Now we know: Second-term Trump is far from invincible. His approval rating is falling fast. So is the public’s assessment of his approach to the economy, and—perhaps most notably—his approach to immigration.

This isn’t just good news for America. It’s an important moment in a debate that has loomed over Democratic and liberal politics for nearly a decade. One of the longest-running, most consequential disputes over Democratic strategy has abruptly produced an apparent winner.

Readers of this website should be familiar with the divide. On one side are a set of pundits, strategists, and politicians who strongly believe that Democrats have to navigate public opinion as it exists at any given moment. If polls report that voters care most about the economy, then that’s what Democrats should talk about. If a policy polls well, Democrats should support it. And there’s a corollary: If Trump or Republicans poll well on a certain topic, Democrats should avoid that topic as unfavorable terrain.

On the other side is a set of voices who see public opinion as dynamic—shaped around a fluid consensus that follows the voices of leaders and opinion-setters. In this telling, it is folly for Democrats to adhere strictly to polling, which only limits their room to maneuver. Instead, Democrats should seek to create new public narratives—especially by going on the offensive against Trump—even if that means attacking him on issues where he is strong.

I am an adherent of the second approach. But admittedly the polling fundamentalists accumulated some evidence during the 2024 election. Kamala Harris ran a pugilistic campaign, attacking Trump as a fascist, pushing hard to stir up alarm. And while she likely performed better than Joe Biden would have, the results speak for themselves. Harris’s warnings went unheeded.

In recent weeks the debate resurfaced again, centered around Trump’s brutally authoritarian extraordinary renditions, particularly his disappearance of Kilmar Abrego-Garcia into an El Salvadoran dungeon.

The moral outrageousness of this case, and of Trump’s subsequent refusal to free Abrego Garcia—even after his administration admitted Abrego Garcia was wrongly expelled—has gradually moved it to the center of national discourse. Democrats, abandoning habitual caution, have continuously escalated the matter. Their efforts were capped by Maryland Senator Chris Van Hollen’s astonishing trip to El Salvador, and equally astonishing success at securing a meeting with Abrego Garcia, which dominated the news for days.

But the polling fundamentalists—writers like Nate Silver and Matt Yglesias—saw disaster in the making. They warned that Trump polled well on immigration, poorly on the economy, and Democrats would be best advised to set aside their moral scruples and simply follow the numbers. Hit Trump where he was weak: tariffs and inflation. Leave Abrego Garcia in obscurity. And some Democratic politicians followed suit, including Gavin Newsom, who described Trump’s deportations as a “distraction” from the economy.

A controversy like this usually goes unresolved, and the respective sides will fight for months or years. But this one has been resolved almost immediately: the polling fundamentalists had it wrong.

Most recent polls show Trump’s approval rating on immigration, far from being fixed, has begun to collapse. One Economist/YouGov poll showed his approval on the issue falling from +6 to -5 in two weeks. Trump’s overall approval rating has also fallen sharply at the same time. Democrats seem to be facing no political penalty for focusing on immigration, and, moreover, they’ve badly wounded Trump on his strongest issue.

The importance of this shift is hard to overstate. It’s proof of concept for dynamic public opinion. We now know, as demonstrable fact, that changing the polls is possible, even when attacking Trump where he’s supposedly strongest.

But why now? What’s different? This isn’t the first time Democrats have hit Trump aggressively, and in the past the public has not always responded. Moreover, Trump has previously seemed to land at a polling floor—roughly 38-40 percent approval—and not budged further.

I would contend that the most likely explanation lies in how Americans get their news. There is a paradoxical quality to modern media, where most stories bounce off, leaving people unaware and unaffected—but every so often, a development breaks through and transforms how Americans see the world. Democrats’ job is to seek out the second kind of story. To do that they need to understand how media has changed.

BREAKING: NEWS

Public opinion can move rapidly in response to events. And it should be obvious that the public learns about events through media consumption. There is simply no other way for most Americans to collect information about national happenings in a brief period.

But which media does the public see? In the past, this was a trivial question. It was expensive to produce and publish news, which meant that there were only a handful of news sources available to each voter, each of which represented a substantial operation. Voters could watch one of a few broadcast channels with enormous audiences and national reach. Or they could subscribe to newspapers, most of which would have hundreds of employees dedicated to covering news in depth. When something happened in the wider world, these outlets would cover it, paying reasonable care to accuracy

That was it. That was all. There was no other practical way to learn about national or international affairs. Everyone tuned into a handful of wavelengths. And so millions of people would hear the same stories, at roughly the same time, and with no alternative sources of mass information. Radicals and conspiracy theorists were limited by the minuscule reach of independent journals, chain letters, and other minor media. Thus, mainstream news narratives were widely shared and generally dominant among the public. If the president was caught in scandal, everyone knew. If the economy was bad, everyone knew.

Today, things are very different. Sources of information have fragmented. Different channels cater to various political leanings. Newspapers have eroded into shadows of their former selves. Most consequentially of all, a huge segment of the public-information ecosystem has moved online, into an endless array of news sites, podcasts, and social-media posts.

I have written about fragmentation before, but it’s worth harping on, because the change is profound and fundamental. No longer does everyone go home and flip on the same channel after dinner. Between TV stations, web sites, podcasts, and social media, each individual audience member assembles a bespoke diet of information. Everyone listens to different sources—sometimes slightly different, sometimes vastly different. This process is probably amplified by the fact that human beings have never spent quite so much time consuming outside media—an average of six hours a day on the internet, half of which is spent on social media, and three hours watching television. We are immersed in information, and—vitally—we choose what to watch and read, from a vast menu of options.

From this torrent of facts and ideas, we populate our minds. This cannot be avoided. Human beings are forced to build their notions of the larger world with secondhand information. It is unlikely you or I will ever meet Donald Trump, travel to Iran, or experience the war in Ukraine. Americans everywhere “know” that Joe Biden presided over a surge of illegal immigration, but most couldn’t intuit that, because most of our communities were untouched and unchanged by it. Many developments are simply too diffuse to be directl