Sometimes it only takes one conversation with one person to crystallize your understanding of something. Such as when I spoke with Moshe Bader in the parking lot of a rally being held by President Donald Trump in August 2018. Bader explained succinctly why he and so many of the thousands of other people there had shown up in hot, rainy weather to hear him speak.

When Trump was pushing back against his opponents, Bader said, “you feel like he’s fighting for you.” And when he scored a victory, his supporters felt like they were, as well.

The personalization of this struggle was important, a central part of Trump’s appeal. It overlapped with a sense common among his supporters that they were increasingly victims in modern America. Trump understood that sense and amplified it.

“We’re all victims,” he told an audience at a rally Dec. 5 in Georgia. “Everybody here, all these thousands of people here tonight, they’re all victims, every one of you.”

It was effective as a tactic for building loyal support. But the fights that Trump wanted his supporters to engage in were often secondhand: He was being targeted on their behalf, he argued, but it was still him, not them, who were being targeted.

At about the time of that rally in Wilkes-Barre, Pa., three years ago, Trump had already stumbled onto a way to make that sense of oppression more personal to his supporters. In the wake of the 2016 presidential election, social media companies like Facebook and Twitter had begun taking a more active role in uprooting the false information and abusive behavior that had to that point been largely unchecked. Twitter, for example, implemented a system in which people who violated certain rules were demoted from searches, limiting the spread of their posts. Eventually, Twitter went further by blocking misinformation from view.

A lot (though not all) of those who were shunted into the background by these changes were conservatives. In July 2018, Trump seized on the popular term for Twitter’s demotion of content: shadow-banning. Trump claimed (on Twitter, ironically) that it was happening to “prominent Republicans” and insisted that his administration would “look into this discriminatory and illegal practice at once.”

What was actually happening was that there was some overlap between the group of bad actors and the group of Trump supporters and the social media company, targeting the former, necessarily also targeted the latter. (An analysis conducted after 2016, for example, found that conservatives were more likely to share false news reports on Facebook.) The following April, Vice reported that Twitter was struggling to figure out how to police white supremacist rhetoric while not disproportionately affecting the political right and its political leaders.

In short order, a narrative — fostered by Trump and his allies — took root: These social media companies were seeking to stamp out conservative thought and rhetoric. Incidents in which posts were flagged or ads banned were seen not as warranted or overzealous but, instead, were elevated as symptoms of political bias.

During the 2020 cycle, social media companies were much more fervent in addressing misinformation. Posts from Trump himself were removed or blocked, often ones including obviously false claims about election security. A sketchily sourced story that appeared in the New York Post and which in broad strokes mirrored the disinformation campaign deployed by Russian actors four years prior was squelched on the platforms, providing a focal point for conservative fury at the heavy-handed oppression of Big Tech.

Last November, weeks after the election and as Trump was battling to convince the world that the 2020 election had been stolen from him (it wasn’t), Pew Research Center asked Americans how they felt about the actions of social media companies. Most Americans (and three-quarters of Democrats and Democrat-leaning independents) approved of efforts to flag misinformation. Three-quarters of Republicans, though, opposed — and those whose most trusted source of election information was Fox News were much more likely to say they disapproved of that behavior.

Interestingly, nearly half of respondents, including most Republicans and Republican-leaning independents, thought that the behavior of social media companies had a major impact on the election. (Among Fox News viewers, that opinion was held by nearly three-quarters of respondents.) This idea that social media was not only stamping out conservative voices but had been effective in shifting the election had taken hold.

We saw what that looked like this week. An undercover activist filmed Rep. Ken Buck (R-Colo.) telling an audience that Google’s anti-Republican bias had resulted in conservative websites being demoted in search results, a shift that could have swung 15 million votes.

There was an overhaul of Google’s search algorithm in May 2020, one aimed at prioritizing quality content. That change, Buck told his audience, meant that sites like the Washington Times were getting less traffic.

Of course, the idea that pushing the Washington Times down in search results would swing millions of votes is nonsense — as is the idea that downplaying the New York Post story was a central factor in Trump’s 2020 loss. Trump was viewed negatively by a majority of Americans, and millions came to the polls in order to send that message.

The idea that Google has such inordinate power that it can swing elections is rooted largely in a study from a researcher named Robert Epstein, who measured how vote preferences changed after participants were shown a biased set of search results. From there, he extrapolated out to a 15-million-vote effect. Of course, the idea that people primarily base their decisions on what they learn from Googling candidates — particularly at the presidential level — is dismissible on its face. President Biden got a record level of support from Democrats even as Trump earned near-universal support from Republicans. Which of them was influenced by his or her search results?

But, again, both because technology (and even terms like “algorithm”) can be opaque or confusing and because people are quick to attribute nefarious intent to powerful entities, these technology companies became a fertile way for Trump to foster that sense of victimhood with his supporters. Any weird behavior on a Facebook post was proof that the company was trying to lock out conservatives (instead of the reality, which is that Facebook is a central conduit for conservative political power and that the company is very responsive to the political right).

Recent polling from Fox News shows that most Americans, including many on the left, think that these technology companies have too much power. Most Republicans also think that Facebook and Google should be broken up into smaller companies.

Author and venture capitalist J.D. Vance, running for the Republican nomination for U.S. Senate in Ohio, told Time’s Molly Ball that this was something that came up repeatedly in his conversations with voters.

“Voters really want us to do something about the tech industry,” he told Ball, adding that people often tell him things like, “I love what you said, but why don’t we break up these companies and put all the CEOs in jail?”

One might normally be skeptical of such an anecdote, but Vance is necessarily interacting with people who self-select in going to hear a Senate candidate speak a year before the actual election, so who knows. This sense that Big Tech is oppressing conservative voices overlaps with one of the most prominent strains of the right-wing culture-war obsession: this idea that left-wing “cancel culture” seeks to silence any dissenters. Republicans are far more worried about “cancel culture” than Democrats (in part, certainly, because it’s been an obsession of Fox News’s). In a February poll, most Republicans said it was a very serious problem, and what better immediate example of grass-roots conservatives being muffled than that Twitter asks if you want to read an article before you tweet it?

Often, though, the actual intent of the social media companies’ efforts shine through. On Friday, Twitter removed white supremacist and Holocaust denier Nick Fuentes from its platform. Vance — who as a conservative White man was by no means the central target of Fuentes’s anger — used Twitter’s decision as an example of the company’s overreach. In doing so, he was suggesting that a private company choosing to muffle a racist was more problematic than a racist spreading racism. It’s seen as more politically useful to oppose Big Tech censorship than it is problematic to side with a white nationalist, which to some extent gets at the heart of all of this.

On more than one occasion, these decisions by social media companies have been cast as somehow running afoul of the First Amendment, as though they are extensions of the federal government to which that amendment applies. In fact, a lawsuit filed this week by Trump against the companies argues that they are essentially “state actors” to whom the amendment should apply. Unfortunately for Trump’s chances, this idea has already been rejected by the Supreme Court, in an opinion written by a justice Trump appointed.

Why is Trump filing this lawsuit now? In part, obviously, because he was banned from Twitter and Facebook after the violence at the U.S. Capitol on Jan. 6. In part, too, it’s because he still wants to be the face of the fight against his base’s oppressors. The lawsuit was structured in the manner of a class-action lawsuit, with Trump and his allies encouraging people to join in the former president’s fight, albeit most immediately by giving money or providing email addresses.

Cartoonist and Trump superfan Ben Garrison made a cartoon showing Trump battling the social media giants, casting Trump as Don Quixote and the tech firms as a windmill. The metaphor here doesn’t really work as intended, given the actual story in that book, which has Quixote attacking only invented enemies. But Garrison nonetheless defended his drawing in a post on his website, arguing that he was “free to take liberties with any metaphor I like.”

Embedded in the middle of that post? A fundraising appeal, asking people to “[h]elp us expose the Big Tech’s Censorship of Americans.” The victimhood at play is broadly useful.