For years now, scholars have expressed alarm at the tendency of government officials to use informal means, rather than democratically enacted laws, to pressure the social media companies to take down what they consider to be harmful or offensive speech. The term commonly used to refer to this kind of informal (but often quite effective) practice of government speech suppression is “jawboning.” While by no means unique to the digital public sphere, jawboning has come to be a particularly common tactic of government regulation of the social media platforms, in part because the government has few other means of regulating what the social media companies do. Scholars have worried, for good reason, that the practice of jawboning allows government officials to evade the stringent constraints on their power to regulate speech imposed by the First Amendment. But relatively little attention has been paid to the constitutional question of whether, or rather when, government jawboning itself violates the First Amendment.
Two recent events have pushed this question to the front page. The first of these events was President Trump’s decision in early July to file class-action lawsuits against Facebook, Twitter and YouTube. These lawsuits accuse the companies of violating Trump and the other class members’ First Amendment rights when they took down, deprioritized, or shadow banned the plaintiffs’ speech. Although the lawsuits have attracted much derision from legal scholars for getting the “First Amendment exactly wrong” by failing to recognize that it applies only to government actors, not private corporations, the core argument the Trump complaints make is not that Facebook, Twitter and YouTube are generally bound by the First Amendment but, rather, that the companies “censored” Trump and other class members’ speech because of what the complaints describe as the “overt coercion” of Democratic members of Congress. In other words, the Trump lawsuits make a First Amendment jawboning argument and one that clearly identifies Democratic members of Congress as the agents who were ultimately responsible for the violation of the plaintiffs’ First Amendment rights (even though, as is sometimes the case, the complaints name no government officials as defendants).
The second jawboning-related event was the July 15 release by the U.S. surgeon general, Vivek Murthy, of a health advisory warning of the perils to the national public health of social media-disseminated misinformation related to the coronavirus. The advisory and accompanying press conference, in which White House Press Secretary Jen Psaki announced that 12 people were producing 65 percent of the anti-vaccine misinformation on social media platforms and called on Facebook, in particular, to take “faster action against harmful posts,” generated a vigorous debate about whether the White House’s actions violated the First Amendment rights of Facebook and its users.
Both of these events raise, in somewhat different contexts (the first a court case, the second a public relations kerfuffle), the question of where the line falls between permissible government pressure and impermissible government coercion when it comes to efforts to get social media platforms to suppress harmful speech. Whatever answer courts give to this important question will shape the relationship between government officials and the social media companies that play such a powerful role in contemporary public life. It will help determine, in other words, the balance of power between the old-fashioned governors of the brick-and-mortar public sphere and the “new governors” of the digital public sphere.
It is not an easy question to answer, however, either doctrinally or normatively. This is because the relevant Supreme Court cases are quite inconsistent in the rules they apply to determine when government efforts to pressure private speech intermediaries into suppressing speech violate the First Amendment. This inconsistency can be blamed in part on the fact that there are good reasons to both support and oppose a broad First Amendment rule against jawboning. (Beware those who say the First Amendment issues are easy here.) In this post, I sketch out the doctrinal landscape as it currently stands before exploring the difficult normative questions that jawboning cases force us to grapple with.
The Messy Constitutional Law of Jawboning
The rules that courts apply in First Amendment jawboning cases vary considerably from jurisdiction to jurisdiction, and from case to case. This is largely a consequence of the fact that the Supreme Court’s two most important cases regarding government coercion of private parties employ markedly different approaches to the question of when government efforts to encourage or pressure private parties into doing, or not doing, something implicate the First Amendment.
The first of these two cases is Bantam Books v. Sullivan (1963). In that case, the court held that a government agency, the Rhode Island Commission to Encourage Morality in Youth, violated the First Amendment when it sent letters to local booksellers that informed them that they were selling book and magazines that the commission had identified as objectionable reading material for youth, thanked the booksellers for their “anticipated cooperation,” and reminded them of the commission’s duty to recommend obscenity prosecutions to the state attorney general. Although the commission lacked any power to sanction the booksellers itself, and acknowledged at trial that at least “several” of the items it identified in the letters were constitutionally protected speech that could not have been successfully prosecuted by the attorney general even if the commission had recommended that he do so, the court held that the letters formed a scheme of “informal censorship” because the commission’s intent was “clearly to intimidate” the booksellers into removing the identified books from their shelves. The court concluded, in other words, that government efforts to intimidate private intermediaries into suppressing other private persons’ speech by threatening them with bad consequences if they did not comply violated the First Amendment—at least when the threats have the intended result—even if the threats the government makes are implicit, attenuated and ultimately empty.
Bantam Books remains good law today. But a decade later, a significantly reconstituted court handed down another decision that suggested a much more restrictive test of unconstitutional government coercion. In Blum v. Yaretsky (1973), a class action of Medicaid recipients, all of whom received state-subsidized care at private nursing homes, claimed that the nursing homes violated the residents’ constitutional due process rights when they discharged or transferred them to other nursing homes that provided less intensive medical care without giving them notice or a hearing. The plaintiffs argued that the nursing homes’ actions counted as state action that violated their due process rights because the homes were motivated to act by federal regulations that, in an attempt to save taxpayer dollars, strongly encouraged nursing homes to transfer or discharge Medicaid patients whenever medically possible.
Justice William Brennan (author of the Bantam Books opinion) wrote a dissent in which he argued that the plaintiffs had demonstrated that the government was ultimately responsible for the transfer and discharge decisions. But a majority of the court disagreed. Because the nursing homes were not specifically ordered by the government to transfer or discharge the patients, the court concluded that it did not matter that the nursing homes did what the government intended them to. “Our precedents indicate,” Justice William Rehnquist wrote for the majority, “that a State normally can be held responsible for a private decision only when it 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.” What this meant, the opinion made clear, was that government coercion that led private actors to deprive others of constitutionally protected rights did not count as state action, or implicate the Fourteenth (or First) Amendments, so long as the private actors retained the ability to decide for themselves what to do in individual cases.
Blum and Bantam Books obviously articulate very different approaches to the problem of distinguishing unconstitutional government coercion from entirely permissible government pressure or encouragement. And while it is surely true that a clever lawyer could find a way to reconcile them, the lower courts that have been asked to decide First Amendment jawboning cases in the years since these two cases were handed down have shown little interest in doing so. Almost without exception, they have instead treated either Blum or Bantam Books as the controlling precedent and ignored whatever complications the other case might add to the analysis. (Indeed, courts that cite to Blum tend to entirely ignore the fact that Bantam Books exists, and vice versa.) The result is a deeply inconsistent body of precedent.
Those decisions that treat Blum as controlling have construed the First Amendment to prohibit only extremely coercive government jawboning. In one case, for example, a district court (in an opinion affirmed by the U.S. Court of Appeals for the D.C. Circuit) held that government threats to cut off funding to National Public Radio (NPR) if it did not cancel the broadcast of an interview with journalist, political activist, and death-row inmate Mumia Abu-Jamal did not transform NPR’s subsequent decision to cancel that broadcast into state action because they did not leave NPR with “no choice but to act in a way compelled by the [government].”
In another case, the U.S. Court of Appeals for the Third Circuit rejected claims that a city violated a billboard owner’s First Amendment rights when it “politely but firmly suggested” that the bank that owned the plot of land on which the billboard stood “remove the unsightly billboards” and noted the possibility that, if the bank did not prove “interested in being responsive to this community[,]” the city might be forced to resort to costlier “legal procedures.” Because the city’s letters to the bank contained no “enforceable threats,” and the only sanctions the city could bring to bear against the bank were “civil or administrative proceedings under a zoning ordinance not yet drafted[,]” the Third Circuit held that the decision to terminate the billboard owner’s lease was private action beyond the scope of the First Amendment and that this was true even though the bank admitted that it removed the billboard in order to maintain its good relations with the city.
Courts applying the Bantam Books standard, in contrast, have interpreted the First Amendment to prohibit government pressure tactics that seek to intimidate rather than persuade private actors into suppressing objectionable speech, even when those tactics are not so strong as to leave their target with essentially no choice but to comply. In Okwedy v. Molinari, for example, the U.S. Court of Appeals for the Second Circuit held, in an en banc opinion, that a reasonable jury could conclude that the decision by a private company to take down several anti-gay billboards after the Staten Island borough president wrote to the company to express his view that the messages were “unnecessarily confrontational and offensive” and “convey[ed] an atmosphere of intolerance which [was] not welcome in [the] Borough” violated the First Amendment rights of the group that erected the billboards. A reasonable jury could conclude as much, the Second Circuit concluded, even though the borough president lacked all “direct regulatory or decisionmaking authority” over the billboard company, and even though the letter contained no explicit threats. This was because, the court explained, the jury could conclude that the letter the borough president sent the billboard company “contained an implicit threat of [economic] retaliation if [the company] failed to accede to [the borough president’s] requests” and that it represented an attempt to coerce the company into complying with the president’s demands, rather than an attempt to convince the company of the correctness of his view.
A few years later, in Backpage.com v. Dart, the U.S. Court of Appeals for the Seventh Circuit relied on both Okwedy and Bantam Books to conclude that letters that Cook County Sheriff Thomas Dart wrote to credit card companies, in which Dart requested that the companies “immediately cease and desist from allowing [their] credit cards to be used to place ads on websites like Backpage.com” and promised to continue to “harass” them about the matter, violated the First Amendment rights of the website and its users. The Seventh Circuit held as much even though the credit card companies obviously had a choice in whether to comply with the “request” Sheriff Dart made in his letters, and even though the company spokesmen flatly denied that the companies acted in response to government pressure. and even though it was extremely unlikely that the sheriff would have been able to prosecute the credit card companies for any crime. Like the Second Circuit, the Seventh Circuit made clear in the opinion that, in its view, what mattered to the First Amendment analysis was not whether the targets of the government’s pressure campaign faced imminent harm if they did not comply but whether the government’s actions represented an attempt to coerce, rather than to persuade, the private parties into complying.
As these case descriptions suggest, what conclusion courts reach about the merits of First Amendment jawboning arguments will depend, to a significant extent, on which line of precedents they rely on. Take the Trump case for example.
If the Florida district court in which Trump filed his lawsuits applies the rule from Blum, it is almost impossible to imagine the jawboning argument the complaints make surviving a motion to dismiss. Although it is true that a number of Democratic senators called on the social media companies throughout 2020 and 2021 to either temporarily or permanently ban Trump’s tweets, posts, and videos, and to ban the speech of other habitually misleading or incendiary users, Facebook, Twitter and YouTube retained considerable freedom to choose to listen to those appeals or to ignore them. This was made evident by the fact that, for months and months, all three companies ignored the Democratic appeals—perhaps because they were pressured to do so by the jawboning campaign that Republicans in Congress mounted at the same time, to pressure the social media companies into being more lenient when it came to conservative speech. Whatever the reason, the companies were clearly not deprived of all freedom of choice by the efforts of Democratic politicians to push the companies to more aggressively take down misleading or violence-inciting speech. Nor of course did members of Congress issue the kinds of “enforceable threats” that the Third Circuit required, before it was willing to treat government requests to private parties as unconstitutional coercion. After all, what enforceable threats could individual members of Congress issue? Thus, under Blum, Trump’s lawsuit is a loser.
But matters might be different if the district court treats Bantam Books as the governing precedent. When then-senator and presidential candidate Kamala Harris tweeted to Twitter CEO Jack Dorsey that it was “[t]ime to do something about” an incendiary Trump tweet, or when Sen. Richard Blumenthal threatened to “break up [the] tech giants because they’ve misused their bigness and power” by allowing “the President [to] use [the] microphone [of social media] to spread vicious falsehoods and ... overturn the will of the voters,” were they attempting to persuade or intimidate the platforms into removing Trump’s speech? Surely a little bit of both.
Of course, to succeed on his jawboning claim, Trump would presumably have to convince a jury that Facebook, Twitter and YouTube censored his speech because of governmental pressure, not for some other, independent reason. This he is unlikely to be able to do. All the evidence suggests that the platforms chose to deplatform Trump because the reputational and internal costs of not doing so became too difficult to bear in the wake of the Jan. 6 invasion of the Capitol. Nevertheless, the jawboning claim is a lot stronger if Bantam Books applies.
Bantam Books also strengthens the claim that the Biden administration violated the First Amendment when it pressured Facebook to suspend the accounts of 12 people the administration believed to be responsible for the majority of coronavirus-related misinformation circulating on the platform. Nothing the administration said or did when it released the health advisory qualifies as unconstitutional coercion under the Blum standard. But it is much less clear that this is true if Bantam Books, Okwedy and Dart set out the governing law. The language the press secretary used was not as legalistic or threatening as the language Dart used in his letters to the credit card companies, but it was not so different from the language the Staten Island borough president used in his letter to the billboard company in Okwedy. Like the borough president, Psaki’s remarks at the press conference identified specific speech the administration wanted removed from the platforms; like the borough president’s letter, Psaki’s remarks made clear that the administration believed the decision to keep that speech in public view was unacceptable. (Indeed, when asked at a press conference the day after the release of the health advisory whether Facebook’s announcement that it has removed 18 million pieces of coronavirus-related misinformation from the platform was sufficient to satisfy the administration, Psaki replied unequivocally that Facebook had not done enough because the accounts of the 12 frequent misinformers had not been suspended.) It is quite possible to read into these comments—and in President Biden’s assertion the following day that Facebook was “killing people”—an implied threat of economic retaliation if the social media companies did not respond more cooperatively to what Psaki described as the administration’s “asks.”
Of course, it is possible to distinguish what the Biden administration said and did from what the Rhode Island commission did in Bantam Books and the borough president did in Okwedy and Sheriff Dart did in Dart. The fact that Psaki’s remarks—and Biden’s, for that matter—were made to a public audience, rather than in a private letter, might suggest that the administration was attempting to pressure Facebook to act by mobilizing public sentiment against it, rather than by implicit threats. Facebook’s size and power also might make it less susceptible to pressure than the booksellers in Bantam Books were, or Backpage.com was, although it is not clear from the cases that this matters. Furthermore, at her press conference, Psaki was careful to note, as neither the Rhode Island commission nor Sheriff Dart nor the Staten Island borough president did, that Facebook was a “private sector company” that was ultimately entitled to make its own decisions about what steps it should take to combat misinformation.
These facts might, and perhaps should, push the administration’s actions back across the line from unconstitutional coercion to perfectly constitutional government speech. But whether or not they do is hardly clear-cut. This uncertainty is partly the result of Bantam Books and its progeny drawing an inevitably fact-intensive distinction between government persuasion and government coercion. But it is also partly the result of the courts failing to agree on what the test of government coercion in First Amendment cases should be. This failure to settle the basic terms of the constitutional analysis reflects the difficult normative questions swirling beneath all of this black letter law.
The Difficult Normative Questions Raised by First Amendment Jawboning Claims
One reason courts might be unable to agree on whether to apply a broad or a narrow test of unconstitutionality in jawboning cases might be that there are good reasons to favor both an expansive and a narrow First Amendment rule. That this is so has to do with the age-old problem of the speech intermediary.
Private speech intermediaries—radio stations, television networks, newspapers and, of course, the social media platforms—play an enormously important role in the day-to-day operation of the public sphere in the United States. But as scholars have recognized, these intermediaries are often highly susceptible to political pressure to suppress speech that powerful groups find objectionable, since the intermediaries do not internalize the full costs of speech suppression. For intermediaries like booksellers, cable networks, and social media platforms, profit lies in getting customers to consume as much of their product as possible, not in showcasing the widest variety of speech. Hence, if an easy way to stay in business is to exclude the most objectionable content, they may be quite happy to do so.
This is likely to be as true—if not more true—of the social media platforms as it is of any other for-profit speech intermediary. So much speech flows through Facebook, Twitter and YouTube on a daily or hourly basis, what does it cost them to remove one tweet or post or video—or maybe thousands of tweets or posts or videos—if a government regulator asks them pretty please? The answer may be, a lot, if the speaker is a powerful politician like Donald Trump. But you or me? Or someone with less social or political capital, like Mumia Abu-Jamal? The cost to the platform in those cases might be very small indeed. Meanwhile, the financial benefits it might receive from cooperating with government requests might be significant, or the platform might hope them to be.
This explains why the social media companies often operate, as Daphne Keller puts it, in “anticipatory obedience” of the dictates of government regulators. And it provides good reason to disfavor a narrow test of unconstitutional government coercion, like the one the D.C. Circuit approved in the Abu-Jamal case. Given the incentives of private speech intermediaries, a test that treats as unconstitutional only government pressure campaigns that leave the private speech intermediaries with no choice but to comply is likely to make it quite easy for government officials to do what the First Amendment is supposed to prevent them from doing: namely, use their economic and political power to limit the diversity of public debate, by “requesting” that private intermediaries suppress speech on their behalf.
On the other hand, the sheer power that private speech intermediaries wield in the digital public sphere makes it important that government actors have some ability to advise and, yes, pressure private companies into adopting content moderation rules that promote the public good rather than just the companies’ economic self-interest. The facts at issue in the Trump complaint make this clear. Surely, during the run-up to the 2020 election and its contentious aftermath, and at a time when the nation was facing a deadly and growing pandemic, it was permissible, even desirable, for members of Congress to interrogate the officers of the large social media companies about the choices they made when it came to misinformation, political advertising, incitement, health information, and the like, and to criticize those choices when they were bad ones. If the result of their actions was to reconsider their willingness to tolerate false or misleading political speech, or health misinformation, or harmful content in general—well, isn’t that good, rather than bad?
We might worry therefore that a broad definition of unconstitutional government jawboning—something like the definition applied in cases like Okwedy and Dart—might make it too difficult for government officials to play the important supervisory role we might want them to play in the digital public sphere. Of course, the distinction that both decisions draw between government actions that seek to compel and government actions that seek to persuade is intended to leave room for government officials to advise, supervise and prod private speech intermediaries to act. But, as the above discussion of the Trump lawsuits shows, and the administration’s health advisory suggests, it is not always easy to tell when government speech is an attempt to convince, or an attempt to coerce. And that in turn raises the possibility that widespread adoption of the standard might chill valuable government speech, whereas the much more clear-cut Third Circuit standard might not.
This is not to say that courts should adopt the Third Circuit rule rather than the more expansive Second or Seventh Circuit approaches. The Third Circuit test fails, just like the D.C. Circuit test, to realistically take account of how the government wields its power. Intimidation tactics rarely require the government actor responsible for them to issue an “enforceable threat.” And yet the harm they pose to the diversity and vitality of public expression is no less real.
Of all the options on the table, the Bantam Books line of cases are, to my mind, the most likely to protect the free speech rights of both the social media platforms and their users against the informal use of government power, without entirely depriving officials of what the Seventh Circuit described as their “freedom of government speech.” But more could and probably should be done to flesh out the distinction that these cases draw between permissible and impermissible jawboning. Does it matter if the jawboning occurs in public and private? Does it matter if the government identifies specific instances of objectionable speech, or speaks in more general, policy-oriented terms? Does it matter if the government official is an expert? At present, it is difficult to find any answer to these questions in the cases. Yet answering them would provide a good deal more clarity than we presently possess about how the First Amendment shapes the interaction between the powerful individuals who exercise government power and the powerful corporations that control the digital public sphere.
Of course, any rules that the courts devise will help constrain government power only if there is a public record of the government’s actions, or if the private speech intermediary chooses to fight the action (as occurred, for example, in Dart). This is because neither the private intermediary nor the government officials will ordinarily have much motivation to acknowledge when jawboning occurs. People whose speech has been suppressed will therefore not know that they can challenge that suppression on constitutional grounds. Elucidating the rules that apply in jawboning cases thus can do only so much to prevent the private exercise of government power when it comes to online speech, absent much more robust transparency about the reasons why platforms take down or otherwise discriminate against individual speech acts or speakers. Nevertheless, something is better than the confused and confusing doctrinal landscape we have right now. Moreover, a clear rule against jawboning could help motivate, and perhaps justify, the imposition of transparency mandates as a kind of prophylaxis against otherwise hard-to-discern First Amendment violations.
This means that even if a clearer First Amendment rule against jawboning would do only so much to prevent the perhaps pervasive exercise of soft government power when it comes to the platforms, fleshing out the rule would nevertheless be well worth it. The good news, in that respect, is that courts are likely to have many more opportunities to do just that. Popular and political dissatisfaction with the current state of the digital public sphere makes it likely that efforts like those undertaken by the Biden administration over the past few weeks to pressure the platforms into suppressing harmful speech are unlikely to go away anytime soon. Rather than decrying these efforts as patently unconstitutional, or embracing them as an unproblematic exercise of the government’s supervisory power, they should be treated as an opportunity to think more deeply about what kinds of rules are necessary to protect free speech values in a public sphere in which both private entities and government officials possess considerable power to determine who can and cannot speak.