fox files

Counting Backwards, The Deafening Skew Behind America’s Flawed Political Violence Data

Public discourse about political violence in the US is now driven by a single claim, that right‑wing actors commit the lion’s share of attacks. That thesis has migrated from activist reports into journalism and then into official talking points. Yet its footing is weaker than advertised. The proposition depends on datasets with moving definitions and selective scopes. It also depends on a habit of turning non‑political crime into political intent when the offender happens to have the wrong affiliations, while discounting ideologically charged offenses when they flow from left‑wing or pro‑Palestinian causes. When we examine how the numbers are built, we see a pattern. Definitions, inclusion criteria, and coding choices are doing more work than the underlying events.

Begin with first principles. A fair account of political violence must track two simple ideas. First, political motive, not the identity of the offender, is what makes an act political. Second, comparable acts must be counted on comparable terms. If a right‑wing offender’s ordinary bar fight is listed as political because he once shared extremist memes, then a left‑wing offender’s riot‑linked arson must be counted as political when it was plainly undertaken for an ideological purpose. If a database counts propaganda stickers as violent extremism on the right, it must also count left‑wing vandalism of memorials and offices as violent extremism on the left. If a study focuses only on fatal attacks, it must explain why non‑fatal bombings, arsons, beatings, and attempted assassinations, many of them left‑coded, do not count. These are not partisan demands, they flow from basic standards of inference. Like cases should be treated alike.

The most aggressive inflation starts with what gets labeled right‑wing by theme rather than by motive. Some compilers treat any identity‑biased crime as quintessentially right‑wing, even when the offender’s own rhetoric and associates place him in pro‑Palestinian or left‑wing circles. In that frame, antisemitic offenses are assigned to the right by definitional fiat, because the target is a protected group and because the right is said to be the natural home of bigotry. That approach reverses the direction of explanation. We are supposed to infer political ideology from the identity of the victim. The method equates theme with motive and then motive with right‑wing identity. Such reasoning would be rejected in any other empirical domain. It lets preconception fix the labels in advance and it protects the labels from correction when the facts of a case cut the other way.

Next, there is the tactic of counting everything around the right while counting only a narrow set of events on the left. One widely cited stream of reports counts every homicide committed by a person with white‑supremacist interest, including domestic disputes and intra‑gang murders with no political purpose. In the same breath, it excludes left‑wing violence that does not produce a corpse. The result is a double filter, add ordinary crime to one side and subtract ideologically driven, non‑fatal violence from the other. Add enough of the former and subtract enough of the latter and the headline becomes inevitable. The data will perform as designed.

A third move is the curated time window or the one‑off outlier exclusion. In some tallies, a single Islamist megattack that reshaped modern history is removed as exceptional. Removing it reduces the non‑right body count by thousands, which predictably enlarges the relative share of right‑wing violence. The rationale is presented as methodological prudence, but the consequence is political arithmetic. The new denominator makes right‑wing violence look like the dominant fraction by construction. If the goal is to measure danger and reality, there is no justification for erasing the single most consequential terrorist attack in US history. If the goal is to win a talking point, exclusion makes sense.

To see how these three moves work in practice, look closely at a few studies that shape the public conversation. Some academic‑adjacent databases operationalize political violence by category rather than by motive. Identity‑focused offenses are called right‑wing regardless of the offender’s own statements. Trivial or non‑violent acts, such as flyers or stickers, are counted alongside serious violent crimes. Meanwhile, ideologically driven left‑wing violence is discounted when it occurs during riots or in anarchist zones that officialdom preferred to frame as spontaneous unrest or mutual aid. The effect is a spectacular asymmetry. The right swallows even apolitical crime by offenders with the wrong associations. The left sheds political motive in cases where violence was plainly part of a cause. Inferences about national danger are then built on this misaligned scaffolding.

A second cluster of reports focuses on murders by extremists and then treats all killings by a person with extremist ties as extremist killings. Consider what that means. If a white‑supremacist gang member murders his girlfriend in a domestic dispute, the death is credited to right‑wing political violence. The political story gets a data point, but there was no political motive, there was only a crime that would have occurred regardless of ideology. Multiply this across a year and you can generate a lopsided pie chart. Then look at the inverse. Left‑wing attacks that injure, burn, intimidate, and terrorize but that do not result in death are omitted because no one died. The chart does not budge. The public sees the chart. The chart says the right is the problem. The construction of the chart does the work.

A third tranche of analysis focuses on the narrow category of terrorist murders. In one prominent version, only events with at least one fatality are counted. Plots are excluded, foiled attacks are excluded, attempts are excluded, arsons are excluded unless someone dies, riots are excluded unless a specific homicide is tied to political motive defined in a narrow way, and the September 11 attacks are placed in a separate box. In addition, the classification of several offenders as right‑wing is made on loose criteria, sometimes on the presence of racist postings or confused manifestos that do not articulate a political plan. When critics scratched the surface and re‑coded ambiguous cases, the large gap between right and left nearly vanished. Correct a few design choices and the headline dissolves into parity or into a more complex distribution that resists sloganeering.

In any rational inquiry the cure for definitional bias is casework. We must test the rules against particular incidents that the public has been taught to treat as examples of right‑wing political violence. When we do, many do not fit. They are either left‑coded, mixed, or non‑political. They often show untreated mental illness rather than doctrine. They often show radical milieus that have little to do with conservatives. They often show offenders who never voted in a Republican primary, who never donated to Republican candidates, and who told friends they had progressive or anti‑establishment views.

Consider the case of Vance Luther Boelter. He was appointed by a Democratic governor to a state workforce development board. He moved in Democratic circles. When he erupted in murderous violence, he targeted Democratic officials who had voted with Republicans on a specific immigration measure. He did not hunt Republicans. He hunted Democrats who in his view had betrayed a cause. The material recovered from his car included anti‑Trump flyers tied to a coordinated protest theme and other standard progressive paraphernalia. Sympathetic reporting later attempted to rebrand him as a Republican or a marginal Trump voter based on contested claims by acquaintances with obvious motives to sanitize the politics of the incident. The uncontested facts tell a simpler story. This was a politically motivated attack, but it was intra‑Democratic retribution over immigration policy. In any balanced dataset, the incident would count as left‑coded or at least as non‑right. It has instead been recycled as an instance of right‑wing violence because the victims were Democrats. This is definition by target again, not by motive.

Now take David DePape, the attacker in the Paul Pelosi case. The public was assured that he was a specimen of right‑wing rage. That claim folded fast when his history emerged. He was a Canadian national who was living and voting in the United States illegally. He lived for years in a progressive enclave with a left‑coded partner known for street protests and for far‑out radicalism. His home displayed a BLM flag and LGBTQ imagery. He had registered to vote with the Green Party and once cast a Green vote for a socialist candidate. He drifted into conspiracism and apparent psychosis, telling people he thought he was Jesus. None of this suggests a coherent right‑wing identity. It suggests a volatile mixture of mental illness and fringe ideology with leftist antecedents, followed by a paranoid fixation that eventually incorporated anti‑Pelosi fantasies. It is not hard to see why a media ecosystem primed to find a MAGA archetype fastened on that angle. It is harder to explain why serious compilers continue to code this event as right‑wing. If motive and milieu matter, the classification should be mixed or indeterminate at best. If the presence of a partisan target is enough to fix the label, then we are back to definition by victim rather than by motive.

Turn to Cody Allen Balmer, the arsonist who attacked the Pennsylvania governor’s residence. In real time, several commentators and officeholders offered the ritual line, another example of far‑right political violence. The details contradict the script. Balmer described himself as a Marxist. He expressed pro‑Palestinian themes and targeted the governor because he believed that the governor would harm Palestinians. His record shows serious mental illness, including bipolar disorder and schizophrenia, and he had a trail of domestic violence and criminal charges. He never registered as a Republican, never voted in a Republican primary, and there is no record of Republican donations. When precise facts are inconvenient, the narrative retreats to ambiguity. Maybe he had some right‑wing sympathies. Maybe he saw posts on 𝕏. Maybe he was disturbed by current events in a way that aligned with conservative anger. The facts remain. Marxist self‑description, pro‑Palestinian motive, mental illness, and no partisan ties to the GOP. A fair coder would place this event on the left or mark it as non‑right. Yet the incident continues to be invoked in public as evidence for the thesis that right‑wing violence predominates. That is not data, it is branding.

Finally consider Anderson Lee Aldrich, the Club Q shooter. The instant narrative labeled the attack anti‑LGBTQ political violence from the right. The emerging record will not cooperate. Aldrich identified as non‑binary and asked to be addressed as Mx. Aldrich. He frequented Club Q and other gay venues. He never voted Republican, never participated in a GOP primary, and was never a donor to Republican candidates. His life showed serious dysfunction and suicidality, an arrest following threats involving a homemade bomb, and a trail of psychiatric treatment. In the courtroom, the picture was of a disturbed young person with violent fantasies and a warped relationship to identity, not a doctrinaire activist from any organized right‑wing scene. No fair reading of his history yields the conclusion that he was a conservative extremist. The rush to brand him as such flowed from the theme of the attack and the identity of the victims. The method is the same as before. Reverse engineer motive from target, then paint the act with the broadest possible brush.

These four cases are not cherry‑picked. They are prominent illustrations of a wider tendency. Where the facts point left or toward non‑political pathology, coders and commentators still push right. Where left‑wing or pro‑Palestinian attacks are unambiguous, the event is reframed as criminal violence with no ideology or it disappears into the gray spaces of data design. In the aggregate the skew compounds. Trivial propaganda acts inflate counts on the right. Non‑fatal left‑wing attacks are excluded. Ambiguous lone offenders are labeled right‑wing by default. Islamist and eco‑extremist events are minimized by time slicing or by outlier exclusions. Once the machinery is assembled, the conclusion is guaranteed. The right will look like the predominant source of political violence even if the underlying reality is mixed or if the greater share of routinized street violence has flowed from the left.

What would a sound methodology look like. Begin by coding motive, not identity, and require clear evidence for political intent. If the offender cannot articulate a political goal and there is no credible public record of one, do not count the act as political. Next, treat like cases alike. If domestic homicides by extremist affiliates count on one side, count them on both sides, or better, exclude them on both sides unless there is evidence the killing was carried out for political reasons. Third, include serious non‑fatal political violence, including arson, bombings, beatings, and attempted assassinations, and then weight incidents by severity. The public cares about danger, not only about death statistics. Fourth, avoid definitional shortcuts that infer ideology from target identity. Fifth, publish full incident lists with coding rationales so that outside reviewers can audit classifications. If your conclusions depend on hidden spreadsheets and shifting labels, they are not conclusions, they are talking points.

One might object that the exact labels do not matter because the trend is the same no matter how you count. That is false. Labeling shapes resource allocation and legal focus. When the data tell the public that right‑wing violence dwarfs left‑wing or Islamist violence, agencies are pressured to divert attention and funds accordingly. That may be wise in some periods. It is reckless if the numbers were built to sell a narrative rather than to inform about risk. It also warps civic understanding. Citizens begin to see ordinary conservatives as adjacent to violent fringe actors. Speech is chilled. Political engagement is stigmatized. The result is a brittle public square in which statistical fog is used to distress one side of the aisle.

Another objection says that it is unfair to distinguish between violent neo‑Nazis and conservatives because the former draw on a right‑coded tradition. The answer is simple. Fringe racists reject the central principles of modern conservatism and are expelled from mainstream conservative institutions. They are not part of the Republican coalition. They are enemies of it. Counting their apolitical crimes as right‑wing political violence smears millions of citizens by association. It is intellectually lazy and morally corrosive.

A third objection says that Islamist violence and left‑wing violence are red herrings, because the object of current concern is domestic extremism by whites. This reply repeats the selection problem at a higher level. The question is not whether we should ignore white offenders, the question is whether we should ignore other offenders, other ideologies, and other patterns of violence in order to uphold a single storyline. A government that can only see one danger is a government that will miss the next danger.

A final objection is rhetorical rather than empirical. It says that scrutinizing the numbers is an attempt to excuse violence on the right. The response is closure. No one is excusing anything. Violence for political ends is wrong. It should be punished. The claim under review is narrower. We are asking whether the claim of a dominant right‑wing share is supported by neutral counting. When we track motive, when we code like with like, and when we stop converting ordinary crimes into political statements, the dramatic right‑dominance story collapses. What remains is a complex landscape in which left‑wing and Islamist offenders, along with non‑political violent actors, account for a great deal of harm and pressure. The conservative point is not special pleading. It is a request for sobriety and standards.

Returning to the four cases. A Democratic appointee murders Democrats for voting with Republicans on immigration, a left‑coded conspiracist with visible progressive markers attacks the husband of a Democratic leader, a Marxist arsonist targets a Democratic governor over a pro‑Palestinian grievance, and a non‑binary club regular with a history of mental illness commits a mass shooting at a gay venue. None of these fit the template of organized right‑wing political violence. All four have been placed into that template anyway. If that is how the corner cases are handled in public view, imagine how less visible cases are coded. Imagine how many times the label is fixed by target, not by motive. Imagine how many times non‑fatal left‑wing violence is thrown out of scope. The dataset is not a mirror of reality, it is a machine for producing a preferred answer.

The remedy is not to flip the sign and declare that most political violence comes from the left. The remedy is to build an honest ledger. If we do, two conclusions will follow. First, much of what is today labeled right‑wing political violence is either non‑political crime by people with ugly affiliations or it is ambiguous lone‑offender pathology. Second, a large share of ideologically motivated street‑level aggression, from riots to arson to targeted intimidation, has been left‑coded or aligned with left‑wing and pro‑Palestinian causes in recent years, and it has been discounted by the very studies that purport to measure the phenomenon. Those conclusions do not vindicate anyone. They force us to see the shape of the problem without partisan blinders.

This is not an attempt to shock the conscience with graphic anecdotes or to turn data into propaganda. The aim has been clarity. Will stricter definitions and transparent coding erase right‑wing political violence. Not at all. They will do something better. They will put it in its proper proportion alongside left‑wing and Islamist violence and alongside non‑political violent crime. Only then can citizens and officials’ reason about risk without falling for the rhetoric of the spreadsheet. Only then can we protect the republic without sacrificing the truth to the fashion of the moment.