ANALYSIS: “Sanewashing,” or How the Press Makes Trump Seem More Coherent Than He Is

Where Information Fits into the Text’s Hierarchy

The New York Times article from September 5, 2024, is titled “Trump calls for an efficiency commission, an idea pushed by Elon Musk.” This headline is accurate. Trump did indeed call for the creation of such a commission, and the idea did come from Musk. No falsehood was committed. That is what makes this case interesting rather than merely anecdotal. The problem lies not in the reported fact, but in the order in which it is presented. The incoherent response regarding childcare services is mentioned only once, with the words “jumbled and meandering,” and it appears in the twenty-fourth paragraph. Given the limited attention span of an online reader, the twenty-fourth paragraph is like a basement. Few people venture down there.

Researchers have observed a difference in vocabulary between different types of media. Online publications use sharper language—“weird,” “bizarre.” Traditional media favor more neutral terms—such as “meandering” and “erratic.” Their interpretation—presented as a hypothesis rather than an established fact—is that journalists at major publications feel greater pressure to adhere to standards of impartiality and therefore feel less free to describe what they hear in blunt terms. This is plausible. It is not proven by the data cited in the text. The distinction is worth maintaining, because blurring it would amount to committing precisely the very sin being criticized: presenting a reasonable interpretation as an established fact.

A true fact placed in the wrong context becomes a form of silence. The classification of information is an editorial act as weighty as the choice of words themselves.

Two Pitfalls Facing Each Other

The authors raise an objection to their own thesis, and that is what makes their text useful. If “sanewashing” involves filtering too much, the opposite also exists: stenography, which consists of reproducing everything word for word without hierarchy or judgment. No one seriously demands that a daily newspaper publish the entirety of every raw political statement. Editorial work is, by definition, a process of sorting. We remove what is deemed less relevant and highlight what is deemed crucial. This process is not a flaw; it is the very essence of the profession. “Sanewashing,” as they put it, could therefore be an inevitable byproduct of journalism itself.

This concession shifts the question. It ceases to be “should we filter?” and becomes “according to what criteria do we filter?” And these criteria are never neutral. Choosing to treat a 374-word response devoid of content as noise presumes that content matters more than form. Yet in the case of a presidential candidate, the ability to structure a response on a major issue is itself a piece of political information. It reveals the candidate’s preparation, their grasp of the issue, and what will happen when that issue comes before the Oval Office. Treating form as incidental is to decide that only the prepared text matters, and that improvisation says nothing about the person improvising. It is a choice. It should be acknowledged as such.

Selection is legitimate. Selection that does not acknowledge itself as such is less so. Between the two lies only a single methodological sentence that almost no one ever writes.

The example the authors use to undermine their own argument

This is the strongest passage in the text—and the most uncomfortable for anyone looking for a one-sided argument. The researchers revisit the New York Times headline they had criticized and show that it was spot on. The efficiency commission announced by Trump that day became the Department of Government Efficiency, or DOGE. This agency was subsequently linked to the dismantling of the U.S. Agency for International Development (USAID), as documented notably by The Conversation. The headline, deemed too cautious, thus highlighted the element that would have the most far-reaching consequences of the entire speech.

The scale of these consequences is quantified in a study published by The Lancet in 2025, which projects up to 14 million excess deaths worldwide attributable to the reduction in U.S. funding for international aid. This is a projection based on modeling—with the uncertainties inherent in such exercises—rather than a tally of deaths that have already occurred. The distinction is crucial, and the authors themselves make this point by comparing it to “a failed response to the childcare crisis.” Their conclusion is clear: journalists must report on Trump’s gaffes without allowing those gaffes to divert attention from the serious issues buried beneath them. The flaw in “sanewashing” is not that the headline focused on the commission. It is that the rest of the story was buried in the twenty-fourth paragraph.

The headline was right, and the article was wrong at the same time. This dual truth is harder to bear than a simple indictment of the press—and infinitely more honest.

The Political Cost of the Impartiality Reflex

A professional standard designed to protect the reader can, under certain conditions, have the opposite effect. The mechanism deserves to be described precisely rather than bandied about as a slogan. The rule of formal impartiality requires reporting without judgment. Applied to a structured discourse, it works: the reader receives the substance and makes their own judgment. Applied to a discourse that lacks substance, it produces a summary that invents a coherence absent from the original. Not out of malice, but because the act of summarizing presupposes that there is something to summarize. The summary creates a thread where none existed.

This interpretation remains an inference, and I present it as such. The text by the three researchers does not claim to have measured the effect of “sanewashing” on electoral perceptions. It describes a disparity in treatment between online and traditional media, and it interprets that disparity. Other explanations are possible: space constraints, deadlines, a particular editor-in-chief’s choices, or the editorial line of the political desk. None of these hypotheses is ruled out by the data cited. What is established is the difference in vocabulary itself. What remains open is its exact cause. A columnist who claimed to settle the matter here would be going beyond his sources and would, in turn, be doing exactly what he denounces.

There is a way to denounce the whitewashing of coherence that consists of whitewashing one’s own analysis: asserting more than one knows, in a confident tone, while hoping that no one will check the facts.

A Diagnosis Without a Prescription

The term “sanewashing” has a strength: it brings to light a process that used to be seen as mere formatting. It also has a weakness: it suggests intent where the researchers’ text primarily describes an automatic process. “Whitewashing” implies a whitewasher. Yet nothing in the New York Times example demonstrates a desire to make Trump look presentable. The journalist who placed “jumbled and meandering” in the twenty-fourth paragraph did indeed report the fact. He reported it in a place where it carries little weight. There is a chasm between deliberate omission and mechanical placement, and the word does not bridge it.

This imprecision comes at a cost. A concept that attributes an unproven intention becomes a reusable weapon against any coverage deemed too lukewarm—including when that coverage is simply rigorous. The risk is real in an environment where accusations of media complacency circulate across all camps. The authors do not fall into this trap: they conclude with a call for balance, not an indictment. Their position can be summed up in one sentence: point out the digressions without letting them distract us, because what follows the digressions sometimes costs millions of lives. This is a methodological requirement, not a condemnation.

A concept that explains everything no longer explains anything. “Sanewashing” accurately describes a specific, dated, and documented case. When applied to the entire press, it becomes yet another oversimplification.

What Remains When a Summary Replaces the Source

The full transcript of the September 5, 2024, response is public. It exists. It is available for review. And yet virtually all citizens who have formed an opinion about this speech have not read it: they have read a headline, sometimes the first paragraph. This is the normal state of mass media—not a scandal in and of itself. But it transforms every decision about what to highlight into a decision about reality for the vast majority of the public. The twenty-fourth paragraph exists for the archives; it does not exist for the reader in a hurry.

This observation does not lead to any simple solution, and I am not going to invent one. We cannot demand that every article reproduce every statement in its entirety. Nor can we demand that readers systematically go back to the transcripts. What remains possible is more modest: that the decisive fact not be systematically relegated to a place where no one goes. That “jumbled and meandering” be moved up a few paragraphs when the issue at hand affected millions of families. The work of the three Queensland researchers asks for nothing more. It asks that the order of the text reflect the order of importance, and it shows—using just a single day of the campaign—how much the two can diverge.

Between the full transcript and the headline read by everyone, there is a gap of twenty-three paragraphs. This gap is not an oversight. It is the normal architecture of contemporary news.

A Transferable Mechanism

The case study is American, but the mechanism described is by no means specific to the United States. It stems from a professional standard shared by the Western press: report without qualifying, summarize without judging, and let the reader decide. This rule is sound. It protects against creeping editorialization and against journalists who slip their opinions into a report. It becomes counterproductive in only one scenario: when the form of the statement itself is the main piece of news. In that case, a summary faithful to the content distorts reality, because the content was not reality.

The Australian researchers are working on a project funded by the Australian Research Council titled “Understanding and Combatting Dark Political Communication.” Timothy Graham also discloses funding from Meta Platforms Technologies, a fact that the article openly states. This transparency is to their credit and deserves to be highlighted rather than glossed over. It does not discredit the analysis; rather, it allows readers to interpret it with full knowledge of the facts. What their work suggests—without proving it in the strict sense—is that a standard of impartiality designed for one type of political discourse reaches its limits when faced with another type. The debate is not about Trump alone. It is about what the press does when a political actor stops playing by the rules that the press assumes are being followed.

The standard is not to blame. It is outdated. It was written for a world where political discourse had a structure, and where summarizing it meant adhering to it.

The sentence we’ll never read in the twenty-fourth paragraph

There remains one detail that the researchers’ text mentions without elaborating on, and which deserves to be followed through to the end. The New York Times headline was accurate. The commission announced that day became the DOGE, and the DOGE has been linked to the dismantling of USAID, which The Lancet projects will result in up to 14 million excess deaths worldwide. The journalist who chose that headline saw, back in September 2024, what would matter. He simply placed in the twenty-fourth paragraph the evidence that the man announcing this commission was unable to formulate a coherent sentence about daycare centers. Both pieces of information were in the same article. Only one was read.

There is no specific moral to be drawn from this story, and I would be wary of anyone who suggested one. Sanewashing is not a conspiracy; it is a habit. A professional habit, inherited and applied in good faith, which under certain conditions produces exactly the opposite of what it intended. The three researchers from Queensland are not calling for its abolition. They are asking that we examine its cost, case by case. What the example of September 5, 2024, leaves in its wake is not an accusation against a newspaper. It is a question that remains open: how many times must a fact be tucked away in the twenty-fourth paragraph before it ceases to exist?

The transcript is still online. It hasn’t changed. It was the summaries that circulated, and it is those summaries that the world has come to regard as the thing itself.

Signed, Jacques PJ Provost, columnist