Graeme Wood (b. 1979) published “What ISIS Really Wants” in The Atlantic, and built the article from Haykel’s analysis. Wood took a sentence that meant one thing in Haykel’s field and let it mean another somewhere else. Inside the study of religion, “ISIS reasons from Islamic legal materials” is a report. It carries no verdict. Released into journalism and politics through “What ISIS Really Wants“, the same words became a charge, that the group speaks for the faith. Haykel had said nothing of the kind. He lost the sentence at the border between fields.
The usual lesson drawn from the episode is that scholars should guard their phrasing. The deeper lesson is about location. A finding turns dangerous not when it surprises but when it lands beside the premise of a live political coalition. Surprise alone does nothing. A strange result with no movement built on its denial sits in the journals and troubles no one. Set a routine result next to the origin story a coalition needs, and it goes off.
That yields a procedure. Name a coalition’s founding story. Then find the mature subfield that treats that story as partial, contingent, or false. The intersection is where the next Haykel case waits. A map of explosive findings is a map of which myths still carry weight.
The conversion runs through three moves, and only three. A claim about an average gets read as a claim about an individual. A description gets read as a recommendation. A partial cause gets read as the sole cause. Learn the three and you can predict the misreading before it arrives.
Take three findings that are furniture inside their fields. In the legal academy, few propositions are more settled than the realist one. Judges do not find law waiting in the text. They make it, and their choices carry their politics and their histories. Scholars treat this as a feature of any legal order. A journalist or a senator treats the same sentence as an indictment, a sign that a court has no standing to bind anyone. The scholar describes. The politician hears a verdict on legitimacy.
History supplies the second. Eric Hobsbawm (1917-2012) and Terence Ranger (1929-2015) collected the essays in The Invention of Tradition
to show how nineteenth-century elites manufactured national custom to bind populations to new states. Nations are recent work, most of their ancient pedigree assembled within living memory of a great-grandparent. Historians hold this without heat. Call a national myth an invention in the political field and you have said something near treason, because the myth is doing work, holding men to a flag and a border.
International relations supplies the third. John Mearsheimer (b. 1947)
argues that states chase power because no authority stands above them, and that talk of rights and democracy often dresses the pursuit of interest. Realists debate this without alarm. A journalist reads it as cynicism, or as apology for whoever does the killing. A description of a system becomes, on export, a confession of the analyst’s own soul.
The three share a move. Each misreading takes a description and hears a prescription. Judges make law becomes judges should rule as partisans. Nations are constructed becomes nations are fraudulent. States seek power becomes the analyst approves. Hold the trio in mind and the other two strip-operations show up on schedule in the cases below.
Almost every such list runs in one direction. Genetics, sex, group gaps, immigration, policing, empire. Every item threatens the left’s account of the world. A list like that reads as a weapon, and it should, because a man who catalogs only the vulnerable premises of the people he opposes has not stepped outside the field problem. He has performed it. The correction is symmetry. So the findings below run in pairs. One burns the progressive coalition. One burns the traditionalist or populist coalition. The pairing carries the argument.
Start with stereotype accuracy. In social cognition, a body of work associated with Lee Jussim finds that many demographic stereotypes track group averages with fair accuracy, better than most predictions the discipline produces. The scope condition governs everything. An accurate group average tells you nothing about the man in front of you and licenses no treatment of him. This is the average-read-as-individual move, and inside the field a hundred cautions block it. Strip the cautions and the finding reads as a defense of prejudice. It lands beside the progressive premise that a stereotype is an error imposed by power. So it burns.
Against it, higher criticism of scripture. Composite authorship, pseudepigraphic attribution, redaction across centuries, dating that places texts long after the events they narrate, these have been ordinary in the academic study of religion for more than a hundred years. It mirrors the Islam case. Inside the field the findings pass no ruling on whether a tradition is true or good. Released to believers, they read as an assault on revelation, because they land beside the premise a religious coalition needs, that the text arrived as it claims. The shocked party here sits on the right.
The second pair. The Implicit Association Test measures something with poor test-retest reliability and thin power to predict conduct, as the Oswald meta-analysis laid out. Mandatory diversity training produces little lasting change and sometimes provokes backlash, as Frank Dobbin and Alexandra Kalev have shown across decades of firm data. In psychometrics and organizational sociology these are routine. Exported, they read as “bias does not exist,” which the research nowhere holds, and they threaten the budget of a large training industry, so they travel hot.
Against it, descriptivist linguistics. Nonstandard dialects are rule-governed to the same depth as prestige dialects, and no test inside the field ranks one variety as more correct than another. Ordinary since Leonard Bloomfield (1887-1949). It burns language-purist conservatism, and the coalition it wounds wears the other color.
Some findings decline to pick a side. Applied economics comes near consensus that rent control lowers the long-run supply and quality of housing, and that tariffs impose broad losses to hand concentrated gains to a few producers. The first burns the urban left. The second burns the protectionist right. One profession, one method, two coalitions cut by the same blade. A man who accepts the tariff finding and flinches at the rent-control finding, or the reverse, shows that his objection was never about method. It was about whose myth took the hit.
Development economics adds another two-sided case. After the aid-effectiveness debates, the weak link between foreign aid and national growth is familiar ground; William Easterly pressed the point, with Angus Deaton (b. 1945) adding his cautions. Exported, it reads as “stop helping the poor,” and it unsettles a humanitarian consensus that runs from the center-left through the religious right.
My supply of clean right-burners ran shorter than my supply of left-burners. I reached higher criticism and descriptivism with ease and then had to hunt. The extra left-burners came without effort: terrorists are not disproportionately poor or uneducated, and are often the reverse, as William Easterly pressed the point, with Angus Deaton (b. 1945) adding his cautions paid for reporting; the age-crime curve; the adjusted sex pay gap that shrinks to a fraction of the raw gap once hours, occupation, and career continuity enter. The shortage on the other side is not proof that the right holds fewer vulnerable premises. It reflects where the academy’s own coalition sits. Its taken-for-granted findings more often contradict the left’s origin stories, so its exported banalities more often burn the left. That is the field problem operating one level up, inside the discipline that produces the banalities, and the analyst who forgets it will mistake a demographic artifact of the professoriate for the structure of reality.
Two findings sit above the pairs and threaten the trust both coalitions need. A large share of published results in psychology, biomedicine, and economics fails to replicate; John Ioannidis (b. 1965) put the case in a paper titled “Why Most Published Research Findings Are False.” Metascience treats fragility as a normal property of the literature. Exported, it becomes “science cannot be trusted,” the inverse of the field’s own careful conclusion, and it now arms the anti-vaccine and climate-denial politics the metascientists mostly oppose. And free-will skepticism is common among neuroscientists and philosophers, along with doubt about the folk model beneath retributive punishment. Released into the criminal law, it reads as “no one is responsible for anything,” and there it detonates.
Return to the rule. The danger tracks coalitions, not counterintuitiveness. The sentence stays grammatically fixed while its social function changes at the border, and the border is wherever a coalition has staked a premise it cannot afford to lose. Haykel’s trouble was never mainly that Wood misquoted him. A statement meaning “ISIS holds an internally recognizable legal argument” entered a field where it could be heard only as “ISIS represents authentic Islam.” The next cases will cluster around genetics, sex, migration, crime, group gaps, scripture, and national antiquity, because those are the places where academic description and political legitimation now sit close enough to touch.
There is a test that follows from all of this, and it is aimed at the analyst rather than the coalition. List the findings that burn your enemies. Then list the ones that burn your friends. If the second list is short, the trouble is not the world. It is the length of your reach, and the coalition you forgot you were standing in.
