NCLA Site Search

Abridging, Not Coercing, Is The First Amendment’s Yardstick for Speech Violations

March 25, 2024

Philip Hamburger, a professor at Columbia, is the CEO of the New Civil Liberties Alliance, which represents most of the individual plaintiffs in Murthy v. Missouri.

Hamburger wrote the following post in response to a post at Volokh by Ilya Somin:

Is coercion the First Amendment’s measure of the freedom of speech? In commenting on Murthy v. Missouri, Prof. Ilya Somin takes the view that “coercion is the appropriate standard.” To this he merely adds that “veiled, but credible threats of retaliation by government officials qualify as such coercion.”

So far does he take this emphasis on coercion that, from his perspective, there is no First Amendment violation even when the government uses “significant encouragement” to get the private party to concede “active control” over its speech decisions. In the absence of coercion (including credible threats of retaliation) he apparently sees no limit on the power government can exercise over speech if it gets consent.

1. Abridging vs. Prohibiting

The First Amendment, however, has something to say about coercion. Prof. Somin recognizes the argument I make in Courting Censorship, that the First Amendment bars government from “abridging” the freedom of speech, and thus bars reducing that freedom. But he fails to note that the amendment also bars government from “prohibiting” the free exercise of religion. The amendment’s contrasting uses of abridging and prohibiting are meaningful. Id, at 254.

The contrast reveals that Prof. Somin’s coercion argument misattributes to free speech the standard that the amendment uses for free exercise. The word prohibiting seems to refer to various forms of coercion. So, when the First Amendment instead speaks of abridging the freedom of speech, it would seem to be adopting a different measure of government action for that right. The freedom of speech is violated by a mere reducing of that freedom, whether or not through coercion.

It is therefore unconvincing to suggest that coercion is the measure of freedom of speech. That contradicts the plain meaning of the First Amendment when it contrasts abridging and prohibiting.

Where did Prof. Somin go wrong? He gets to his coercion standard from the word “freedom,” arguing that if a private party “cut[s] back on speech voluntarily, the freedom of speech has not been abridged.” His theory seems to be that you have not had your freedom abridged if you give consent, and you are acting consensually unless you are coerced. Thus, although the First Amendment’s very words make clear that coercion is not the standard for freedom of speech, he reintroduces a coercion standard on the theory that it is the opposite of freedom and consent.

But does it make sense to introduce a coercion standard into a right when the Constitution carefully speaks of it in terms of abridging and contrasts that to prohibiting? Such reasoning defeats the Constitution’s words and meaning.

2. Abridging Was Part of the Traditional Understanding of the Freedom of Speech and Was a Deliberate Choice

In light of scholarship by Nicholas Rosenkrantz on the Constitution’s nouns, it seems time to attend to its verbs. Here, of course, I want to examine abridging and prohibiting.

Some verbs in the Bill of Rights are generic. To say, as in the Second Amendment, that a right shall not be “infringed” is to use a verb that could be used as to any right. But other verbs are specific to particular rights—for example, the Third Amendment’s “quartered.” That verb goes far in defining the substance of the particular right.

Tellingly, the verb “abridging” was not generic. For one thing, “it was a familiar locution to speak of abridging the freedom or liberty of speech or the press and to associate this with reducing or restraining the freedom.” Courting Censorship252. Moreover, that word and its contrast to prohibiting was a self-conscious choice:

The difference in the First Amendment between abridging and prohibiting was deliberate. In July 1789, the draft Bill of Rights contained adjacent paragraphs guaranteeing, in the first, religious rights and, in the second, speech, assembly, and petitioning rights—saying in each that the rights shall not be “infringed.” In early September, however, the Senate combined the two paragraphs. The resulting new paragraph barred Congress from making any law “prohibiting” the free exercise of religion or “abridging” the freedom of speech, or the press.

Id, 254. The word abridging was specific to certain rights and evidently was meant to be significant.

So, once again, it is a mistake to conflate the standard for free speech with that for free exercise. Whereas the verb prohibiting requires us to ask about degrees of coercion, the verb abridging demands that we ask whether there has been any reduction in the freedom. It is therefore improbable that the freedom of speech was meant to be measured by coercion.

3. Abridging Cuts Off Evasions

The Supreme Court’s overemphasis on coercion has invited censorship. As I explain in Courting Censorship, the doctrine laid out in Blum and other such cases leaves government confident that it can suppress speech simply by working not too coercively through private parties.

Indeed, much of the current censorship is cooperative:

There is much overlap in the censorship agendas of the government and the Platforms—so the government has exerted pressure only at the margins. That pressure is serious enough, for it concerns the material that is most crucial, politically and medically, and because it has suppressed millions of postings.

Id, 246. Thus, there is plenty of coercion, but at the same time the greater bulk of the censorship is not coercive, and thus (from the coercion perspective) seems outside the scope of the First Amendment.

It is clear, however, already from the Supreme Court’s cases on unconstitutional conditions that consensual arrangements can violate the First Amendment. In Trinity Lutheran v. Comer, even the meager temptation of a negligible amount of money was understood as prohibiting the free exercise of religion. Although there was not a chance that the church would abandon its faith for the piddling inducement, the Court recognized that the condition in that case came with “indirect coercion.” The lesson is that any penalty (even a mere 3 pence tax) on the free exercise of religion amounts to coercing and prohibiting.

Cooperative arrangements are all the more clearly covered when it comes to abridging the freedom of speech, because the relevant government action involves reducing the freedom, not necessarily prohibiting or coercing it. Under this standard, even when the government acts through entirely voluntary cooperation, without even a hint of coercion, it can still be abridging the freedom of speech. Indeed, even to adopt a law or policy reducing the freedom of speech, without any suppressive effects yet, runs afoul of the injunction against making any law abridging the freedom of speech. Courting Censorship, 256-58.

Abridging has the virtue of clarifying that the First Amendment bars all of the current evasions of the Court’s coercion model, including the genuinely cooperative arrangements to suppress opinion.

Consider the government’s coordination of private censorship.

Even when imposing their own private censorship, the Platforms face a coordination problem. A Platform will sometimes be aiming merely to sanitize its own site by removing opinion it considers distasteful; but it still needs to limit the risk of losing users who seek the suppressed opinion elsewhere. It therefore must coordinate with the other Platforms to make sure they suppress the same sort of opinion. The need for coordination is all the greater when a Platform aims to influence politics or opinion. For that purpose, it needs to ensure that what it suppresses will not appear on another dominant Platform—at least not one nearly as large and with substantially overlapping users. Otherwise, its censorship will not effectively shape the public mind.

Although the Platforms therefore often need to coordinate, they cannot do so by themselves without antitrust difficulties. The government solves this problem by offering them coordination—by supplying them with guidance as to what is worthy of suppression, thus allowing the Platforms to align their censorship. Indeed, the government increasingly provides the coordination on a massive scale by subsidizing and working with . . . private censorship and misinformation outfits.

All of this coordination, whether done directly or through cutouts, is a serious abridgement of the freedom of speech. . . .

Id, at 246-47.

4. The Anti-Evasion Principle

Although the word abridging bars evasions, there’s also a longstanding constitutional principle that government cannot use private parties to do its dirty work:

In Cummings v. Missouri—an unconstitutional conditions case—the Court declared that “what cannot be done directly cannot be done indirectly. The Constitution deals with substance, not shadows.” In Frost & Frost v. Railroad Commission—another unconstitutional conditions case—the Court declared: “It is inconceivable that guaranties embedded in the Constitution of the United States may thus be manipulated out of existence.” Later, in Norwood v. Harrison—an equal protection case concerning state aid to private schools—the Court added that it is “axiomatic” that the government “may not induce, encourage or promote private persons to accomplish what it is constitutionally forbidden to accomplish.” Perhaps the most notable expression of such ideas came from Chief Justice John Marshall in Wayman v. Southard: “It is a general rule that what cannot be done directly from defect of power cannot be done indirectly.” Together, these admonitions reveal a general principle against evasion, including the evasion accomplished through private parties.

Courting Censorship 247.

Tellingly, the courts have interpreted rights in ways that cut off such evasions. As Eugene Volokh observes, the Fourth and Fifth Amendments are often understood to limit government officers from using consenting private parties to do what the officers cannot. Even the First Amendment is interpreted to block evasions—as evident from cases on unconstitutional conditions.

It’s therefore puzzling to read arguments that the government can work through corporate cutouts to suppress the speech of individuals. Even when it’s done entirely cooperatively, with full consent from the platforms, it looks like the sort of evasion that doctrine forbids.

5. The First Amendment’s Words Help Distinguish Unlawful Censorship from Lawful Government Communication

One might protest that if the abridging standard reaches even consensual arrangements, it doesn’t leave room for government to talk to newspapers about stories that pose a threat to national security. But that assumes that every time the government reduces speech, it is making a law or policy reducing the freedom of speech. The First Amendment expressly avoids that mistake.

It may seem improbable that an eighteenth-century text can offer moderate and helpful guidance for twenty-first century problems. The First Amendment’s phrasing, however, is valuable.

First, the First Amendment bars government from abridging the freedom of speech, not just abridging speech. So, it cannot be assumed that every government action reducing speech violates the First Amendment, and this leaves room for government to persuade newspapers or Platforms to drop some speech.

Second, the First Amendment applies to law and, by extension, policy. It thereby permits government to engage in persuasion diminishing speech as long as it does not rise to a law or policy.

These are just summaries of complex points; you can read further details in  Courting Censorship at 259-63. Here, the general point should be enough: The First Amendments words, unlike the coercion standard, offer some valuable first steps for sorting out the difference between unconstitutional censorship and lawful communication.

What does all of this mean in practice? A one-off unthreatening conversation with an editor (for example, about national security) is not barred by the First Amendment, but a policy of any sort to suppress lawful speech—whether because it is false or offensive—is forbidden. The government cannot make itself the arbiter of truth or offensiveness, and it cannot adopt a policy to suppress what is lawful.

The bar against a policy reducing any lawful speech is especially clear from the First Amendment’s pre-administrative assumption that only Congress would be making laws restricting speech. From that perspective, speech had the protection of the representative process; it could be assaulted only if legislators from across the country publicly passed a law suppressing speech. The First Amendment, in short, assumed that the executive could not adopt any policy against lawful speech.

Nowadays, of course, administrative and sub-administrative mechanisms enable the executive to adopt policies against speech. But that is all the more reason to carry forward the First Amendment’s assumption that only unlawful speech would be at risk from the executive. It therefore would be entirely appropriate, even necessary, for the injunction in Murthy v. Missouri to bar the executive from making or carrying out any policy reducing lawful speech.

As things stand, the coercion standard fails to draw a plausible line. It seems to permit all that is not coercive, and because this seems so lax as to invite censorship, the coercion standard ultimately requires judges to backtrack from coercion. Being unable to draw an accurate or even clear distinction, the coercion standard only confuses judges. They need a better guide for distinguishing between the government’s unlawful censorship and its lawful persuasion. So, even just as a practical matter, the First Amendment’s text deserves the Justices’ attention.

6. Private Consent Cannot Wash Away a Legal Limit

Rights are limitations on power, not tradable commodities. This is especially clear in the First Amendment because of its phrasing: “Congress shall make no law . . .” Being a law empowering and limiting government—a law made by the people—the Constitution cannot be escaped with any amount of lesser consent, whether from individuals, companies, or states.

Prof. Somin urges that the freedom of speech is unabridged as long as an individual consents—and thus as long as the government does not use coercion. He thereby treats the freedom as a matter of personal discretion, which an individual can relinquish as he pleases. In the founding era, however, the freedom of speech was understood (at least in theory) as a natural right—meaning not an individual’s natural physical freedom or discretion, but his non-injurious freedom of speech. See Philip Hamburger, Natural Rights, Natural Law, and American Constitutions 908-09. The point is not to insist on a belief in natural rights, but merely to observe that the freedom of speech was understood as a sphere of freedom or discretion that was the same for all individuals and their associations. By guaranteeing it against government abridgment, the First Amendment barred government from reducing that sphere of that freedom. And by imposing this barrier as a legal limit on government, the amendment prevented it from being adjusted by any consent less than that of the people in the amendment process.

At the same time, consent can have a role within the First Amendment. Most constitutional rights let government do some things with consent that it otherwise could not and yet also limit what consent can accomplish. The government, for example, can search your house with your consent because that is a reasonable search, not because your consent can give the government a power that the Fourth Amendment denies to it.

Bringing the question back to the freedom of speech, the First Amendment leaves room for government to act with consent, but only as long as the consensual restrictions do not abridge the freedom of speech. When does that happen? The word abridging is very informative:

When a law directly constrains speech, it can be difficult to sort out whether it violates the First Amendment, for this amendment does not specify the difference between a law that abridges the freedom of speech and one that does not. But when a condition restricts speech, the inquiry can be easier, for if the condition confines speech more severely than the government could do directly, then the condition is abridging the freedom of speech.

Philip Hamburger, Purchasing Submission 169 (Harvard 2021). The government cannot use consent to impose restrictions it could not impose directly because that would abridge the freedom of speech.

This argument obviously relies on direct abridgments as a baseline for determining what is an abridgment done by consent. This baseline makes sense because direct regulation is the archetypical mode of violating rights, and the First Amendment comes close to expressly recognizing this when it declares: “Congress shall make no law . . .”

One way or another, what is unconstitutional when done directly establishes an apt baseline for measuring when consensual limits abridge the freedom of speech. It follows the text, it offers clarity, and it helps to avoid the evasion that occurs when the government turns to consensual arrangements to do what would be unconstitutional if done directly.

(Incidentally, a brief note on vocabulary—in particular, the word persuasion. It surely is a mistake to contrast unconstitutional abridging, let alone unconstitutional coercion, with constitutional persuasion. I must admit that I’ve been guilty of casually using the word persuasion in this manner. But that usage misleadingly suggests that consensual restrictions on speech are generally constitutional. In fact, they are not constitutional if they go beyond what could be directly imposed and thus abridge the freedom of speech.)

The full implications of this analysis are complex and are surveyed in Purchasing Submission. Suffice it to say that the role of consent within the freedom of speech is important but limited. When a consensual restriction on speech goes beyond what would be constitutional if done directly, it abridges the freedom of speech.

7. The Freedom of Speech Is a Public Good

The inability of consent to cure speech restrictions beyond what could have been imposed directly is confirmed by the public interest in speech. The First Amendment protects speech not merely because speech is valuable to the speaker, but more broadly because it has value for the whole of our society, if not the world.

Just one individual’s speech can reshape our understanding of government (James Harrington) or even of the universe (Galileo). Today, individual scientists can illuminate our health and the dangers to it from government policy (Drs. Jayanta Bhattacharya and Martin Kulldorff). All such individuals have been censored at high cost not just for them, but for the rest of us, who benefit from learning the truth, even if officialdom condemns it as errorheresy, or misinformation.

We learn from the truth, even from untruths, and most commonly from the partial truths that constitute most speech—as John Stuart Mill explained in his On Liberty. So, all censorship harms the public, not just the censored individual.

This matters for consent. Even if a censored individual consents to having his speech suppressed, that does not cure the harm, because the damage from censorship is to all of us, not merely the censored individual. As argued by Prof. Daniel Farber, free speech, with its informational benefits, is not just a private right but also a public good. This point is developed more broadly by my colleague Prof. Thomas Merrill.

It has been seen that under the First Amendment, if a restriction cannot be imposed directly, such a restriction should not be considered justified by consent. And this makes sense when one considers the public interest in each individual’s freedom of speech. The First Amendment is a limit on government power for the benefit of all of us, and the government therefore shouldn’t be able to whittle it away by making “a separate peace with those who can be induced to consent. Private deals allow the government to buy off political opposition.” Hamburger, Purchasing Submission 105.

8. No Consent from the Speakers

Prof. Somin applies his ideas about coercion and consent to justify the government’s suppression of speech on the social media platforms. This seems odd because the government pressures and cooperates with the platforms to suppress the speech of third parties, who aren’t consulted by the government, and who don’t consent. One might have thought that even under his coercion theory, Prof. Somin would view this nonconsensual censorship as unconstitutional.

But no. He seems to think that when government gets the plaforms’ consent, it has avoided coercion and so has acted constitutionally.

In the archetypical instance of “jawboning,” a government official calls up a newspaper editor to ask him to delay or stifle a news story that would harm the national interest. In other words, government seeks the consent of the speaker.

Yet that is precisely what does not happen in the current censorship regime. Instead, government officials or their private cutouts urge the platforms to suppress the speech posted by members of the public. In some instances, the platforms genuinely consent; in other instances, they relent under pressure. But even when the platforms consent without pressure, the censored individuals have not consented—indeed, they often are not even informed that they are being censored.

You might protest, as does Prof. Somin, that when the government persuades Reason not to let him post on this website, the government does not thereby violate his First Amendment rights. In his view, “that’s no different” from when the government persuades the platforms to suppress Dr. Jayanta Bhattacharya’s posts. Really, no different? Prof. Somin publishes here as a member of the blog, who has been personally invited to publish under its masthead. Whereas newspapers publish their choice of submitted editorials, blogs publish anything a member of the blog posts, but either way, nothing gets published or posted except what has been chosen by the newspaper or blog through its selection of an editorial or blogger. Thus, to the extent an editorial or post appears in the newspaper or blog, it becomes the speech of the newspaper or blog.

In contrast, any individual can post on the platforms, at least until his posts are removed, and that is the individuals’ speech, not the platforms’ speech. Not even the platforms claim otherwise. (Rather than claim speech rights in what their users post, the platforms merely assert that they have editorial discretion—a speech right in expressively discriminating against their users.) So when the government consensually gets a platform to remove posts, it is suppressing the speech of third parties without their consent.

You might respond that users agree to the platforms’ terms of service. But the government has often pushed the platforms to suppress speech that did not violate those terms. In any case, the private consent to the terms of service cannot give government a power that the First Amendment denies to it.

The lack of consent from the individuals who speak on the platforms should be a sobering problem for Prof. Somin’s theory. His underlying principles of coercion and consent mean that the individual speakers—for example, the distinguished scientists Drs. Jayanta Bhattacharya, Martin Kulldorff, and Aaron Kheriaty—should have been consulted by the government to get their consent to their suppression. But they were not asked for their consent, and they would never have given it.

9. Consent Does Not Show an Absence of Coercion

A further difficulty is that Prof. Somin’s coercion-consent dichotomy presents false alternatives. He takes his dichotomy so far as to question the “significant encouragement” test on the ground that “[i]f the private party gave the governmental actor such control voluntarily, that may be bad media ethics, but it is not a violation of freedom of speech.”

Although Prof. Somin speaks of voluntary consent, he might have acknowledged that, in familiar legal doctrine, a wide range of pressures and influences less than coercion can vitiate consent.

Put another way, consent is the just the first stage of analysis in contract law. A party can consent, and yet a court can find that the consent was obviated by other considerations. To be sure, in the nineteenth century, the doctrine on duress spoke of a gun to the head and other compulsion. But that was a long time ago. Since then:

[T]he law has become profoundly sensitive to the complex pressures—including economic and informational imbalances—that can deprive apparently consensual arrangement[s] of their voluntary character.

Although contract law historically paid little attention to such imbalances, notions of economic duress and informational disparities have gained much credence, even if not consistently, in the past half century.  Similarly, medical treatment is not considered truly voluntary unless doctors disclose enough information to their patients for them to exercise informed consent. Researchers are required to secure informed consent from their human subjects, and payments to impecunious volunteers, even payments as low as $20, prompt soul searching as to whether the resulting economic pressure renders their participation nonconsensual. Most recently, pressures less than coercion have come to the fore in sexual relations. In the context of employment or education, a range of economic and peer pressures for sex or for conformity to sexual stereotypes can be legally significant, and a wide variety of words can create a hostile work environment.

Admittedly, some of these developments can go too far. . . . But the underlying insight—that economic and personal pressures can compromise consent—is undeniable.

Hamburger, Purchasing Submission 193. In short, it would be very odd for an analysis of the freedom of speech “to ignore the past century of expanding legal sensitivity to the range of pressures less than coercion. When the law recognizes such threats to consent across the legal landscape, it does not make sense to pretend that such questions do not arise” under the First Amendment. Id.

10. Secrecy and Scale

It’s also important to recognize the secrecy and scale of the current censorship regime. Murthy v. Missouri is not about an individual officer calling up a newspaper editor to chat about one of the paper’s stories. It is a massive covert operation against the American people.

This was confirmed just a day after the argument in Murthy v. Missouri in further revelations from Elon Musk:

This is coming from multiple parts of the government. From the State Department, the FBI, Homeland Security, from really many, many parts of the government. It wasn’t just one arm of the government.

He explained that:

[T]here’s a little-known agency in the state department called the Global Engagement Center, which most people have never heard of, but they might have been the single worst offender because they demanded the suspension of over 250,000 accounts which I think all Twitter largely complied with. . . .

The suspension demands were so broad that they accidentally demanded a suspension of a journalist on CNN and an elected Canadian politician. It was just an incredibly broad swath.

As for the secrecy, “‘There was this FBI portal that auto-deleted all communications after two weeks, so we actually don’t know what was said there.’”

It is therefore utterly disproportionate, even unreal, to compare the current censorship to an old-style phone conversation between an official and a newspaper editor. As I explain in Courting Censorship at 259-63, that sort of conversation does not necessarily violate the First Amendment. In contrast, a policy to use the platforms to suppress points of views critical of government policy across vast numbers of posts—indeed, without the consent of the speakers—is grossly unconstitutional, regardless of whether the government coerced or cooperated with the platforms.

11. The Last Straw

A coercion standard would be the last straw for the freedom of speech. It already is at much greater risk than in the past because of the rise of administrative and especially sub-administrative power (the sort of power that operates below the level of notice-and-comment rules and ALJs). With such power, government can engage in wholesale suppression:

Administrative and especially sub-administrative power have facilitated the substitution of wholesale suppression for the old retail suppression. Traditionally, the federal government could not actually suppress speech, but could only punish the speaker. And it had to charge him with seditious libel or some other offense and prove to a judge and jury that his words violated the law. This was retail adjudication, and it was essential for the protection of speech and other rights.

Nowadays, however, . . . government can press for administrative licensing of speech or, worse, can use informal mechanisms such as sub-administrative threats, raised eyebrows, and offers of coordination to get the dominant Platforms to suppress their users’ speech. Although the government doesn’t have to prove anything against anyone, it can make vast amounts of speech just disappear. In such ways, retail punishment of individuals through seditious libel prosecutions has been abandoned for wholesale suppression of opinion. . . .

In shifting from retail prosecutions to wholesale suppression, the government places the onus of going to court on the censored. Whereas the censored once merely had to defend themselves when prosecuted, they now need to go to court to stop the censorship. . . .

[T]he government’s administrative and sub-administrative mechanisms flip around the burdens of proof and persuasion. Whereas government once had to prove the guilt of each of us before we could be punished for our speech, now each of us has to prove that the government unconstitutionally censored us and that the court should grant a remedy—just in order to make our words visible.

Courting Censorship 218-19.This wholesale suppression and its shift in burdens is bad enough, but there is more.

The freedom of speech is becoming nearly a right without a remedy. Qualified immunity makes it difficult to get damages for past censorship. The limits on injunctions mean it isn’t easy to get a remedy against the breadth of future censorship. And, of course, neither damages nor injunctions are prompt and effective against secret censorship—as evident from the fact that it took half a decade to get the first injunction against the current suppression.

On top of all of this, if coercion becomes the standard for speech violations, there would be no remedy at all against the host of censorship mechanisms that are not overtly coercive. One would have thought that Supreme Court doctrine should stop censorship in its tracks. Instead, as detailed in Courting Censorship, doctrine is becoming almost an instruction manual for how to get away with it.


The coercion-consent measure of free speech is utterly mistaken. It is wrong about coercion, it is wrong about consent, and it practically invites government censorship. So, if the Supreme Court takes such an approach in Murthy v. Missouri, the case will stand out as one of the most abysmal First Amendment decisions in the nation’s history.

In Murthy, two states and five individual plaintiffs urged the Supreme Court to reject the coercion standard and recognize that the First Amendment “capaciously protects the freedom of speech from any ‘abridging’ (i.e., diminishing) of that freedom.” Respondents’ Brief 48. The Court should heed this argument and fall back upon the First Amendment. If necessary, it should invite further briefing and argument on this point. Neither the Court nor the country can afford any mistake about it.

The Court should also invite briefing and argument from the censored individuals, Drs. Jayanta Bhattacharya, Martin Kulldorff, and Aaron Kheriaty, and Jill Hines and Jim Hoft. Remarkably, in both Murthy v. Missouri and NetChoice v. Paxton—the two cases that may determine the fate of free speech in America—the censored individuals had no opportunity to argue for themselves. None. The individuals who have been suppressed, who have the most intense First Amendment interest, should have a voice at the Court.

Originally Published in Reason