct and indirect deaths. The estimate includes deaths caused directly by fighting as well as indirect deaths associated with the destruction of infrastructure, public health systems, economies, and essential services. It encompasses the consequences of wars involving multiple governments and armed organizations, not deaths attributable solely to U.S. forces.
The scale is staggering. But the appropriate response is to investigate military policy, command responsibility, strategic decisions, violations of international law, and the broader consequences of warfare. It would not be reasonable to conclude that Americans, Christians, or white people possess an inherent propensity for mass killing. Now consider Israel’s military campaign in Gaza, where the distinction between individual responsibility, state policy, and collective blame is no less important.
The International Criminal Court issued arrest warrants in November 2024 for Israeli Prime Minister Benjamin Netanyahu and former Defense Minister Yoav Gallant, alleging responsibility for war crimes and crimes against humanity, including starvation as a method of warfare and attacks against civilians. These are allegations underlying arrest warrants, not criminal convictions. The court also issued a warrant for a Hamas military commander over alleged atrocities committed during the October 7 attacks; proceedings against that commander were subsequently terminated following confirmation of his death.
Human Rights Watch separately documented extensive forced displacement and destruction in Gaza, concluding that Israeli authorities had committed war crimes and crimes against humanity. Israel has rejected major international accusations concerning its conduct, while the underlying allegations remain central to international legal and political disputes. The relevant subjects of investigation are identifiable political leaders, military commanders, government policies, operational decisions, and institutional practices.
No intellectually responsible person should use these findings to assert that Jews as a religious or ethnic population are inherently violent. Nor should anyone use American military crimes to make equivalent declarations about Christians. The point is not that one side should adopt the same prejudicial reasoning against another population. The point is that the reasoning is indefensible regardless of the population involved.
Yet when some Western political commentators discuss violence committed by Muslim extremists, the distinction between perpetrators, institutions, ideologies, and entire religious populations often collapses. Islam becomes the explanatory category. Muslim identity becomes the relevant variable. The actions of particular organizations become reflections of a much larger civilization. When examining the conduct of Western states, however, explanations often become more specific. Political leaders made decisions. Intelligence agencies adopted policies. Military commanders followed or violated orders. Governments pursued strategic objectives.
Suddenly, causality matters. The same methodology should apply in both cases, and this is where the hypocrisy becomes impossible to ignore. The problem isn’t simply that some right-wing commentators fail to recognize patterns in Western violence. It’s that many of the contextual distinctions they demand when examining Western conduct are precisely the distinctions they refuse to extend to populations they already regard with suspicion. The scientific method does not contain a provision allowing you to change the rules depending on which civilization you like.
Even the Terrorism Statistics Complicate the Story
The asymmetry becomes harder to defend when we examine domestic terrorism using consistent definitions. In September 2025, the Center for Strategic and International Studies published an analysis of ideological terrorism in the United States, drawing on a dataset of 750 terrorist attacks and plots from January 1994 through July 4, 2025.
The researchers found that, over the preceding decade, right-wing terrorist attacks had killed approximately 112 victims, compared with 82 killed by jihadist attacks and 13 by left-wing attacks. The study also documented a rise in left-wing attacks and plots during 2025, although their lethality remained comparatively low. Its findings varied by time period, demonstrating why the selection of a particular window matters when comparing different forms of political violence.
These figures do not establish that conservatives as a population are terrorists. They describe violence associated with particular extremist ideologies under the researchers’ definition of terrorism. But that’s precisely what makes them relevant. If ideological association and repeated violent conduct justify collective suspicion toward Muslims, why wouldn’t the same approach justify collective suspicion toward conservatives?
Because it would be ridiculous. Millions of people can share broad political or religious identities without sharing the extremist commitments of violent individuals who claim some relationship to those identities. The same distinction protects ordinary conservatives from being collectively blamed for white supremacist terrorism and ordinary Muslims from being collectively blamed for jihadist terrorism.
It’s a remarkably straightforward principle, and somehow, for a certain category of political commentator, it only seems to become intellectually available when the population under examination is their own.
Your Algorithm Is Not a Representative Sample of Humanity
The contemporary version of this problem has acquired an additional complication: social media recommendation systems. Imagine spending six months following accounts devoted exclusively to crimes committed by immigrants. Every day, another video appears. Another assault, another robbery, another horrifying incident accompanied by inflammatory commentary. You collect the examples, share the clips, and begin to develop the impression that immigrant criminality represents an unusually pervasive problem.
But your experience does not establish the prevalence of immigrant crime. It establishes that you have been repeatedly exposed to selected examples of immigrant crime. You don’t know how many comparable crimes were committed by native-born Americans. You don’t know the relative population sizes or comparative rates. You don’t know whether the incidents were selected systematically or whether the accounts deliberately exclude contradictory information. You may not even know whether every incident has been accurately described.
Nevertheless, the collection feels compelling because the examples are real enough, numerous enough, and emotionally powerful enough to create the impression of overwhelming evidence. This is where the Cambridge research becomes especially useful. Rodriguez-Louette describes how racialized memes exploit familiar cognitive shortcuts while operating within platforms that reward repeated exposure, emotionally provocative material, and recognizable visual formats. Screenshots and mugshots acquire the appearance of documentary authority even when the selection of material is profoundly unrepresentative.
The result is an information environment in which repetition can become confused with confirmation. Other research provides empirical support for the role of recommendation systems in amplifying politically divisive material. A 2025 study published in PNAS Nexus used a preregistered audit to examine Twitter’s engagement-based ranking system. Compared with a reverse-chronological feed, the engagement-based system amplified emotionally charged and politically hostile content, including material attacking opposing groups. The researchers also found a mismatch between the content selected by the algorithm and what users said they actually preferred to encounter.
In May 2026, researchers published another experiment in Nature involving approximately 2,000 participants assigned to different custom social-media feeds over eight weeks surrounding the 2024 presidential election. The researchers found that engagement-based ranking amplified moralized, emotional, and hostile content relative to chronological ranking. It also increased perceptions of hostility between political groups. Importantly, the experiment used custom-designed feeds rather than claiming to measure every commercial platform’s current recommendation system. Its value lies in demonstrating how ranking choices themselves can alter the information environment and perceptions of other groups.
These findings connect directly to the pattern-recognition phenomenon. A person scrolling through an algorithmically curated feed isn’t observing a random sample of human behavior. They’re observing a selection influenced by their previous interactions, the incentives of content producers, and the platform’s ranking mechanisms. The feed may contain authentic incidents while still producing a deeply distorted sense of how common those incidents are.
As Testset previously examined in Why Does the Right Love AI Slop So Much?, the modern political information environment increasingly supplies material designed to reinforce existing expectations rather than challenge them. Synthetic imagery adds another layer of distortion, but the fundamental problem does not require artificial intelligence. Selective exposure was capable of manufacturing false impressions long before generative tools arrived. What AI contributes is the ability to produce endless additional material that looks consistent with the narrative.
At that point, the person proudly announcing that they’ve developed superior pattern-recognition abilities may simply be recognizing the preferences of the system feeding them information. They’ve confused their algorithm with a representative sample of reality, and they’re calling the resulting confusion intellectual courage.
The Difference Between a Pattern and an Explanation
We should be careful not to replace one intellectual error with another. Pattern recognition is not inherently unscientific. It is an essential part of cognition and an indispensable component of many legitimate analytical methods. Nor are statistical disparities automatically meaningless. Different populations can have different measured outcomes in education, health, employment, crime, income, and countless other fields. Researchers examine those differences because understanding them matters.
Some disparities are substantial. Some persist even after researchers account for commonly examined variables. Others disappear, shrink, or change direction when the comparison becomes more rigorous. The existence of a measurable relationship does not by itself establish the mechanism producing it. And even when a relationship has predictive usefulness, that does not mean it justifies broad moral judgments about individuals who share a particular demographic characteristic.
Consider an ordinary medical example. Every time I’ve developed strep throat, I’ve consumed water. The association is completely consistent across my personal experience. Nevertheless, drinking water is not what causes strep throat. The relevant organism is group A Streptococcus, and the fact that I drink water while also occasionally developing bacterial infections reveals nothing meaningful about the relationship between the two. The apparent pattern tells us considerably less than it seems to.
The same distinction becomes far more consequential when people begin making claims about racial groups, immigrants, or religious populations. A collection of incidents cannot establish relative risk without an appropriate comparison group. A disparity in outcomes does not establish its cause. A correlation does not necessarily demonstrate that changing one variable will change another. And no amount of confidence can compensate for inadequate evidence.
These are not elaborate prohibitions invented by progressive academics to prevent conservatives from discussing uncomfortable subjects. They are elementary principles of statistical inference. A serious researcher who identifies a demographic difference should be able to explain what was measured, how the population was selected, how the comparison was conducted, what alternative explanations were considered, and which conclusions the evidence actually supports.
The same standards should apply whether the subject is immigrant crime, white supremacist violence, Islamist extremism, American war crimes, or the behavior of politically connected corporations. If the standards change when the population changes, the methodology is no longer neutral.
The Political Convenience of Selective Intelligence
The extraordinary usefulness of pattern-recognition rhetoric lies in its ability to reverse the relationship between prejudice and skepticism. Someone makes a sweeping negative claim about an entire population. Another person asks for representative evidence demonstrating that the claim is accurate. Rather than provide the evidence, the original speaker announces that the critic is refusing to recognize obvious patterns.
The demand for statistical rigor becomes proof of cowardice. The request for context becomes evidence of ideological contamination. Skepticism becomes denial. This is an impressive piece of rhetorical engineering because it allows people to retain the confidence associated with empirical knowledge without accepting the obligations that make empirical knowledge reliable.
The Cambridge study identifies this maneuver in the material it examines. By presenting racial generalizations as supposedly self-evident observations, the memes transform objections to their conclusions into evidence of social suppression. The person making the claim becomes a courageous truth-teller. The person asking whether the claim is actually true becomes a censor.
There is another advantage. Selective pattern recognition allows the observer to choose which categories are meaningful. A Muslim terrorist can be classified by religion. A white supremacist terrorist can be classified by psychological history. An immigrant criminal can be classified by immigration status. A native-born criminal can be classified by individual circumstance. All four classifications may contain accurate information. But selecting different explanatory categories depending on the identity of the perpetrator makes the resulting comparisons profoundly misleading.
One cannot reasonably compare the inherent characteristics of an entire religious population with the personal circumstances of selected individuals from another population. Yet this is precisely the kind of asymmetrical reasoning that makes racialized political narratives so durable. A disliked group becomes the explanation for the conduct of its worst members. A favored group becomes a collection of individuals whose misconduct requires careful contextual analysis.
Good behavior by a disliked group’s members can be dismissed as exceptional. Bad behavior by the favored group’s members can be dismissed as unrepresentative. The arrangement is practically unfalsifiable. Every incident that supports the prejudice counts as evidence. Every incident that contradicts it becomes an exception. Every demand for better data becomes an ideological attack. And the whole little circus proceeds under the impressive banner of pattern recognition.
The Eighth-Grade Intellectual Revolution
The most amusing element of this supposed intellectual awakening is the extraordinary self-regard of its participants. Some of its most enthusiastic proponents present themselves as fearless investigators willing to confront realities that journalists, academics, and ordinary citizens supposedly lack the courage to acknowledge. Yet the signature method frequently consists of assembling selected examples, omitting relevant denominators, disregarding competing explanations, and declaring the resulting interpretation self-evident.
The absence of rigorous analysis is presented as intellectual liberation. The refusal to examine alternative explanations becomes independence from institutional orthodoxy. And confidence substitutes for the work ordinarily required to determine whether a conclusion is correct. It is a peculiar form of pseudo-intellectualism because its practitioners have adopted the vocabulary of science while rejecting many of the safeguards that make scientific reasoning useful.
They want the prestige of objective observation without the inconvenience of objective investigation. They want the authority of statistics without the burden of statistical literacy. They want the satisfaction of being right without the possibility of discovering they’re wrong. Most revealingly, they want to apply sweeping demographic explanations to people they already dislike while reserving individualized explanations for themselves.
That is not a discovery about human behavior. It is a demonstration of how prejudice organizes attention. There are certainly serious conservatives who engage with crime statistics, immigration research, terrorism data, and the complexities of causation. The errors examined here are neither universal among conservatives nor exclusive to the political right. But within the far-right rhetorical ecosystem that has embraced pattern recognition as a justification for demographic suspicion, these failures are not incidental. They are central to the argument’s political usefulness.
The objective is not simply to identify recurring phenomena. It is to establish a permission structure in which selected observations can justify conclusions that would otherwise require a much more demanding evidentiary defense. The vocabulary gives old prejudices a new costume, and the costume is doing a great deal of work.
Congratulations, You’ve Recognized a Pattern. Now Finish the Assignment.
The entire enterprise depends on confusing the beginning of an investigation with its conclusion. Recognizing that two things appear together is not the same as establishing why they appear together. Collecting examples is not the same as measuring prevalence. Identifying a disparity is not the same as explaining its cause. And selecting which populations deserve collective suspicion based on one’s political preferences is not statistical sophistication. It is prejudice.
The genuinely intelligent response to an apparent pattern is curiosity disciplined by skepticism. What evidence supports the observation? What might be missing? How does the rate compare with an appropriate population? Which alternative explanations survive investigation? What would convince us that our original interpretation was wrong? These questions are not obstacles to truth. They are how we distinguish truth from whatever bullshit happens to flatter us.
Which brings us back to Homer Simpson and his tiger-repelling rock. Homer sees an absence of tigers and accepts an explanation that attributes it to an ordinary stone. Lisa understands that the observation does not establish the proposed cause. Thirty years later, a collection of political commentators has effectively adopted Homer’s position while congratulating itself for being smarter than Lisa.
But the real lesson of that episode extends beyond its famous joke about faulty reasoning. Springfield eventually turns its frustrations toward immigrants because an opportunistic politician discovers that blaming outsiders is easier than explaining the city’s problems. The reasoning is defective. The politics are useful. And that combination is considerably more dangerous than mere stupidity.
The far right has not invented a superior form of reasoning. It has found a way to dress selective observation, elementary statistical errors, and long-established prejudices in the language of scientific inquiry. Its adherents can recognize a pattern in a collection of immigrant crime videos but somehow overlook the comparative crime statistics. They can identify collective characteristics in Muslim extremists but rediscover individual responsibility when confronting violence committed by white Americans. They can attribute entire civilizations’ moral character to particular atrocities while treating the documented violence of Western governments as the product of specific policies and historical circumstances.
The intellectual standards expand and contract with the identity of the accused. That inconsistency is the most important pattern in the entire exercise. And unlike the claims being made about whole populations, this is a pattern we can investigate by examining actual statements, comparative data, and the standards of reasoning used to connect them.
The right’s new pseudo-intellectualism has mistaken the appearance of analytical sophistication for the substance of it. Its practitioners have discovered that saying pattern recognition sounds considerably more intelligent than admitting that they formed an opinion about millions of people after watching several videos on their phones. But a scientific vocabulary does not make an argument scientific, and no matter how enthusiastically you announce that you’ve recognized a pattern, you still have to demonstrate that you understand what it means.
Homer wanted to buy the rock. At least he had the good sense not to start a podcast explaining why everyone else was too stupid to recognize its extraordinary powers.
Research and Methodology
This article draws primarily on Sarah Rodriguez-Louette’s 2026 research, Memeing Scientific Racism: The Digital Reframing of Racialist Ideologies, published in Evolutionary Human Sciences (Volume 8, e27; DOI: 10.1017/ehs.2026.10059). The study’s 68 memes were purposively sampled and should not be interpreted as a representative measure of all conservative online activity. The cognitive-psychology framework draws on Hamilton and Gifford’s 1976 research on illusory correlation, Pettigrew’s 1979 work on intergroup attribution, and Primi and Agnoli’s 2002 experiments involving children and adults. The media analysis relies on two related studies by Kearns, Betus, and Lemieux examining the volume and framing of American terrorism coverage between 2006 and 2015.
The statistical comparisons rely on the Violence Prevention Project’s mass-shooting database; CSIS research on ideological terrorism; the 2020 PNAS study of felony arrest rates in Texas; the 2024 American Economic Review: Insights analysis of immigrant incarceration from 1870 through 2020; and National Institute of Justice-funded research examining immigration and crime. War-related figures are attributed to Brown University’s Costs of War project, while allegations of war crimes are identified according to the specific findings or legal actions of the institutions cited. Claims about social-media recommendation systems draw on peer-reviewed experimental and auditing research rather than an assumption that all platforms function identically.

Comments