There is no "censorship industrial complex"

Biased and misguided online censorship is a genuine problem. However, popular claims about a sinister "censorship industrial complex" are hysterical, conspiratorial, self-serving, and dangerous.

“I believe that we actually do have a threat to democracy in this country, but unfortunately, it’s not the threat to democracy Kamala Harris and Tim Walz want to talk about. It is the threat of censorship!” - J.D. Vance

In recent years, there has been a heated culture war over online misinformation, censorship, and free speech. As Jemima Kelly observes in the Financial Times, this conflict often takes the form of a simplistic shouting match of “‘misinformation bad!’ from the left, and ‘censorship worse!’ from the right”.

In fact, framing things in terms of left and right is understandable but misleading. It would be more accurately described as a conflict in Western countries between what you might call the “liberal establishment” (what Nate Silver calls “the Village”) and various anti-establishment forces. Although the former leans progressive on social issues, it is defined not primarily by ideology but by allegiance to the most prestigious institutions of modern liberal democracies (e.g., elite mainstream media, academia, government, the arts, etc.). And although the latter includes right-wing populist movements (e.g., Trump, Brexit, etc.), it also includes anti-establishment figures who defy easy political categorisation (e.g., Glenn Greenwald) and excludes establishment conservatives (e.g., Mitt Romney, Rishi Sunak) and their supporters.

Misinformation bad!

I have been critical of the liberal establishment’s emphasis on “misinformation bad!”. Of course, misinformation is bad. Nevertheless, since 2016, the year of two populist revolts (Brexit and then Trump), there has been a sustained moral panic over its prevalence and impact, treating online misinformation as a leading societal threat and the driving force behind populism, attacks on democracy, declining trust in institutions, and much more.

Against this narrative, research consistently shows that clear-cut online misinformation like fake news is relatively rare and mostly consumed by a minority of extremists and conspiracy theorists who seek it out because it aligns with their pre-existing beliefs. In other words, such content is largely symptomatic of the problems (e.g., anti-establishment attitudes and institutional distrust) it is supposed to explain.

Nevertheless, such findings have done little to temper alarmism about online misinformation. Partly, this is because of an unfortunate concept creep in which terms like “misinformation” and “disinformation” are increasingly used to cover a large and expanding category of uncongenial content, including problematic opinions, accurate information carrying the wrong message, and satirical memes.

However, it is also because of the enduring appeal of misinformation alarmism. As with all good narratives, such alarmism is rooted in an important grain of truth: if misinformation is defined extremely broadly, it obviously is and always has been a real problem. Moreover, the claim that populist movements are rooted in online misinformation coheres with the widespread (false) belief that those who support such movements are gullible fools, and it explains such movements in the most self-serving way possible—not in terms of any connection to legitimate grievances or establishment failures, but as a simple result of infection by bad information.

Censorship bad!

In writing criticisms of misinformation alarmism and the anti-misinformation industry, I have been aware that the arguments partly align with those advanced by pundits like Matt Taibbi, Glenn Greenwald, Michael Shellenberger, and Jacob Siegel. These writers are most responsible for articulating and defending the “censorship bad!” side of the modern culture war.

Nevertheless, I have never wanted to associate my views or arguments with theirs. Partly, this is for ideological reasons. Whereas they are all staunchly anti-establishment, I am not. For all the faults and failures of modern liberal institutions—and there are many such problems and much self-serving denial over them—they constitute extraordinary historical achievements, and anti-establishment movements on the left and right are much worse than what they seek to replace.

More concretely: Yes, modern academia and, say, the New York Times are very far from perfect, but their flaws pale in comparison with the torrent of self-serving lies and demonstrable falsehoods that come from figures like Trump, Elon Musk, and Tucker Carlson.

However, the deeper reason I have not wanted to associate my arguments with those within the “censorship bad!” discourse is simply that the most influential writings and commentary about a sinister “censorship industrial complex” involve extreme exaggerations and misrepresentations, low-quality reporting, smear campaigns, and cherry-picking.

These flaws are bad enough on their own, but hysterical discourse about “totalitarian” censorship is being used to justify Trump's dangerous lie that the 2020 presidential election was fraudulent and his claim that the real threat to American democracy comes from the Democratic Party’s support for online censorship.

In fact, one of the most remarkable things about the “censorship bad!” discourse is that the very people who complain about the liberal establishment’s alarmism about online misinformation advance far more hysterical and unsupported claims about the prevalence and impact of online censorship.

Censorship alarmism

Consider, for example, a highly influential essay by Jacob Siegel, in which he calls “the counter-disinformation complex” an “industrial-scale censorship machine”, part of a “totalitarian system” orchestrating the “hoax of the century”, and

“one of the most powerful forces in the modern world… which the government uses to direct a “whole of society” effort that aims to seize total control over the internet and achieve nothing less than the eradication of human error.”

Similarly, Michael Shellenberger, who coined the term “censorship industrial complex”, suggests that “if you read 1984…. it is eerily prescient to what we are facing today” and that “the true peril of our current moment… is that we are headed toward totalitarianism”.

Echoing this, Matt Taibbi declares that when it comes to

“risks to the First Amendment… That battle is lost. State censorship is a fact in most of the West… The endgame is not controlling speech. They’re already doing that. The endgame is getting us to forget we ever had anything to say.”

Not to be outdone in alarmism, Mike Benz, whose “reporting” (see below) is cited more than any other in discussions about the censorship industrial complex, alleged that online censorship in America has involved a “scale of censorship the world has never experienced before.”

Today’s article

As I will show in this article, these claims and countless similar examples of alarmist discourse about online censorship are preposterous exaggerations reliant on falsehoods, ignorance, low-quality reporting, and conspiracy theorising. (To keep things manageable, I will restrict my focus to the USA).

Areas of agreement

First, though, it is important to flag some areas of agreement with the leading pundits of the anti-censorship movement.

I have already mentioned one: like Taibbi, Shellenberger, and Greenwald, I think the liberal establishment greatly exaggerates the impact of online misinformation and the gullibility of ordinary people.

Similarly, I am strongly opposed to censorship.

In addition to the many benefits of a culture of free speech, the most dangerous forms of misinformation typically come from elites, which should make us sceptical of handing such elites power to censor misinformation.

Moreover, even when official views (e.g., concerning the safety and efficacy of vaccines) are very well-supported by evidence, censorship of dissenting information tends to erode trust in establishment institutions, exacerbating the very problems it aims to solve. This would be true even if censorship decisions were perfectly reliable, but of course censorship is always enforced by fallible and biased human beings in ways that make error and partiality inevitable.

I also agree that the liberal establishment is bad on this issue. For complex and depressing reasons, liberals have come to place less value on free speech in recent years, at least in rhetoric. As I return to below, this is partly a misguided but understandable response to the existence of dangerous falsehoods about topics like election fraud, vaccines, and immigration. However, it also reflects the fact that free speech tends to favour anti-establishment views and movements. As liberalism has become nearly hegemonic within many establishment institutions (e.g., mainstream media, science, universities, public health, etc.), many liberals have therefore come to value free speech less.

Finally, I agree that objectionable forms of online censorship have occurred and continue to occur, and that such censorship has disproportionately targeted the falsehoods and mistaken narratives associated with anti-establishment populism.

Clarifying “censorship”

Admittedly, discussions about censorship are complicated by the fact that censorship comes in different forms, and anti-censorship pundits use the term even more promiscuously than establishment liberals use the term “misinformation”.

One important distinction is between state or government censorship (i.e., censorship enforced by government representatives or agencies), which would violate the First Amendment, and private censorship (e.g., censorship enforced by social media companies), which would not.

Another distinction concerns the kind of censorship in question. Social media companies often rely on a “remove, reduce, inform” framework. Removing content is a clear example of censorship. Intervening to reduce people’s exposure to content without removing it (typically justified by the slogan “freedom of speech is not freedom of reach”) could also reasonably be characterised as censorship. However, it is far less serious than the outright removal of content. To mark the distinction, I will call such removal “hard censorship” and the mere reduction in people’s exposure to content “soft censorship”.

“Inform” refers to cases where social media companies deploy counter-speech, adding a label or context to claims (e.g., tweets, images, videos) deemed misleading. Such policies and their enforcement might be biased and objectionable in many ways, but (contra Shellenberger) do not amount to censorship on any reasonable interpretation of that term.

The “censorship industrial complex” thesis

With these distinctions in mind, what are the main complaints advanced by those writing about a “censorship industrial complex”?

Private censorship

First, there are many complaints about private censorship by major social media platforms. Although such complaints take different forms, they overwhelmingly focus on the idea that such platforms have employed hard and soft censorship in ways that favour the liberal establishment and disproportionately target anti-establishment views.

In defence of this, anti-censorship pundits point to various examples, including: Twitter’s “shadow banning” (i.e., soft censorship) of accounts like Covid contrarian Jay Bhattacharya and right-wing firebrand LibsofTikTok; the decision to remove Donald Trump from major social media platforms in the aftermath of January 6th; Meta’s temporary banning of any discussion of the idea that SARS‑CoV‑2 escaped from a Wuhan lab; and the decision by major social media companies to temporarily remove or reduce exposure to a New York Post story about the contents of Hunter Biden’s laptop in the run-up to the 2020 presidential election.

Such decisions are well-documented. Moreover, I think almost all such decisions were wrong, both because of general problems with censorship (even when it targets demonstrably false and harmful content) and because in some of these cases (e.g., the lab leak hypothesis, the New York Post story, and the arguments of Covid contrarians) the censored content was either accurate or debatable.

Nevertheless, although it is reasonable to object to biased and misguided censorship by major social media companies, they are ultimately private companies that should be free to set whatever content moderation policies they want. The argument that they are “public squares” that should be treated as public utilities is unconvincing, especially in a thriving marketplace where different social media platforms (Truth Social, Substack, 8chan, Rumble, etc.) enforce very different policies.

Given this, even if one thinks (as I think) that the censorship decisions enforced by major social media companies have often been objectionable, biased, and misguided, this judgement would not license alarmist claims about a “censorship industrial complex”.

Government censorship?

In response, those who posit the existence of this complex raise a second complaint: that government organisations (e.g., the FBI, CIA, and CISA) and representatives (e.g., presidential advisors) have pressured social media companies to censor content.

It is undeniable that government agencies and representatives have engaged in extensive contact with such companies requesting censorship decisions. For example, the Twitter Files documented that the FBI frequently sent Twitter names of accounts alleged to be spreading misinformation, and it also revealed extensive communication between both the Trump and Biden administrations and the company.

Moreover, in many such cases, this communication clearly involved “pressure” on any reasonable interpretation of that term. To take only one example, in April 2021, Meta’s head of global affairs, Nick Clegg, reported a conversation with a senior adviser to Joe Biden who “was outraged—not too strong of a word to describe his reaction—that we did not remove this post,” referring to a Leonardo DiCaprio meme with an anti-vaccine message.

Once again, it is entirely reasonable to complain about such things. Nevertheless, nobody writing about the “censorship industrial complex” has provided any evidence that government representatives have demanded or coerced such platforms to agree with their censorship requests, which would (except in extremely rare circumstances) violate the First Amendment.

In fact, as the quote from Clegg and much other reporting in the Twitter Files and elsewhere illustrates, companies often resisted government requests.

Once again, it is therefore very difficult to see anything like a “censorship industrial complex” at play here.

Nevertheless, anti-censorship pundits have responded to this objection by positing various sinister ways in which government agencies and representatives have covertly influenced censorship decisions by social media companies—for example, by funding or directing non-governmental organisations to censor content, by lying that certain things are Russian disinformation, and by deploying secret CIA agents among prominent academics researching misinformation.

In defence of such allegations, such pundits repeatedly identify two examples of allegedly sinister government involvement in censorship: the Election Integrity Partnership and the censorship of the New York Post’s story about Hunter Biden’s laptop.

However, their reporting on these topics involves misrepresentations, exaggerations, conspiracy theorising, and baseless smear campaigns.

The Election Integrity Partnership

In Michael Shellenberger’s congressional testimony about online censorship, he describes the Election Integrity Partnership (EIP) as “the seed of the censorship industrial complex”.

The EIP was a coalition formed in 2020 during the Trump presidency. It aimed to monitor and respond to election misinformation harmful to the democratic process, focusing specifically on misinformation and rumours designed to suppress voting, reduce participation, confuse voters, or delegitimize election results without evidence. It ultimately involved Stanford University’s Internet Observatory, the University of Washington, the National Conference on Citizenship, the Atlantic Council’s Digital Forensic Research Lab, and Graphika, a social media analytics firm.

The allegations

According to claims and insinuations by Shellenberger, Taibbi, and others writing about the censorship industrial complex, the EIP essentially channelled censorship requests from the government and federal agencies to platforms, thereby providing a means by which the government and federal agencies could get around the First Amendment and censor disfavoured content.

In defence of this allegation, they point to the fact that the founders of the EIP had consulted with the Department of Homeland Security’s Cybersecurity and Infrastructure Security Agency (CISA) before their launch and that CISA was a project partner. Moreover, drawing on reporting by Mike Benz, Taibbi, Shellenberger and many others have alleged that the EIP got nearly 22 million tweets labelled “misinformation” during the 2020 election. Shellenberger has also suggested that Renée DiResta, an influential misinformation researcher and the technical research manager at the Stanford Internet Observatory during the EIP project, is a secret CIA agent.

Misleading reporting

It is difficult to overstate just how misleading this narrative is.

First, one thing omitted in almost all the claims about the EIP by anti-censorship pundits is a fact one might think is highly relevant: Before, during, and after the 2020 presidential election, Trump and his supporters launched an aggressive disinformation campaign to delegitimize an election he lost, one of the most shocking, authoritarian, and anti-democratic attacks in American history.

As I have already noted, I do not think the existence of such disinformation campaigns justifies censorship. Nevertheless, it is understandable that government agencies would be interested in such a dangerous attack on American institutions and why researchers would want to monitor, study, and combat it.

If I were submitting official testimony about a project designed to monitor and respond to misinformation surrounding an election, it would feel dishonest not to mention that the project responded to the existence of dangerous election misinformation. In fact, I would struggle to look at myself in the mirror if I omitted that fact in testimony invited by Jim Jordan, someone who has systematically spread Trump’s lies about the 2020 election and who tried to prevent the certification of the election result.

Second, the EIP did not have the power to label any tweets or social media posts as “misinformation”. Instead, the project involved extensive research on the prevalence of election misinformation and could, at most, notify social media companies of content on the platform alleged to violate those platforms’ own policies, something which anybody has the freedom to do.

Third, the EIP did not notify Twitter of 22 million tweets. Instead, in research after the 2020 election, they published a report that counted the total number of tweets they judged to be involved in misleading claims about the election, which summed up to 22 million. During the election campaign, the team had only tagged 2,890 tweets as violating Twitter’s policies. Because the EIP had no authority, Twitter only took action on 65% of those tweets, just 13% of which were removed.

It is worth pausing on this point: Not only did the anti-censorship pundits misrepresent the power of the EIP, implying it could censor or label content when it could not, but they over-estimated the number of attempts to get content classified as misinformation by over 21,990,000.

When this was pointed out to Matt Taibbi, he at least had the good sense to remove the misinformation he had placed in a Twitter Thread seen by millions of people. In contrast, Shellenberger doubled down.

Fourth, although the EIP was formed after consultation with CISA, a project partner, there is no evidence that CISA founded, funded, or controlled the EIP or sent content to the EIP to analyse or flag to social media platforms.

Finally, what about the extraordinary allegation that Renée DiResta, an influential misinformation researcher who was part of the EIP, is a secret CIA agent? The only evidence Shellenberger provides is that she had done an undergraduate student fellowship at the CIA, ending in 2004.

Overall, the quality of the reporting here—the falsehoods, misrepresentations, exaggerations, and conspiracy theorising—is frankly appalling. Most of it derives from the work of Mike Benz, someone Shellenberger claims (mistakenly) was “head of cyber at the State Department”, who he has not found “making a single unsubstantiated accusation, much less peddling conspiracy theories.”

Legitimate grievances

To be clear, these problems with the anti-censorship pundits’ reporting on this project do not mean there is nothing to object to in the EIP’s operations. As I have written about extensively, there are many problems with what Joseph Bernstein calls “Big Disinfo”, a loose collection of researchers, journalists, and non-governmental organisations throughout the West focused on studying and combatting online misinformation, and some of these problems might also characterise the EIP’s work. (I have not looked into it enough to know).

Although Big Disinfo is partly a response to real dangers, it functions too much like an intellectual monoculture, one which is often naive about the challenges of figuring out the truth in a complex world and frequently produces politically biased and low-quality work. For example, although there is an enormous amount of focus and attention on misinformation associated with conservatives and populists, there is almost none focused on false or misleading narratives peddled by the liberal establishment.

Moreover, it is understandable that Big Disinfo’s close interactions with governments and social media companies make many people uncomfortable. If one imagined something like the EIP focusing specifically on areas where misinformation is peddled by progressives—for example, misinformation used by anti-racist protestors to justify protests and riots, climate alarmist misinformation used to support bad climate policies, or misinformation about the prevalence and impact of misinformation peddled by mainstream media—progressives would be furious.

In other words, one can have many legitimate concerns about initiatives like the EIP. The point is that such concerns have nothing to do with the misinformed and conspiratorial alarmist narratives about a “censorship industrial complex” advanced by Benz, Shellenberger, Taibbi, and others.

Hunter Biden’s Laptop

Three weeks before the 2020 presidential election, the New York Post published a story covering emails from a laptop allegedly left by Hunter Biden at a Delaware computer shop. According to the Post, information from the laptops revealed corruption by Joe Biden.

Not only did most other mainstream news outlets decline to publish the story due to concerns about its veracity, but both Twitter and Facebook acted to censor it based on fears that it involved Russian disinformation, in Facebook’s case by reducing exposure to the story. These fears were shared by a group of 51 former senior intelligence officials, who published an open letter claiming that the emails reported in the Post had “all the classic earmarks of a Russian information operation”, although they also stressed that they “do not know if the emails… are genuine or not and… do not have evidence of Russian involvement.”

Their fears have turned out to be unsupported: no evidence of Russian involvement has emerged, the laptop seems genuine (the FBI had already confirmed this in late 2019), and forensic analysis has authenticated many of the emails from it.

There is now an almost universal consensus that the decisions by Twitter and Facebook to censor the New York Post story was misguided, including by those in charge of such companies when the decisions were made.

In fact, it provides an almost perfect illustration of the extreme risks and pitfalls of a censorship approach to dealing with uncertain threats of disinformation. I would bet that the significant amount of anger, distrust, and polarisation caused by this bad censorship decision outweighs any conceivable benefits of attempts to censor Russian disinformation in recent years (not least because there is little evidence such disinformation efforts have had any significant impact in the first place).

Nevertheless, does the episode provide evidence of a sinister censorship industrial complex? According to Shellenberger, Taibbi, Siegel, and many others, it does.

Whether they are correct matters. For example, Donald Trump routinely cites this episode of censorship as unambiguous evidence of election interference that undermines the legitimacy of the 2020 election result. And when J.D. Vance was recently asked whether Trump lost this election, he responded with a question: “Did big technology companies censor a story that independent studies have suggested would have cost Trump millions of votes?”

Private censorship ≠ a censorship industrial complex

There are two problems with this narrative.

First, there is no compelling evidence that online censorship of the New York Post’s story cost Trump the 2020 election, any more than there is compelling evidence that Russian disinformation cost Hilary Clinton the 2016 one.

Second, although it is legitimate to criticise bad censorship decisions by influential private companies, such decisions do not qualify as “election interference”. And in fact, in Taibbi’s initial reporting on the story as part of the Twitter Files, he explicitly acknowledges that he found no evidence of government involvement in Twitter’s actions.

Given this, those who use the story as an example of a sinister “censorship industrial complex” are, once again, forced to rely on conspiracy theories about covert influence operations by government agencies.

The primary theory in this context is that the FBI and perhaps other government agencies deliberately lied about the laptop as a way of getting social media companies to censor it. In defence of this theory, pundits mostly point to the following facts:

• The FBI knew about the laptop and its authenticity since 2019.

• The FBI and other government agencies warned social media companies to be on guard about the possibility of a Russian information operation involving hacking and leaking materials.

• The FBI did not alert anyone to the fact that the laptop was authentic when the Post published their story.

• Highly influential former intelligence officials, some of whom had worked at the FBI, published an open letter suggesting the laptop’s materials looked like the result of a Russian information operation.

Evaluating the conspiracy theory

I am not one of those people who rejects conspiracy theories out of hand. Historically, government agencies in the USA, like the FBI and CIA, have routinely conspired to act in harmful and objectionable ways, and it is completely conceivable that they continue to do so.

Nevertheless, this specific theory is very implausible.

First, the theory asks you to believe the following story: the FBI learned about the laptop in December 2019, decided that it did not want the American public to know about its contents, and then conspired to prime social media companies to be vigilant for Russian disinformation in ways that would lead them to censor any story that focused on the laptop.

Even if one assumes that the FBI wanted Biden to win the election and would conspire to bring this outcome about, this story involves pretty weird motivations, actions, risks, and calculations—and for what? To somewhat reduce exposure to a story about Hunter Biden’s laptop that is not particularly damaging to Joe Biden? (In fact, the decision to censor the story likely increased attention to it, as censorship often does, but the conspiracy theory asks us to imagine that the FBI were too dim to anticipate that possibility).

Second, the conspiracy theory omits numerous relevant facts. Most obviously, Russia really does try to influence American elections with disinformation campaigns and “hack and leak” operations. That happened very prominently in 2016, including when a Russian cyber attack led to the release of emails by John Podesta. Given this, even in the complete absence of a conspiracy, both the FBI and social media companies were always going to be on extremely high alert about the possibility of Russian information operations.

Third, the truth about the Hunter Biden laptop story is bizarre. The son of a presidential candidate dropped off a laptop with lots of self-incriminating material to a random computer shop owner in Delaware, who could not confirm that the person who had dropped it off was Hunter Biden because he is blind. And then the laptop somehow finds it way into the possession of Rudy Giuliani, who hands it over to an extremely unreliable right-wing tabloid newspaper for use in an October surprise to shift the election in Trump’s favour, an action which coincidentally aligns with Putin’s geopolitical goals.

That is a strange and suspicious chain of events. I find it easy to understand why smart, well-meaning people would find it more plausible that the emails resulted from a Russian hack and leak operation.

Finally, although it sounds bad that the FBI knew the contents of the laptop were authentic from 2019, it sounds less bad when you remember the following facts:

• The FBI is a massive organisation with tens of thousands of workers and many departments.

• There is no evidence that any representative of the FBI ever told social media companies that the Post story is fake or instructed them to censor it.

• The people who wrote the open letter suggesting that the Post story looked to be based on Russian influence operations were former intelligence officials not working for the FBI (or other government organisations).

• The FBI must obey laws that often prohibit it from publicly discussing ongoing investigations, which would have likely prevented it from disclosing that it knew the laptop was real when the Post story was published.

For these reasons, the influential conspiracy theory about the FBI’s covert involvement here seems very unlikely to be accurate.

Why any of this matters

Claims about the Election Integrity Partnership and Hunter Biden’s laptop are the main pieces of evidence that pundits appeal to in claims about a sinister “censorship industrial complex”. Given that such claims rest on exaggerations, misrepresentations, low-quality reporting, and unfounded conspiracy theorising, we should conclude that there is no sinister “censorship industrial complex”.

Why does that matter?

As I have already noted, I agree with anti-censorship pundits that there are genuine problems with the liberal establishment’s alarmism about online misinformation and with much of Big Disinfo’s work to combat it. Moreover, I also agree that because liberals tend to be bad on issues of free speech, it is important to criticise possible future moves towards more censorious policies, either by governments or social media companies.

Nevertheless, such legitimate points should not provide a license to characterise censorship and Big Disinfo in evidence-free, conspiratorial, and alarmist ways. In fact, the biased, low-quality, and frequently nasty character of writings by anti-censorship pundits simply simply discredits genuine anti-censorship concerns. At least one reason many liberals do not take such concerns seriously is that the most influential complaints about online censorship are fundamentally unserious.

The market for anti-censorship discourse

Hysterical discourse about a totalitarian censorship industrial complex also has more sinister consequences. This discourse did not arise by accident, or through simple mistakes. There is a lucrative market for it.

Most obviously, it is highly appealing to those motivated to demonize and discredit establishment institutions and their supporters. This demonizing function is transparent in much of the discourse. For example, Jacob Siegel declares that “the censorship regime’s aim is not to censor or to oppress, but to rule”, and Matt Taibbi writes that the intellectual class making up the anti-misinformation industry “pee themselves at the sight of mean tweets. They have no honor, no belief, no poetry, art, or humor, no patriotism, no loyalty, no dreams, and no accomplishments.”

If you care about attention, fame, and Substack subscriptions, this simplistic demonizing discourse can be very appealing. For anyone who cares about truth and nuance, it is less so.

More concretely, hysterical narratives about censorship are a blessing to Trump, who repeatedly draws on exaggerated fears about censorship to defend the Big Lie that the 2020 election was stolen from him, and to characterise Democrats as the real threat to democracy, a claim that many others seem to endorse.

There lies the bitter irony of discourse about a “censorship industrial complex”: those who condemn censorship for its authoritarian suppression of truth are profiting from falsehoods that serve America’s most authoritarian politician.