Introduction

We are awash in lies. Misinformation has always been with us, but the endemicity of social media and the depth of political polarization in the United States and elsewhere has enabled falsehoods to be amplified, monetized, microtargeted, and spread around the world at unprecedented speed and scale. The consequences for democracy, public health, and social harmony are emergent and grave.

Misinformation presents one of the most vexing challenges for content moderation on social media (and off it, such as on cable news) for myriad reasons. Because misinformation can be difficult or controversial to define, policing it risks chilling core political speech, and because misinformation often resists automated enforcement, it cannot be removed or countered nearly as quickly or as cleanly as it spreads. Government regulation of misinformation raises special concerns, both because public officials may be particularly susceptible to political bias and because even outright lies enjoy a measure of constitutional protection.

The orthodox U.S. constitutional response to harmful speech, including false speech, is counterspeech. As Justice Brandeis wrote in his canonical concurring opinion in Whitney v. California, “[i]f there be time to expose through discussion the falsehood and fallacies, to avert the evil by the processes of education, the remedy to be applied is more speech, not enforced silence.” The view that corrective speech is preferable to censorship resonates with Justice Holmes’ famous intimation that “the best test of truth is the power of the thought to get itself accepted in the competition of the market.” The digital age has revealed Holmes’ aphorism to be as absurd as ever, but he was prescient in saying “[t]hat at any rate is the theory of our Constitution.” So it has become.

And yet, there are reasons to think U.S. constitutional law might not leave as much room for counterspeech as the misinformation problem demands. Prior to 2011, Arizona had a public campaign financing law under which candidates for state office who opted into the system could receive additional state matching funds beyond the default amount in proportion to money spent above that amount by a privately financed opponent or the opponent’s independent supporters. The law was challenged successfully as a violation of the First Amendment. Rather than treat the additional spending as a form of government speech—and, in this case, counterspeech—that the government could direct as it pleased, the Supreme Court treated the government as impermissibly punishing the speech of the privately financed candidate.

Taken seriously, the Arizona case, Arizona Free Enterprise Fund’s Freedom Club PAC v. Bennett, seems to imply that the government may not deploy its resources to counter or dispel the speech of identified private speakers. The idea that the government must adopt a neutral posture toward competing ideas may reside somewhere within Justice Jackson’s famous identification of the notion that “no official, high or petty, can prescribe what shall be orthodox in politics, nationalism, religion, or matters of opinion” as the most fixed of stars in the American constitutional constellation. At the same time, the Supreme Court has said outright that “government speech is not restricted by the Free Speech Clause” of the First Amendment. In other words, one line of Supreme Court doctrine sees government counterspeech as impermissible propaganda or punishment of speakers, and another sees it as unconstrained government expression.

Resolving the apparent tension between these two doctrinal lines is important to assessing the range of possible responses to the misinformation crisis. Some of the most promising possibilities in addressing misinformation involve the use of public resources. For example, Ethan Zuckerman has proposed that the government levy taxes on digital advertising to support independent and public service digital media. Others have advocated for application of the now-defunct fairness doctrine to counter low-value, false, or one-sided speech on internet platforms. Does the First Amendment permit the government to use its own speech to say a private person’s speech is wrong, or to direct others to do the same? If so, subject to what conditions and constraints may it do so?

Any satisfactory answer to these questions requires an assessment of the constitutional status of government propaganda. Propaganda gets a bad rap, but governments engage in it pervasively, from routine press conferences to public school curricula to campaigns to promote policy positions or industry. It cannot be illegal per se for governments to push the political views of public officials, even if they spend resources to do so. At the same time, it is easy to imagine purely partisan or self-regarding instances of government speech—propaganda, that is—that would be so abusive as to exceed reasonable limits on state power.

As I elaborate below, distinguishing impermissible from permissible instances of government propaganda requires a functional analysis that relies on several factors: the purpose of the government speech; its degree of separation from the “private” speech of public officials; its transparency as to its source; the availability of alternative information channels; its electoral implications; and what I call the “generality” of the speech, or the degree to which it is directed at particular individuals or identity groups. Government speech does not need to be neutral—how could it be?—but it ideally should not be personal.

It would be fair to object that a “totality of the circumstances” inquiry of this sort is unmanageable by judges. That may be so, but there are nonetheless ways in which intentional thinking around these principles is useful. Judges might be able, for example, to use these factors as part of a kind of constitutional avoidance inquiry that helps shape existing First Amendment categories. And of course, judges need not be the audience. Building out principles of public morality that both constrain and motivate the behavior of government officials is important even in the absence of judicial review. Indeed, even when judicial review is appropriate, political development of those principles ordinarily should precede judicial involvement and shape the exercise of judicial power.

More broadly, tailoring the inquiry to questions of justiciability tends to reinforce the pernicious idea that the appropriateness of government speech turns on whether it infringes the rights of an objecting listener or competitor. As the Arizona public finance case demonstrates, we have come to understand free speech rights through the “pathological perspective” that Vince Blasi memorably identified, where the measure of a right is the dystopian world that would result from its wanton abuse by government officials. This is a mistake. Pathologies of governance can run in both directions. We should devote at least as much attention to the sorts of constitutional norms that are needed to make government effective as we do to the norms that are needed to avert tyranny.

I. The Misinformation Problem

Misinformation on social media is a serious problem. During the COVID-19 pandemic, unverified or false information about the virus’s origins, severity, treatments, mitigation measures, and vaccines routinely circulated and was amplified on Facebook, Twitter, and YouTube, among many other sites. This misinformation led directly to loss of life, likely on an enormous scale. Other forms of health-related misinformation, especially in relation to vaccination, remain prevalent and popular on social media sites. During both the 2016 and 2020 U.S. presidential election cycles, shares and retweets of fake news stories routinely received more engagement than true stories from reputable news outlets. Famously, in 2017, a conspiracy theory about Democrats running a child sex trafficking ring out of a Washington, D.C., pizzeria led a North Carolina man to break into the restaurant and fire a rifle inside. Research suggests that lies travel online substantially more quickly and reach more and a broader array of people than truth.

There are few effective public or private regulatory responses to the spread of misinformation. Misinformation can resist objective definition or may sit along a spectrum of reliability. Policing it requires judgment calls that we are rightly uncomfortable entrusting either to social media executives or to the underresourced employees of content moderation mills. Misinformation is often socially contextual as well, requiring careful study or localized knowledge to diagnose. This takes time and resources, the cost of which is measured not just on its own terms, but also in the opportunity it gives for fake news to go viral. A half-day’s delay is a lifetime for an online lie.

Apart from the logistics of regulating misinformation, the law limits the scope of permissible government intervention. The First Amendment protects people from being punished for lying, though the existing case law arguably applies intermediate rather than strict scrutiny to such laws. The government can lie with legal impunity as well, the (dubious) assumption being that it is likely to pay a political cost for doing so. Defamation law permits some liability for lies that do reputational damage, but defamation of public figures is constitutionally protected under First Amendment case law, with only narrow exceptions for intentional or reckless conduct. Beyond that, Section 230 of the Communications Decency Act shields social media platforms from liability for content they host but do not create. Suing content creators directly can be difficult, as they may be anonymous, hard to trace, or judgment-proof.

The misinformation problem is social and not just technological or legal. Even if the law permitted the government to regulate misinformation, it is not clear that we would really want it to do so. People tell lies or half-truths all the time. They lie about their appearance, their life experiences, their whereabouts, their drug use, their sexual partners, their beliefs, and their intentions. Studies have suggested that most people lie on a daily basis, and that lying to others comprises a substantial portion of our social interactions.

In politics, the spectrum of deception we have come to expect runs from puffery on the low end to sheer invention on the high, an ancient surmise. “No one has ever doubted that truth and politics are on rather bad terms with each other,” Hannah Arendt wrote in 1967. “Lies have always been regarded as necessary and justifiable tools not only of the politician’s or the demagogue’s but also of the stateman’s trade.” The trick, then, would be to regulate the lies we think should be regulated but to leave ourselves free to speak in the usual way, through varieties of truth, and to do so when we all disagree about what’s what. It’s a neat trick.

II. Counterspeech

The problem just described riffs on a familiar theme in the law of free expression. According to an influential view of the justification for protecting free expression, speech cannot be restricted solely out of a worry that people will believe the speech and act upon it. Curtailing speech in the name of a greater good is sometimes permissible on this view, but doing so because adults of sound mind will evaluate its truth and reach a different conclusion than the government’s undermines the idea that those adults are equal, autonomous citizens. It substitutes the government’s reasoning for our own.

Worries of this sort have motivated the common refrain within the U.S. free speech tradition that the best answer to speech is more speech. To Justice Brandeis, among the most celebrated champions of this view, “[t]hose who won our independence believed that the final end of the State was to make men free to develop their faculties.” Rational people given the liberty to develop their capacity for reason could be trusted to evaluate speech and counteract any injurious effects so long as the potential harm was not immediate. “To courageous, self-reliant men, with confidence in the power of free and fearless reasoning applied through the processes of popular government, no danger flowing from speech can be deemed clear and present unless the incidence of the evil apprehended is so imminent that it may befall before there is opportunity for full discussion,” Brandeis wrote in Whitney. “If there be time to expose through discussion the falsehood and fallacies, to avert the evil by the processes of education, the remedy to be applied is more speech, not enforced silence.”

Responding to false speech through counterspeech avoids many of the pitfalls of direct regulation. Concerns over how to identify misinformation or over how much of it we actually want to limit fade when the response is dialogic rather than punitive. The problem of delay remains—countering false speech online can certainly be time-consuming and resource-intensive—but it might be mitigated by the fact that counterspeech can also go viral, even if it typically does so less effectively than its target. With counterspeech, the broader worry that regulation of misinformation undermines citizen autonomy essentially disappears. Responding to misinformation by presenting contrary information relies on free, equal, autonomous, rational citizens to assess the truth on their own, just as Brandeis urged.

There are also no obvious legal barriers to government counterspeech, though there may be less obvious ones, as discussed below. The Supreme Court has stated in its “government speech” line of cases that “government statements (and government actions and programs that take the form of speech) do not normally trigger the First Amendment rules designed to protect the marketplace of ideas.” This freedom has allowed governments, for example, to screen out disfavored personalized license plate logos or designs or to host some monuments but not others in a park on the theory that the government retained ultimate control over its own expression. In those cases, the Court permitted the government to shield its decisions through the invocation of “government speech” even though it was in one instance engaged in overt viewpoint discrimination and in another placing a religiously themed monument on municipal property. The Court has gone so far as to state, unequivocally, that the free speech clause of the First Amendment “does not regulate government speech.”

In speaking in such broad terms, the Court has emphasized that government would be “radically transformed” if it were subject to the same free speech limits in speaking as it is in regulating the speech of private persons. This seems right, as far as it goes. When Justice Jackson memorably said that “no official … can prescribe what shall be orthodox in politics, nationalism, religion, or other matters of opinion,” he did not mean the government could not hold and express its own beliefs about such matters—the Court objected to the lack of an exemption from a compulsory flag salute, not to the salute itself. Governments run schools and universities, which involves crafting curricula, hiring faculty, and admitting students. Governments launch campaigns to improve public health or to support farmers or U.S. manufacturers or the government’s own services. Government subsidies support broadcasting and the arts. All of this is to say nothing of the fact that public officials relentlessly promote their own policy agenda even while on taxpayers’ time, and thus on their dime. It would be absurd to do any of those things while remaining content and viewpoint neutral. To speak while discriminating on the basis of content and viewpoint is simply what it means to speak.

If indeed the government is unrestricted by the freedom of speech clause in its ability to speak, it would seem to follow that it is unrestricted in its ability to counter speak. In the misinformation context, this might include a range of possible responses. Speaking purely hypothetically, imagine that sometime in the future, a novel airborne contagious virus causes a global pandemic that results in the deaths of hundreds of thousands of Americans per year, the shuttering of thousands of businesses, massive loss of employment, and physical, emotional, and financial stress and misery on an unimaginable scale. The government would like to encourage people to wear face masks indoors, which reliable scientific evidence suggests can protect people from being infected and further transmitting the virus. Some combination of grifters, partisan political opportunists, and genuine skeptics of the underlying scientific consensus are flooding social media with at best unverified and at worst invented claims and purported research suggesting that face masks increase rather than decrease one’s risk of infection and transmission. The First Amendment does not permit the government to censor these claims directly, nor does it permit the government to require private platforms to remove them. What other options may the government pursue?

  • At the most basic level, the government could attempt to release its own information asserting the safety and efficacy of face masks. It could post that information to social media and hope that it is widely disseminated.
  • It could act more proactively and engage in “strategic amplification,” such as by asking influencers with large social media followings to share and promote the government’s content.
  • It could formalize these arrangements by paying those influencers to share and promote the government’s content.
  • It could pitch a story to PBS or NPR, both of which receive government funding, in which the misinformation is identified, and the record corrected.
  • It could reply to, comment on, or “quote tweet” the posts of people sharing misinformation in which it states that the misinformation is false and offers a contrary narrative. It could do so respectfully or less respectfully.
  • It could contract with a private business to conduct a targeted operation that deploys bots to promote the government’s content, flood the threads of those who spread misinformation, insult them, and question their intellect, appearance, or loyalty. The company might also bait misinformation spreaders into violating platform community standards and then report them for doing so.
  • It could take any of these actions on a conspicuously partisan, racial, or religious basis.

Apart from these specific variations on counterspeech, other governmental projects could play a role in responding to misinformation. Zuckerman’s advocacy for digital public infrastructure might potentially include social media platforms or search engines “whose owners and managers took their civic roles more seriously.” Were the government to finance and stand up its own platform in order to avoid advertising-driven incentives for engagement—including engagement that sensationalizes, misleads, and misinforms—could it exclude or downrank content it deems to be harmful misinformation?

Some commentators have called for a reinvigorated fairness doctrine that would apply to online speech. Under this Federal Communications Commission (FCC) policy, some version of which was in place from 1949 to 1987, television and radio broadcasters were obligated to present both sides of controversial issues and to offer a right of reply to those who were criticized on the air. The FCC abolished the fairness doctrine in 1987, but it was upheld unanimously in the Supreme Court’s 1969 decision in Red Lion Broadcasting v. FCC. There is reason to doubt that the current Supreme Court would permit a fairness doctrine for the internet. Still, a state-imposed duty of fairness or right of reply applied to internet service providers or social media platforms is arguably consistent with current case law and would offer the government a privileged channel for responding to public health misinformation.

III. Limits on Counterspeech

Still, as some of the examples above imply, the Supreme Court’s apparent confidence that government speech—and hence government counterspeech—knows no First Amendment speech limits seems misplaced. Government speech can take many forms, ranging from the relatively benign and public spirited to much darker harassment, jawboning, or racism. Consider at least three possible constraints.

A. Compelled speech

As noted above, the Supreme Court has said government would be “radically transformed” if forced to conform its own speech to the limits of its regulation of private actors. The quoted language comes from a case in which University of Wisconsin students argued unsuccessfully that the university’s exaction of a mandatory student activity fee to fund student groups with which some students disagreed ideologically amounted to compelled speech in violation of the First Amendment. For students to prevail on such a claim would seem to threaten all of public finance, since the government typically funds its activities through tax revenues. To grant a veto to objecting taxpayers would be to abandon common governance.

And yet, the students had a point. In a line of cases commencing in 1978 with Abood v. Detroit Board of Education, the Supreme Court has found that public employees cannot be required to contribute agency shop fees to unions. In Abood, the Court said such fees had to be limited to nonpolitical activities such as collective bargaining and contract administration, and in Janus v. AFSCME, the Court extended this holding to all union activities, reasoning that a public sector union’s collective bargaining is inescapably political. The Janus Court did not distinguish the University of Wisconsin case, but the only apparent difference seems to be that, in the latter, the politically objectionable activity occurred at one remove from the payments—it was not the school’s own speech being funded but rather private speech funded on a viewpoint-neutral basis. That view would align with the Court’s statement in Johanns v. Livestock Marketing Association that “[c]itizens may challenge compelled support of private speech, but have no First Amendment right not to fund government speech.”

Where the government has singled out private speakers for funding in accordance with its preferences, it is difficult to understand what principle motivates this distinction. Universities, including public ones, regularly adopt political positions. They announce commitments to diversity in admissions and hiring and spend money in pursuit of those commitments. They divest from disfavored industries and invest in others for political reasons. They hire faculty whose views they find acceptable and refuse to hire faculty whose views they find abhorrent. Students are not constitutionally entitled to withdraw their tuition fees, or some pro rata share, in protest of these positions. Insofar as there is no requirement that public institutions themselves be viewpoint neutral, it is unclear why the Constitution would forbid them from using public money to support even private institutions that have a favored political agenda, or why objectors to such support are not simply in the position of taxpayers who disapprove of the government’s policies or contracting decisions. In fact, objectors would be in a better position than taxpayers, because they have a choice whether to associate with the institution.

I don’t mean to suggest that First Amendment compelled speech doctrine offers relief to ordinary objecting taxpayers. It doesn’t. But because the Court has not articulated the difference between this and Janus with sufficient clarity, it is difficult to tell what the road from tax objectors to Janus looks like, or how long it is. It appears that there are at least some contexts in which the government violates the First Amendment if it exacts at least some kinds of money from objectors in support of its favored political positions.

B. Speech as punishment

In a short series of cases involving campaign finance reforms, the Supreme Court has invalidated laws that raise contribution limits or increase public funding for political candidates when their opponents spend above a certain amount. First, in Davis v. FEC, the Court struck down the so-called Millionaire’s Amendment to the Bipartisan Campaign Reform Act. Under the law, candidates who faced self-financed opponents who spent above a threshold could benefit from higher individual contribution limits for their own campaigns. Later, in Arizona Free Enterprise Club’s Freedom Fund PAC v. Bennett, the Court held that Arizona’s public finance system could not increase the public matching funds a participating candidate received in proportion to spending by (or on behalf of) an opposing candidate who has opted out of the system.

The Court’s rationale appears to be that laws of this sort impermissibly burden the rights of candidates to “speak” through the money they spend on their political campaigns. What makes these putative burdens unusual, though, is that they are not prohibitions on speech but rather take the form of facilitating counterspeech. The notion that counterspeech counts as some sort of punishment for speaking, on the same order as direct regulation, appends a puzzling addendum to the American free speech tradition. I had not supposed that the First Amendment confers a right to speak without opposition—this would certainly come as news to people who are not wealthy political candidates. To say that the government is enjoined from preferring opposed over unopposed speech, and acting accordingly, seems in tension both with the idea that counterspeech is a constitutionally preferred remedy and with the idea, frequently incanted, that more speech is preferable to less. It is possible that the threat of counterspeech chills the speaker—as is always the case with counterspeech—but it seems at least as likely that counterspeech inspires a further response.

Though not directly stated, the implication of these cases is that the First Amendment would prevent the government from simply choosing to finance one candidate over another in a political campaign. If the government could do so, it would be difficult to understand why it could not impose the far lesser burden of selectively increasing a candidate’s private contribution limits, and on a viewpoint-neutral basis at that. Indeed, in Buckley v. Valeo, the Court seemed to permit Congress to do just this, by providing more public finance money to major party candidates than minor party candidates, also on a viewpoint-neutral basis. There’s still good reason to think that direct, discriminatory government support of a political candidate is indeed (or at least should be) constitutionally impermissible, but to ground even that prohibition in the First Amendment is flatly inconsistent with the idea that the First Amendment “does not regulate government speech.”

The Court’s case law gives us no resources for determining just what it is about government support for a political candidate that triggers First Amendment concerns. Is government speech in favor of the beef industry constitutionally different from government speech in favor of a political candidate who supports the beef industry? Being able to answer that question would get us some of the way toward determining whether and to what degree the Arizona law or the Millionaire’s Amendment implicates those same concerns. Without an answer to that question, we are left with the impression that facilitating counterspeech punishes speech and therefore is off-limits for the government unless it satisfies strict scrutiny.

C. Propaganda

The Oxford English Dictionary defines propaganda as “[t]he systematic dissemination of information, esp. in a biased or misleading way, in order to promote a political cause or point of view.” So defined, government propaganda is clearly constitutionally permissible. Generally speaking, the government speech doctrine allows the government to disseminate information to promote its point of view. The same doctrine makes clear that the legality of government speech also does not depend on whether it is biased or misleading, a standard that for familiar reasons would be difficult for judges to manage.

In addition to public education and workaday efforts at self-promotion such as buy American campaigns and public health measures, which Congress has long supported, U.S. states and the federal government can boast a storied history of propaganda. The United States government was a pioneer in wartime propaganda, with George Creel’s Committee on Public Information leading an unprecedented effort to win the support of a divided American public for World War I. There was “no medium of appeal that we did not employ,” Creel wrote of his committee, which employed some 150,000 people. “The printed word, the spoken word, the motion picture, the telegraph, the cable, the wireless, the poster, the signboard—all these were used in our campaign to make our own people and all other peoples understand the causes that compelled America to take arms.”

The fact that Mussolini and Hitler took inspiration from Creel’s efforts makes the dark side of mobilized propaganda, already obvious, even more visceral. Edward Ziegler writes that “a characteristic distinguishing democratic from totalitarian government is that while a democracy attempts to facilitate and ascertain public opinion and establish policy in accordance therewith, an autocracy attempts to engineer public opinion in support of its decisions.” Ziegler is wrong about this; all of the world’s democracies try to engineer public opinion to support their decisions. But the distinction he attempts to make captures an intuitive limitation on government speech. At some point, flooding the zone with relentless promotion of the government’s aims turns creepy. Some First Amendment scholars, including Scanlon, have identified the interests of listeners as equally important in justifying free speech as the interests of speakers. We might identify overly aggressive propaganda as the listener’s analog to compelled speech. At some point, compelled listening becomes too much to bear. It’s hard to say precisely what that point is, but you know it when you hear it.

IV. Reconciliation

The above discussion reveals some gaps in our constitutional reckoning with the government’s promotion of its own viewpoints, especially in detriment to the viewpoints of others. These gaps emerge in part out of the Supreme Court’s tendency to make categorical statements about free speech rights. If freedom of speech is presumptively absolute, then it follows that government speech must of course be immune from scrutiny under the free speech clause. Government could not function without making expressive choices between competing policy positions and outcomes. But if the government is presumptively not allowed to discriminate on the basis of viewpoint, then it follows that the government of course may not single out those whose speech it disfavors and selectively direct its resources against them. Both of these conclusions make sense, but they seem irreconcilable.

The first step in resolving this tension is abandoning the strong categorical presumptions that produce it. Both poles on the spectrum from absolute tolerance to absolute prohibition of government speech seem unacceptable. It should be clear by now that neither is there some clear line in between, though I have more to say about this below. The question turns on a range of factors that do not always point in the same direction: The question is not whether but when government speech is unduly coercive and discriminatory. Without committing to a broader theory of government speech, the discussion below surveys several variables that might reasonably influence our assessment of when government counterspeech should be constitutionally suspect or not.

A. The government’s purposes

Consider first the purposes behind the government counterspeech. Government speech intended to correct harmful misinformation or limit its spread is performing a public service that is arguably consistent with the objectives of the First Amendment. Government speech intended to secure the election of a particular candidate or party, to denigrate disfavored individuals or groups, to lie or mislead, or to coerce people into complying with the wishes of public officials lacks any constitutional dimension, if it has any political-moral legitimacy at all.

Government lies are protected speech, as noted, but existing doctrine finds intentions relevant in deciding whether harmful lies can lead to liability. A public official who is telling intentional falsehoods or recklessly disregarding known facts in responding to an individual’s speech should not stand in the same posture constitutionally as one acting in apparent good faith or relying on apparent facts.

We might also wish to distinguish trying to persuade people to believe the government’s version of reality or adopt its position from trying to persuade people directly to act upon those beliefs. Persuading people to act is not illegitimate, per se, but the possibility that it can blur into coercion makes it worthy of greater scrutiny. The line between counterspeech and regulation divided the Supreme Court in the Arizona case. Whichever side was right in that case, it seems fair to say that the farther the government is from that line, the better. Thus, coercive counterspeech by someone empowered to make life difficult for its audience—the police department, the DA’s office, or the chair of the appropriations committee, say—may be viewed differently from the same as similar counterspeech by the public library or the beautification commission.

B. “Public” vs. “private” speech

One complication of placing limits on government speech is that the government comprises individuals who have their own free speech rights. This need to draw a distinction between speech on the government’s behalf and speech by government officials is well understood in the establishment clause and free exercise context. The government is not allowed to enact policy for religious reasons, but government officials are allowed to invoke religion in justifying their policy choices. This line is difficult to draw, but the dual commands of the First Amendment religion clauses require it.

There is no free speech establishment clause, but the best argument against it sounds more in judicial manageability than in political morality. There is much to be said for the idea that the First Amendment limits the ability of the government to use the resources of the state to promote partisan political positions. But the same First Amendment allows government officials to participate in public discourse, as the Supreme Court’s public employment cases recognize. The line between these contexts may often be blurry. The proximity of government speech to the private views of its officials should accordingly offer some protection from First Amendment scrutiny. Thus, a press conference or a public address should not ordinarily be viewed in the same light as a customized license plate program or a subsidy scheme. The subsidy context in particular is one in which spending money in support of speech cannot simply be assimilated to speech.

C. The availability of alternatives

The concern with government propaganda is not just that it may be false or misleading but also that it comes from an official source. Information that carries the imprimatur of the state may, depending on the context, be more trustworthy and may be more likely to capture the attention of the media and the broader public. The response that the marketplace of ideas metaphor offers to speech regulation—more speech and all that—assumes that the state is one participant among many in that market. But if the market is not a competitive one and the state enjoys greater market power than others, whether due to its size, its financial resources, or its coercive power, it undermines the idea that truth is being fairly crowdsourced rather than dictated from on high.

Government counterspeech seems essentially unobjectionable if the state must compete for attention on equal terms with the cacophony of voices present on social media and elsewhere in the speech environment. The resources at the government’s disposal mean the playing field will never quite be level, but there will always be significant imbalances along this dimension. There may even be circumstances in which the government’s voice is needed to correct that imbalance.

We cannot say a priori whether the state’s intervention into public discourse is for good or ill, and we are likely to disagree about it in any event. But if the state’s voice is the only one present, or one of just a few, we have little reason to think First Amendment values are being well served. The idea that one factor in free speech analysis should be the availability of “alternative channels for communication” is of course familiar to First Amendment law as one step in the “time, place, or manner” test. Using a similar test to inform when we should be skeptical that government speech is consistent with the First Amendment makes sense for similar reasons.

D. Transparency

Government propaganda can be overt or covert. We should prefer the former. Gia Lee has descried instances of the federal government surreptitiously reviewing the scripts of broadcast television shows for suitably anti-drug messages or running literal fake news programs on local television stations produced, without attribution, by government agencies. As Lee notes, obscuring the source of speech undermines the means by which humans process it and evaluate its truth or reasonableness. When the government conceals its own role in promoting particular messages, it undermines the ability of citizens to hold it accountable.

The United States, of course, enjoys a venerable tradition of anonymous speech, from Thomas Paine’s Common Sense to the Federalist Papers to the Constitution itself, whose authorship was unknown at the time it was published. Anonymity can protect unpopular speakers or whistleblowers from abuse or retaliation and can promote reasoned evaluation of the content of speech free from the bias that knowledge of the speaker can sometimes produce. These rationales carry less force and are sometimes outright counterproductive in the context of government speech. As Lee writes, “[i]n some respects, it is precisely against society’s interest to insulate government speech from retaliation.”

As strong as the First Amendment’s speech protections are, the Supreme Court has held disclosure requirements to a less demanding standard than strict scrutiny. Given the strong democracy-enhancing reasons for government transparency in this context, government efforts at anonymity or misdirection in expressing counterspeech should be viewed skeptically.

E. Elections

Elections are special. They are our central channels of political change. If the First Amendment imposes any limit on government speech, it must bar the government from specifically directing public resources toward the election campaigns of favored candidates. As Thomas Emerson has said, “[i]t is not the function of government to get itself reelected.” Either the First Amendment prevents the government from speaking in this way or it does not prevent the government from speaking at all. One senses that this norm underlays the Court’s decision in the Arizona public financing case. The Arizona law did not merely increase the public resources available to particular candidates; it did so in direct response to the First Amendment activities of their electoral competitors. The Court somehow did not see the case as implicating government speech perhaps because directing speech in what the Court viewed as a politically discriminatory manner felt so clearly out of bounds.

It is conceivable to locate this norm somewhere other than the First Amendment. Under the equal protection clause, government discrimination solely based on the partisan identity of recipients of state funding would presumably be arbitrary and therefore illegal. The potential for other constitutional homes for this kind of anti-discrimination norm overdetermines the illegality of partisan government speech but does not completely answer the concern. In Wooley v. Maynard, the Supreme Court held that drivers with ideological objections must be permitted to cover up the “Live Free or Die” language on New Hampshire’s license plates. Had the plates read “Vote Democrat or Die,” they should have been disallowed on their face, for the same reason the federal government should not be allowed to project the GOP’s elephant logo onto the Capitol dome. The problem would be less that someone’s equal protection rights were offended than that the government would be using its power and privilege to distort the electoral speech environment. This would “abridg[e] the freedom of speech” even in the absence of a cognizable equal protection injury.

Of course, government speech inescapably may distort the speech environment in other contexts as well. And even in the electoral context, we might believe the government is sometimes justified in its “distortion.” In the Arizona case, for example, we might characterize the public finance law as evincing a government preference for an electoral speech environment that did not overly privilege the interests of wealthy financiers. The Court said in Buckley that a selective public finance scheme was permissible insofar as it “use[d] public money to facilitate and enlarge public discussion and participation in the electoral process, goals vital to a self-governing people.” If this is right, a strong norm against government electoral speech should be specific to partisan speech or favoritism towards particular candidates.

The majority and the dissenters in the Arizona case could indeed be characterized as disagreeing over whether the law at issue fit this description. Writing for the majority, Chief Justice Roberts pointed to the fact that, as in Davis v. FEC, “‘the vigorous exercise of the right to use personal funds to finance campaign speech’ leads to ‘advantages for opponents in the competitive context of electoral politics.’” Justice Kagan’s dissent found it relevant that “[t]he program does not discriminate against any candidate or point of view,” implying that her position might be different if it did.

F. “Generality”

Recall Wooley, the New Hampshire license plate case discussed above, in which the Court held that motorists could obscure the state’s “Live Free or Die” motto. Above, I suggested that “Vote Democrat or Die” should be impermissible on its face. What if instead the license plate had said, “Be White or Die,” or “Joshua Smith of Portsmouth Must Live Free or Die”?

Neither of these cases implicates the electoral process, at least not in the direct sense of the “Vote Democrat” hypothetical. We might instinctively believe that the first may give rise to some equal protection issues, but the Supreme Court has been especially emphatic that racially discriminatory speech is constitutionally protected. The Court has given us little reason to believe that racially discriminatory government speech, absent some nonexpressive injury, receives any less protection than it would if spoken by a private citizen.

Still, both these hypotheticals describe acts that should be legally dubious, and for similar reasons. As a general matter, the government should act impersonally, especially when its conduct is contemptuous. An emphasis on generality pops up at several points in the Constitution. For example, Article 1 bans bills of attainder and ex post facto laws for both states and Congress. As Akhil Reed Amar has explained, “the basic tripartite structure of the federal government reflected a strong commitment to the ideal that legislation, at least if punitive, should be general and prospective. Otherwise, a legislature could simply impose penalties upon political opponents by name.” Designating specific people as criminals or “reverse engineer[ing] an attainder” by retroactively criminalizing conduct that has already occurred upsets a separation of powers logic that assigns adjudication to judges rather than politicians.

Notably, the bill of attainder and ex post facto prohibitions are two of the five provisions that the original Constitution and Bill of Rights apply identically both to states and to the federal government. The other three—the ban on titles of nobility, the due process clauses, and the criminal jury trial guarantees—all touch upon the generality question in their own ways. Titles of nobility tie the exercise of political power to a person’s bloodline—who they are—rather than their actions. The due process clauses ensure that life, liberty, and property are not deprived arbitrarily, as through personal vendettas, without evidence of actual wrongdoing. And jury trials appeal to the sense of the community in part to ensure that people are punished only in ways that could be applied broadly to all.

These constitutional norms do not directly constrain government speech, but they fortify the republican notion that the government should act on the basis of ideas and evidence, rather than identity. Discriminating against a particular speaker because of the ideas they have expressed is qualitatively different from discriminating against them because of who they or their parents are. The closest the existing case law gets to this idea may be in the cases governing the power of political parties to exclude members. The Supreme Court has held that political parties, though quasi-private, lack the constitutional power to exclude members on the basis of race or status as clergy. But parties have been read to have associational rights that permit them to exclude nonmembers from participating in their primary elections and to impose ideological litmus tests on party members and candidates. It’s hard to make sense of these cases without understanding a party’s ideological commitments to be a constitutionally distinct and privileged subset of its other associational preferences, which are fair game for judicial scrutiny.

The distinction between the government denigrating individuals or groups based on their status versus their beliefs or actions gets at a deeper concern about the duties the government owes its citizens. The First Amendment’s putative requirement, from Holmes’ pen, that the government must respond to ideas with ideas rests on the premise that we have a shared and therefore commensurable vocabulary for exchanging views about the world. Responding to an idea with an ad hominem upsets this premise. Jeremy Waldron has written of the “public good of inclusiveness that our society sponsors and that it is committed to.” The formation of democratic culture in which we all have a right to participate suffers when the state marshals its resources so as to undermine or throw in question the free and equal citizenship of individuals or identity groups. The instances in which this happens through the exercise of government speech may not be per se unconstitutional, but they are reason for pause and conscientious assessment.

As with other factors above, applying a norm of generality poses a challenge to judicial review, especially when tossed into a complex mélange of crosscutting considerations and judgment calls. I am not committed to the view that the “test” of government counterspeech I have laid out above is judicially manageable. It might not be. But that does not mean it cannot act as a test of constitutionality. One way in which it can perform that role is as a supplement to existing doctrinal lines. Mark Yudof has written, for example, that principles of government nonpartisanship might be helpful in adjudicating cases implicating existing doctrine such as school speech and public employee cases. Courts can likewise refer to the other factors noted above to give interpretive guidance in the way of an avoidance canon. For example, in cases in which the Court is deciding whether to shield the government from liability by invoking the government speech doctrine over factually adjacent but less permissive doctrines (e.g., public forum, employee speech, or unconstitutional conditions doctrine), the government’s lack of transparency, say, or its selective targeting of individuals or particular classes could be factors in determining whether the safe harbor of “government speech” should be made available.

Beyond judges, public officials can refer to these principles as they make judgments consistent with their oaths of office, and citizens can revert to these factors in their evaluation of officials’ performance. The failure of the Supreme Court even to notice the internal tension in its case law suggests that the norms around government counterspeech remain immature. The challenges of online governance will require those norms to grow up fast.

V. Implications

Part II offers a range of hypothetical actions a government might take in response to perceived misinformation. Bringing the factors discussed above to bear upon those hypotheticals yields some tentative conclusions.

It is obviously unproblematic for the government to do no more than to respond to falsehoods with its own version of reality. The legality and appropriateness of this response constitutes the minimum core of government speech doctrine. It serves as the edifice upon which much of modern First Amendment law is built.

The principles offered above suggest some limits, however, on the aggressiveness with which the government may respond to misinformation. For example, if it relies selectively on other users to promote content, that information should not be a state secret; for good measure, the state might offer a publicly accessible database where it catalogs its online activity, including its efforts at strategic amplification. At a minimum, such information should be available via the Freedom of Information Act and state analogs. Likewise, transparency principles suggest that any government-funded media organization disclose its funding relationship when it reports stories that are conspicuously favorable to the government’s account of events.

Principles of generality demand that the government make its case without ad hominem attacks or racial appeals. The government can argue that content is misinformation, but it should not call someone a liar. This difference matters. The government should also not be selective in the users or speakers it targets for counterspeech. Anyone posting analogous content with analogous reach should receive analogous treatment.

Were the government to create some public infrastructure for hosting online content, courts would likely be skeptical about whether it could engage in viewpoint discrimination. In cases involving the efforts of public officials to block users from government-controlled platforms, courts have generally held that, to the degree these were indeed public forums, first-order First Amendment principles would apply. Other cases point in a different direction. The Supreme Court has, for example, permitted the Texas Department of Motor Vehicles to discriminate based on viewpoint in selecting customized license plates, on the ground that license plate logos are government speech. A public internet platform could perhaps be sufficiently branded and curated to likewise qualify and therefore be empowered to remove misinformation, but there’s no telling. Heightened transparency around the rules of such a platform would offer some assurance that the government was not acting arbitrarily. Outsourcing moderation to a structurally independent body or agency might offer further assurance.

First Amendment discourse no longer may avoid these questions. The misinformation crisis has gotten away from us, beyond the reach of purely private innovation. We are right to be wary of the government’s footprint in this space, but we should be at least as wary of the government staying out of it.

 

© 2022, Jamal Greene. 

 

Cite as: Jamal Greene, Government Counterspeech, 22-16 Knight First Amend. Inst. (Dec. 16, 2022), https://knightcolumbia.org/content/government-counterspeech [https://perma.cc/R6NL-V5C9].