The problem with common sense, goes an old joke, is that it is not so common. A less-recognized problem is that sometimes it doesn’t make sense, either.

Those contradictions were on display during a tense exchange Wednesday between senior Trump adviser Stephen Miller and New York Times reporter Glenn Thrush, who asked Miller to provide some proof that low-skilled immigrants are causing job losses for American citizens. Here’s a short excerpt from the much-longer dialogue:

Miller: Let's also use common sense here, folks. At the end of the day, why do special interests want to bring in more low-skilled workers?  And why historically—

Thrush: Stephen, I'm not asking for common sense. I'm asking for specific statistical data.

Miller: Well, I think it's very clear, Glenn, that you're not asking for common sense, but if I could just answer your question.

Thrush: No, no, not common sense. Common sense is fungible. Statistics are not.

Moments later, having skirmished over studies, Miller once again returned to the powerful justifying force of common sense: “The reality is that, if you just use common sense—and, yes, I will use common sense—the reason why some companies want to bring in more unskilled labor is because they know that it drives down wages and reduces labor costs.”

The recourse to “common sense” is probably not accidental, especially for a student of political movements like Miller. Nearly every contemporary politician is guilty of falling back on the phrase, but for centuries, populist movements in particular have invoked common sense as a justification for policy goals and as an antidote to expert opinion. Like President Trump, the people invoking it have often done so, as Sophia Rosenfeld writes in her book Common Sense: A Political History, as part of “a populist style of conservatism that celebrated authoritarian governance alongside the traditional ways, values, and language of ordinary people.” (This would have been a much simpler, and faster, story about the Thrush-Miller exchange had I trusted my gut, rather than falling into the trap of seeking out expert opinion.)

Even the specifics of the exchange between Thrush and Miller fit the historical pattern: Common sense has frequently been deployed in service of xenophobic and nationalist concepts, and it has often used elite journalists—say, New York Times reporters—as convenient foils. And when you start hearing a lot about “common sense,” it’s often a sign of national crisis, or at least of a serious effort to undermine faith in national institutions—which few people would dispute.

“Common sense has ... served to underwrite challenges to established forms of legitimate rule ... in the name of the special kind of intuition belonging to the people,” Rosenfeld observes. Common sense is typically evoked and held up as authoritative only at moments of crisis in other forms of legitimacy. Revolutions, which, by definition, result in divided loyalties and the upending of the rules to multiple domains at once, are a case in point. Otherwise common sense does not need to call attention to itself.”

Time and again, the Trump administration has embraced solutions that it has labeled common sense, but which are either highly disputed, wholly counter to expert consensus, or flat wrong. This has been true on immigration, on protectionism, on industrial policy, climate change, and a range of other issues.

When Thrush told Miller, “Common sense is fungible. Statistics are not,” he was half-right. Common sense is fungible, and what qualifies in one age sounds preposterous in another. (The title of Thomas Paine’s Common Sense appealed to the wisdom of people in governing themselves, but it cracked an implicit joke, too: The idea of democratic governance was decidedly not commonsensical for the era.) But statistics are also often fungible. Different studies produce different results; different scholars scope their research differently and land in different places.

Miller, for example, cited a study by George Borjas, a highly respected Harvard scholar who’s known as an immigration skeptic. Borjas looked at the effects of the Mariel boatlift, which plugged 125,000 low-skilled Cuban immigrants into the Miami labor market, and found it had depressed wages. But a different study found little impact; other scholars have argued that Borjas’s research doesn’t really prove what he says it does. Thrush also cited a recent National Academies of Sciences, Engineering, and Medicine study that found little effect on wages.

This doesn’t mean that Borjas is wrong, but it shows that finding the single definitive stat that resolves a debate is often impossible. It’s much easier and more effective to retreat to common sense than to get into the trenches of statistical warfare. (Miller, a veteran immigration skeptic, could probably do either, but he also knows that most viewers aren’t ready to dig into the nitty gritty, and thus that common sense is a better political message.)

In other cases, Trump is flouting more uniform expert consensus in the name of common sense. It is common sense that the only way to keep people and drugs from crossing the Mexican border is to build a wall from ocean to gulf. But experts reject that, noting the physical impossibility of sealing some parts of the border and pointing out that unauthorized immigrants and illegal drugs get through the border by various measures—semi truck, for example. (Ironically, border crossings have declined sharply since Trump’s inauguration, offering evidence against the notion that illegal immigration can be controlled only with a wall.)

There’s not only an overwhelming scholarly consensus that the climate is warming because of human activity, but an increasing number of material phenomena—from flooding in Florida to droughts in California—show it. Yet Trump appeals to common sense, saying, for example, that a snowstorm in Washington disproves the idea of climate change.

It makes Trumpian common sense that if the president can intervene to stop the closing of a Carrier plant in Indiana, it’s good for workers. But experts warned that such interventions rarely produce lasting changes if companies think they can cut costs by moving production elsewhere. Given that the plant is still shipping jobs overseas and cutting positions even as it profits from a raft of tax incentives, the common sense has some serious flaws.

It makes Trumpian common sense that if you place high tariffs on foreign goods, it will benefit American industry, protect manufacturing, and help blue-collar workers. One can certainly find economists who will argue that targeted tariffs in specific cases are useful tools, or even that the United States ought to adopt a more protectionist stance overall. But they are also in the minority. Most experts point to the increased costs for American consumers and the risks of starting trade wars.

Common sense can be a useful corrective for experts, who are as fallible as any other group of humans—and sometimes need a gut check. And scientists are susceptible to their own form of common-sense-bias, dismissing novel findings that contradict established paradigms or the weight of the published literature because they just know they can’t be true.  But scientists are working from a method that allows received wisdom to be challenged, debated, and ultimately overturned.

That’s precisely the sort of process that invocations of common sense are designed to short circuit, making it harder for new evidence to modify old ideas.

Meanwhile, common sense is plenty susceptible to error on its own. Take this riff from Trump in August 2016:

Another major issue in this race is foreign policy. Hillary Clinton has made one bad foreign policy decision after another. Beginning with the support for going to war in Iraq, and I opposed it so strongly. Nobody cared. I was a civilian, but I opposed it. I said you will have a total destabilization of the Middle East. It was such common sense and look what happened.

Trump was lying—he supported the Iraq war at the time, but turned against it as it went sour, just like Clinton did. And, in fact, just like most of the American public did. Gallup found that in March 2003, as the war began, three-quarters of Americans supported the war. In other words, “common sense” was in favor of the war—and just like Trump, Clinton, and plenty of foreign-policy experts, it was wrong.

Trump’s Iraq war claim is a tidy illustration of how common sense, even while masquerading as unchanging and disinterested, is just as fluid as any other source of knowledge or legitimacy. When a politician invokes common sense, she is falsely claiming the mantle of apolitical neutrality in the service of fundamentally political ends.

“Claims about common sense are, in public life, almost always polemical: statements about consensus and certainty used to particular, partisan, and destabilizing effect,” Rosenfeld writes. “Common sense, as a presumed form of knowledge, cannot be divorced from power or protest. ”

The problems with appealing to common sense are more serious than simply dressing up political sentiments as apolitical intuition. As Rosenfeld points out, common sense is at heart an assertion (accurate or not) of the majoritarian will that serves to cut off debate—paradoxically undermining one of the great features of democracy, which is its openness to new and counterintuitive ideas. So whenever a politician or politician’s aide stakes any policy proposal primarily to the fact that it seems like intuitive wisdom, that’s probably a sign that their conclusion should be scrutinized skeptically.

It’s just common sense.