Social media is bad for everything and everybody, for humans and journalists and other living things. But at least it reliably generates interesting juxtapositions from which op-ed columns can be made — including columns about how a fixation on, well, social media, is damaging liberalism’s understanding of the world.

The relevant juxtaposition comes from my Twitter feed, which Friday featured — to great acclaim and many retweets — a speech by Sacha Baron Cohen, the erstwhile fake newsman Borat, condemning “a handful of internet companies” for building the “greatest propaganda machine in history” and driving the rise of authoritarianism, demagoguery and bigotry.

At the same time, Twitter also surfaced a recent study from academics in France, Canada and the United States that examined the relationship between social media echo chambers and support for populism in France, Britain and the United States. The authors found that there was either no relationship or a negative one: Populist voters were somewhat more likely to hang out with people of a similar ethnicity or social class offline, but on the internet they were no more likely than other voters to inhabit an echo chamber. And social media use was a strong predictor of opposition to the campaign of President Donald Trump.

This is not the first study to question social media’s supposedly central role in the drama of right-wing populism. Shortly after Trump’s election, economists at Brown and Stanford found that he performed more poorly than Mitt Romney and John McCain among Americans who get their news online, while the voters he converted were often the very offline.

This kind of evidence doesn’t mean that online conspiracy mongering has no influence on populism. People who rarely use the internet might be more easily deceived by fake headlines when they do wander online. Normal partisans may get kookier under the sustained influence of the memeplex. And Cohen is right that small communities of depraved people, from pedophiles to anti-Semites, use online platforms in vicious ways — and internet giants invoke free expression while shirking their responsibility to deny such viciousness a refuge.

But we should be more doubtful of Cohen’s larger narrative, which is commonplace among progressives — a narrative that invokes the “sewer” of social media to explain everything from climate change skepticism to anti-immigration sentiment, portrays Russian trolls and YouTube stars as the crucial actors of the populist era, and proposes the regulation of online speech as the main restorative that the liberal order needs.

Instead, the evidence in the papers cited above hints at a different scenario — in which because educated liberalism is increasingly so very online itself, ensconced in its own self-reinforcing information bubble, liberals end up analyzing populism exclusively through their digital experience even when that analysis is obviously insufficient.

In a recent Boston Review essay, Tufts political scientist Eitan Hersh notes that many American liberals participate in politics through a kind of uber-online “political hobbyism” in which real-world organizing recedes in favor of constant engagement “from behind screens or with earphones on.”

Inside this hobbyist’s world, the centrality of those screens becomes a given. Spend all your time on Twitter and Facebook, and it seems that Twitter and Facebook must be essential to the far right’s appeal and that a better social media ecosystem would suffice to deal with Trump, suppress Marine Le Pen or sideline Nigel Farage.

But if the other side is actually less online than you are, this assumption leads to two mistakes. First, you end up downgrading the obvious real-world forces driving populism’s appeal, persuading yourself that an algorithmic tweak or better fact-checking will deal with deep trends — economic stagnation, social crisis — that would exist with or without fake news.

Second, you lose sight of the ways in which your own information bubble is a potential radicalizing force — including for people observing it from outside, for whom it makes political liberalism seem like an airless world filled with hypereducated ideologues. Indeed, on the evidence of a Democratic primary that seems made for the social media bubble, it’s liberalism that’s being warped by online feedback loops and radicalization cascades.

Which is not to say that conservatism isn’t pretty warped at the moment or that social media has been good for right-wing decency and common sense. (Remember where I began: It’s bad for everything.) But what’s wrong with conservatism has as much to do with old media forces like talk radio and cable news, plus real-world isolation and disconnection, as it does with QAnon. Trump owes his (relative) resilience to the inertia of medium-information voters enjoying a decent economy, not to YouTube radicalism.

And if those voters are the battlefield, then any Democratic strategy will be insufficient unless liberalism realizes that before it regulates social media for other people, it should learn how to resist the internet itself.

Ross Douthat

Ross Douthat is an Op-Ed columnist for The New York Times.