Do The Twitter Files Show First Amendment Violations?
When is corporate censorship a constitutional problem?
“Of course governments want to shape and control the public conversation, and will use every method at their disposal to do so, including the media.”
― Jack Dorsey
“Private speech isn’t free speech.”
― someone on Twitter, probably
In my lifetime, I have seen the Democratic Party go from the party of free speech and fear of corporate power to the party of “private corporations can do what they want.”
The Republican Party, for its part, has always subordinated free speech to social etiquette, orthodoxy, and order. With the Left’s recently growing distaste for open discourse, the Right seized the opportunity to give itself a PR makeover, but its commitment to free speech beyond Republican-friendly speech hasn’t much changed.
And so now both parties LARP as the party of free speech when it suits their culture war goals, but neither believes in free speech as a broadly-applicable principle. This is a precarious position—we are left with no major political force truly in support of free speech.
Free Speech as a Culture War Flashpoint
As free speech has risen to become a preeminent battlefront in the culture war, the biggest and loudest theater being contested is social media. I have argued at length that social media isn’t a marketplace of ideas but a marketplace of engagement with more perverse incentives at odds with truthseeking. Despite this, social media remains the most pivotal hub of the modern public square. Until we fix that, it’s what we’ve got. Social media is where millions of Americans engage in discourse every day. It’s where traditional media determines public sentiment (which in turn drives the focus of news). It’s where the journalists working for traditional media build and maintain their “professional brand.” And it’s where academics cultivate a public image.
In sum, all of our traditional major sensemaking institutions are at this point deeply reliant on—and shaped by!—social media.
And so, for anyone with something approximating a traditional liberal view on free speech, the growing popular calls for mass censorship on social media (and platforms eagerly answering those calls) are alarming. The Twitter Files, an ongoing effort by journalists Matt Taibbi, Bari Weiss, and Michael Shellenberger to report on the contents of internal Twitter documents selected by Elon Musk to leak, brings the issue front-and-center.
In turn, responses from the Left that “they’re a private company, they can do what they want” are alive and well. The argument often seems to go hand-in-hand with an almost comical conflation of free speech and the 1st Amendment. “What about free speech?” a commenter will ask, to which the responses comes in droves, “It’s a private company, there is no free speech issue.”
Of course, free speech is a centuries-old collection of ideas in philosophy and political science concerned with the protection of the individual’s right to freely think and communicate. These ideas predate the 1st Amendment by hundreds of years, and they were never limited only to concerns of government censorship. Corporate power and the tyranny of mobs have historically been just as dangerous to free thought.
But these avenues of censorship aren’t illegal, they’re just bad for the free exchange of ideas. For conduct to violate Americans’ constitutional rights, we need what the courts refer to as “state action.”
What the Law Actually Says
“Congress shall make no law . . . abridging the freedom of speech.” So begins the 1st Amendment of the U.S. Constitution. These words are the entirety of the statutory text constitutionally guaranteeing free speech against government imposition.
If anyone tries to tell you 1st Amendment law is simple, though, they’re either clueless or lying. Hundreds of cases in the judiciary have fleshed out the contours of constitutional speech protections. Too much ink has been spilled on the subject to cover it all, but suffice to say the law is far more developed than the single, solitary sentence of the 1st Amendment would suggest. For example, the 1st Amendment doesn’t just apply to laws, but to any government action. It applies to all levels of government, from local to federal.
And most critical to our discussion, it can apply to private actors working with the government or fulfilling a “traditional governmental role.” As the Supreme Court explained in Edmonson v. Leesville Concrete Co., Inc., 500 U.S. 614 (1991),
"Although the conduct of private parties lies beyond the Constitution's scope in most instances, governmental authority may dominate an activity to such an extent that its participants must be deemed to act with the authority of the government and, as a result, be subject to constitutional constraints."
The test developed by the courts is known as “state action doctrine,” and it implicates the actions of private actors in three main ways:
(i) the private entity performs a traditional, exclusive public function;
(ii) the government coerces or compels the private entity to take an action; or
(iii) the government acts jointly with the private entity.
We can ignore (i) because the meaning of “traditional, exclusive public function” is very narrow and courts would not apply it here. But both (ii) and (iii) are potentially implicated, and so we should examine the Twitter Files with both in mind.
What The Twitter Files Show So Far
It bears emphasis that we still don’t know the full contents of the Twitter Files. The “Twitter Files” story is being released in successive pieces by different journalists who have been given access to the Files. To date, the Twitter Files consist of:
1: The Twitter Files (by Matt Taibbi)
2: Twitter’s Secret Blacklists (by Bari Weiss)
3: The Removal of Donald Trump Oct 2020-Jan 6, 2021 (by Matt Taibbi)
4: The Removal of Donald Trump Jan 7 (by Michael Shellenberger)
5: The Removal of Donald Trump from Twitter (by Bari Weiss)
I won’t cover everything released so far, but I will provide some highlights relevant to our 1st Amendment lens:
Insiders can use backchannels with Twitter to get content banned. Taibbi describes the practice as “constant.” There are examples in the Files of both the Trump Administration and the Biden campaign (before his presidency) successfully lobbying to censor content. According to Taibbi, “[t]his system wasn't balanced. It was based on contacts. Because Twitter was and is overwhelmingly staffed by people of one political orientation, there were more channels, more ways to complain, open to the left (well, Democrats) than the right.” This is true. In fact, according to Shellenberger, “In 2018, 2020, and 2022, 96%, 98%, & 99% of Twitter staff's political donations went to Democrats.”
Twitter took unprecedented steps to censor the Hunter Biden Laptop story, including deleting links, banning accounts, and preventing users from DMing the story. Twitter’s position was that it violated the “hacked materials” policy, but internally the company acknowledged it was mere speculation that the materials were hacked and the rule was used as pretense.
Twitter had a secret blacklist of users whose posts were prevented from trending. Twitter previously publicly stated it didn’t do this. One user blacklisted was Dr. Jay Bhattacharya of Stanford who was vocal about policy concerns related to school closings during covid. Dr. Bhattacharya was not an outlier—the blacklist consistently skewed toward silencing voices critical of the public health response to covid.
A secret group called the Site Integrity Policy, Policy Escalation Support (SIP-PES) handled higher profile censorship within the company. Internal documents show the group acknowledging there were no policy reasons for bans they were handing out. Banning decisions would start with Tweets the SIP-PES disagreed with, and then the group worked backwards to find a policy justification for censoring.
Fusing automated and manual moderation, Twitter also employs ad hoc bots, which it can “spin up” with certain rules to monitor Twitter activity and take specified actions based on the specified rules.
Twitter moderation shifted from a mix of rules-based automated moderation with some subjective decisions made by the SIP-PES to, by around the time of January 6, a more ad hoc system based on more subjective censorship criteria and increasing direct involvement with “federal agencies.”
Those federal agencies were the FBI, DHS, and the Office of the Director of National Intelligence, and Twitter moderation met with them secretly on a weekly basis. Twitter employees involved in these meetings hid the nature of the meetings on their calendars, for optics. In public-facing comments, Twitter vaguely credits “partnerships” used to help determine content for censorship, in lieu of naming the FBI or the DHS, also for optics. Employees understood this should not be public knowledge.
In conjunction with clandestine government “partners,” Right wing Tweeters voicing concerns over election integrity were routinely targeted, whereas similar Leftwing Tweeters voicing concerns over election integrity were given cover. This appears to have happened, at least in part, while Trump was still in office.
Jack Dorsey resisted many of Twitter’s incremental shifts toward more censorship, but his staff continued to work on him. During and after January 6, while Dorsey was on vacation, staff were pushing ad hoc rules changes specifically to try and ban Trump permanently. 300 Twitter employees signed an open letter pressuring Dorsey to ban Trump, which was published by the Washington Post. Despite these calls, internal discussions repeatedly affirmed that Trump was not violating Twitter’s Terms of Service. Of course, employees were ultimately successful in getting Trump banned. In contrast, other world leaders who unequivocally incited specific acts of mass violence have never been banned from Twitter.
Assuming There is Government Action, Does it Violate the 1st Amendment?
With these facts in mind, before we turn back to the state action doctrine, we have to answer a threshold question: even if there were government action, would the kind of moderation detailed in The Twitter Files violate the 1st Amendment?
Put simply: yes. Under the 1st Amendment, government cannot censor speech just because it thinks the claims are false. Government cannot censor speech because it thinks the claims will threaten public health. And government cannot censor speech because it’s worried some people will be inspired to break the law. (The closest legal analog here is “incitement to imminent lawless action” and that is a very high threshold not met by any of the content so far discussed in the Twitter Files).
No 1st Amendment attorney worth their salt would tell you the government could legally engage in the kind of censorship we’ve been able to glimpse from the Twitter Files. And so now to the $64,000 question: was the government engaged in censorship? Or can this conduct only be ascribed to Twitter?
Did the Government Coerce Twitter to Censor?
Under the “compulsion test,” conduct is only deemed state action where the government “has exercised coercive power or has provided such significant encouragement, either overt or covert, that the choice must in law be deemed to be that of the State.”
The best argument I can give supporting coercion is that social media’s legal protections under Section 230 have been a hotly debated public issue. Democratic lawmakers (like Elizabeth Warren) have campaigned on the need to regulate social media more. The companies may believe cooperating with the government on issues of censorship will save it from more comprehensive regulations, which threaten the companies’ profits.
Still, so much of the Twitter Files suggests a strong internally-motivated push for the kinds of censorship coming out of Twitter’s “partnerships.” This is by no means the explicit legal test, but I think of it as the standard for legal entrapment—the question is “without government influence, would they have still done what they did?”
The answer here, at least in most cases, appears to be yes. In other cases, Twitter wouldn’t have censored simply because it would not have known what to censor. In that regard, the relationship between Twitter and government does seem more like a “partnership”—the government points, and Twitter shoots. On the whole, the Twitter Files paint the picture of a back door alliance between Twitter censors and the parts of government ideologically aligned with them. Which leads us to the last prong…
Did the Government and Twitter Act Jointly?
The “joint action test” requires that “the state has so far insinuated itself into a position of interdependence with the private entity that it must be recognized as a joint participant in the challenged activity.”
If there is state action here, it’s under the joint action test. And facts unearthed so far support a joint action theory. For example, there are direct, concrete examples of the Trump Administration using backchannels within Twitter to censor content. We see the same with the Biden campaign, although I didn’t see examples in the Twitter Files of the Biden Administration doing so after he took office. (More on that below.)
More glaring are the weekly meetings between Twitter moderation team members and the FBI, DHS, and the Office of the Director of National Intelligence. The Twitter employees involved knew to keep these meetings secret, and it’s not hard to see why: at the very least, it creates the appearance of active government involvement in censorship.
What’s Missing From the Twitter Files?
To go further, we would really need to know more about what happened at those meetings, and that information is not likely forthcoming in future Twitter Files releases. There would need to be investigations, but by whom? What arm of federal law enforcement isn’t potentially compromised in this, I’ll say it, conspiracy? A massive government-allied corporation with direct, unfettered access to controls on the volume and reach of millions of Americans’ speech, who hears what and who gets to speak at all—this sounds like law enforcement’s wet dream.
We also don’t know how long this censorship partnership lasted, or if it’s ongoing. Glaringly missing from the Twitter Files is any discussion around censorship in the present, as Elon Musk captains the ship. The reason this is glaring, of course, is because Musk leaked the Twitter Files, and he almost certainly would have chosen which files to leak (and which not to leak) with intentionality.
For this last point, we all need to see through the act. Musk is not a principled ideological actor championing free speech. For all the good he has done, he is still a self-interested billionaire who managed to build the world’s (second) largest fortune. That doesn’t happen by accident. He is smart, sophisticated, and in recent years has become increasingly adept at riding culture war waves to increase his publicity and, in turn, his bottom line.
Whether Musk leaked the Twitter Files simply to distance himself from the previous, unabashedly pro-Democratic version of Twitter and is still actively “partnering” (read: conspiring) with federal agencies to enact massive censorship campaigns is a question we can’t yet answer. But maybe there are hints out there in the public domain. Maybe I’ll leave you with one straight from the horse’s mouth:
