Abstract

The First Amendment tolerates—has long tolerated—the regulation of certain kinds of false speech. Indeed, regulable lies are not limited to traditionally less-protected categories of speech like defamation and commercial deception. They include an array of other established speech regulations, administered by government institutions every day, from criminal laws barring perjury and other lies to government officials, to disciplinary measures by elected bodies sanctioning members for false or otherwise objectionable speech. Yet while it is easy to identify the kinds of lies that existing doctrinal categories make regulable for the personal, physical, or reputational harms they inflict on individuals¸ it has been far less clear what other constitutionally “cognizable” harms support this separate class of lies equally long subject to regulation. This essay argues that those other cognizable harms are best understood as including the structural interests of constitutional democracy—a concept that extends not only to preserving the integrity of elections but also to protecting the functional operation of other governing institutions, and to maintaining an information environment sufficient to support a system of constitutional democratic governance. While existing jurisprudence and long-recognized principles of constitutional theory offer ample support for this distinct category of democracy harms, the doctrinal fiction that no such category exists reinforces a perverse conception of the civil right to free expression—as one that tolerates burdens in the interest of remediating private harms (like injuries to reputation), but not in the interest of public harms (like injuries to the institutional mechanisms of democratic accountability). Furthermore, focus on the maintenance of existing categories as the critical safeguard against excessive speech regulation obscures the Court’s chronic failure to develop a consistent or meaningful understanding of the potentially far more consequential limit on government regulation of speech: The empirical requirement that regulators demonstrate a content-based rule is actually necessary to achieve their democracy-protecting goal and is carefully tailored to achieving it.

Introduction

As partisan polarization and social violence have risen in recent years, and public faith in governing institutions has declined, scholars across a remarkable range of academic disciplines have embraced concerns that one or more features of our current information ecosystem is functioning to widen cracks in core structures of U.S. constitutional democracy. The causes for concern are not hard to see. Social media and other online communications platforms have featured centrally in repeated, high-profile attacks on democratic processes directly—from the Russian disinformation campaign aimed at influencing the presidential election of 2016, to the organization of the physical invasion of Congress based on false claims of election fraud in 2021. At the same time, partisan cable media preferences, the algorithmic influence of online platforms, and longstanding professional conventions of traditional media have amplified the spread of false information or on occasion declined to circulate true information, seeming to operate in more diffuse ways to undermine prospects for the kind of compromise or even shared perception that democratic governance requires. And indicators that the existing information environment drives, or at least reinforces, sharp partisan divisions among Americans abound. Today, more than two-thirds of Republicans (but just 3 percent of Democrats) believe that Joe Biden’s election was the product of voter fraud.

Yet while these developments have given rise to profound concerns about the future of constitutional democracy—a system of government featuring not only regular popular elections but also independent political and judicial authorities, respect for the rule of law, a free press, and at least baseline protections for individual freedom and civic equality —debate over how we might remedy the current state of affairs in the United States has been hamstrung by the persistence of two enduring fictions surrounding how our Constitution protects the freedom of speech.

The first is the expectation that the harms caused by most any problematic speech can best be corrected by, as Justice Brandeis famously put it, “more speech.” If the goal is correcting false perceptions of reality or misguided beliefs, the theory goes, natural competition in the “marketplace of ideas” best ensures that truth will out. But as a growing number of scholars, and at least two Supreme Court justices have lately noted, reasons abound to believe that expectation has become “obsolete.” Confronted with a choice among effectively infinite sources of information, Americans’ behavior has reflected long-recognized cognitive habits like confirmation bias and conformity bias, which lead us to seek out information sources that are shared by our own identity groups and that confirm our existing views. Worse, empirical studies have demonstrated that corrections of misinformation do not invariably lead listeners to abandon reliance on misinformation heard in the first instance. Indeed, because corrections are filtered through listeners’ pre-existing beliefs and allegiances, they can have the effect of further entrenching listeners’ commitment to the original, mistaken belief. In the meantime, organized disinformation campaigns—by states and private actors, foreign and domestic—have proven highly effective in leveraging various contemporary communications tools not just to amplify falsehoods but to flood the “market” with contradictory information, raising concerns that the predominant effect is that listeners simply tune out.

A second fiction has proven equally persistent: the belief that the First Amendment prohibits the content-based regulation of speech in all but a handful of specifically defined categories including defamation, incitement of imminent lawless action, fighting words, true threats, commercial fraud, child pornography, and obscenity. These categorical exceptions, all geared toward preventing traditionally recognized, personal, physical, or reputational harms to individuals, have played a powerful rhetorical and conceptual role for both the Court and would-be federal regulators grappling with the challenges of the current information ecosystem. As the Court insists, its jurisprudence leaves no room for the “startling and dangerous” notion that any regulation of speech is subject merely to some “free-floating … ad hoc balancing of relative social costs and benefits,” or that any other categorical exception, for false speech or otherwise, exists.

Yet here too, reality shows the Court engaged in far more “free-floating” balancing than it likes to admit—whether in the growing set of decisions in which the content-based regulation of protected speech has survived its strictest (once thought unsurvivable) test of constitutional scrutiny, or in the arguably greater number of cases in which the Court has pretended (wrongly by its own terms) that a content-based rule is actually content-neutral. Add to these a rich array of established and court-supported laws—from criminal prohibitions of perjury and other lies to government officials, to disciplinary measures by elected bodies sanctioning members for falsehoods —and it becomes impossible to insist that the Court believes either that unrestricted speech always best promotes truth, or that the only “legally cognizable” harms for which speech may be regulated are the particular individual harms captured by the usual categorical exceptions. Taken as a group, these cases and laws make plain that the First Amendment tolerates a range of regulatory safeguards to shield against specific structural harms to democratic governance, from protecting the functional operation of government institutions, to preserving the integrity of elections, to maintaining some meaningful “marketplace of ideas.” Formally categoryless this array of speech regulations may be, but functionally, they amount to their own categorical exception—an exception for what we might broadly call structural democracy harms. Reflected in existing doctrinal understandings in U.S. constitutional law, as well as familiar principles of First Amendment theory, and the international human rights law the United States helped shape, this body of law recognizes a proposition so well settled it commonly goes overlooked: that while freedoms of thought, opinion, or belief may never be subject to government restriction, freedom of expression can be subject to narrowly tailored regulation if it is “actually necessary” to preserve a democratic public order.

Recognizing these democracy harms as equally among the list of “legally cognizable” justifications for speech regulation has several implications. For one, it helps expose how conventional statements of First Amendment doctrine have the effect of privileging speech regulations aimed at remedying harms to individual interests alone. Regulations against defamation and true threats are certainly important. But the law’s more or less arbitrary habit of relegating democracy harms to the status (at best) of undesignated other “speech integral to criminal conduct” reinforces a misguided assumption that there is little the government can permissibly do to regulate features of the information ecosystem that damage core structural concerns of constitutional democracy. Breaking free of categorical expectations opens room for reformers to think more broadly—and more critically—about solutions to our current information dysfunction. In the interest of protecting election integrity, for example, might it be constitutional to charge a domestic political organization with criminal fraud for engineering a disinformation campaign regarding vote counts or results? The existence of a democracy harms justification suggests such a prosecution might be defensible, even under existing law.

But embracing democracy harms as a “legally cognizable” justification for speech regulation answers only one of the questions courts ask in checking such an application’s constitutionality. The other question is harder, and woefully underdeveloped in existing First Amendment doctrine and scholarship: whether and how one might demonstrate in any persuasive way that a particular regulation is “actually necessary” to cure or mitigate the identified harm and is adequately tailored to that end. Focus on the maintenance of existing categories as the critical safeguard against excessive federal speech regulation has obscured the Court’s chronic failure to develop a consistent or meaningful understanding of the potentially far more consequential limit on government regulation of speech: the evidentiary requirement that regulators demonstrate, as the Court has on occasion put it at its most stringent, “a direct causal link between the restriction imposed and the injury to be prevented.” While multiple off-the-shelf bodies of law and legal institutions offer a menu of evidentiary burdens, standards of proof, and process requirements that might provide helpful analogies to guide and constrain an adjudicator’s evaluation of this question, attention to categorization of speech has made it easier for the Court to avoid applying any fixed standards—and for regulators to avoid having to build a record to meet them. This should change. For while First Amendment jurisprudence to date has rested heavily on a fictional account of how truth prevails in constitutional democracies, preserving an information environment capable of sustaining such a system going forward requires more serious contact with the empirical.

Fake News About False Speech

While the First Amendment is written in absolute terms as a constraint on government power—“Congress shall make no law abridging the freedom of speech, or of the press” —the Court has never taken the amendment at its word. As it has with all constitutionally protected rights, the Court has subjected government restrictions on speech to a balancing test. When the government seeks to regulate speech on the basis of its content, it bears the weighty burden of demonstrating that its proposed rule is “necessary” to achieve a “compelling” government interest and is indeed “narrowly tailored” to that end. The only way of avoiding this strict and near-certainly fatal degree of constitutional scrutiny, the standard story goes, is if the speech being regulated falls into one of a handful of categorical exceptions of “unprotected” or “less protected” speech—common law torts like libel and defamation, criminal offenses like “true threats” against individuals and incitement to imminent violence, and regulatory prohibitions of obscenity and commercial fraud. Speech capable of destroying someone’s reputation, inciting a crowd to violence, or the like, causes such serious harms to individuals, the Court has reasoned over time, that society’s interest requires bending the usual First Amendment rule in order to prevent anyone from suffering them. Laws aimed at this kind of speech are thus regularly upheld so long as the government can show the law is more or less plausibly well aimed at remedying the (mostly private) harm it seeks to correct.

Yet this standard summary of the doctrine is a far cry from the more complex reality of the Court’s actual case law in this realm. Both state and federal governments have, over time, enacted a host of laws that regulate speech for reasons other than preventing injuries to reputation, protecting children, and the like. And as it turns out, the Court has upheld a great many of those other speech regulations in the interest of protecting one or another feature of constitutional democracy: preserving the function of independent judicial and political institutions, protecting the integrity of elections, and promoting public confidence in the existence of a system of constitutional governance. Take speech regulations justified as protecting the basic functioning of government institutions, for instance. To ensure the effective operation of the judicial system, criminal laws prohibit perjury on the grounds that such lies “undermine[] the function and province of the law,” “because it can cause a court to render a ‘judgment not resting on truth.’” The Constitution equally permits courts to impose various content-based controls on the speech of lawyers, police officers, and even ordinary witnesses—also in service of ensuring the effective functioning of the judicial branch. States can likewise prohibit candidates for judicial office from personally soliciting campaign funds—because “it is the business of judges to be indifferent to popularity,” because courts must be able to function with the “actuality of judicial independence,” and because the structure of the Constitution requires the maintenance of a “distinction of judges from politicians.”

The judicial system isn’t the only independent institution protected by speech regulation. Longstanding federal laws prohibit false statements to executive agency officials and likewise prohibit falsely representing that one is speaking as a government official or on behalf of the government. Indeed, more than 100 federal criminal statutes punish false statements made in connection with areas of federal agency concern —all designed to, among other things, ensure that communications purporting to count as official authorization are reliably taken at their word. To preserve institutional effectiveness, the government can impose significant speech restrictions on military bases, in prisons, and in schools. It is to ensure the same kind of functional stability that the U.S. House and Senate have likewise asserted the power to discipline members for their speech (through, for example, expulsion, censure, reprimand, or fine), including speech on the House floor. Whether the punishment of member speech is intended to protect the decorum or orderly function of legislative proceedings, or to avoid the proliferation of falsehoods in committee business, Congress’ behavior has at least since the 19th century reflected its understanding that neither the First Amendment nor any other provision of the Constitution bars the legislature from regulating speech by its own members if a legislative majority believes the speech could compromise the institution’s democratic purpose.

The Court has proven equally willing to tolerate content-based speech regulations to the extent it is persuaded they are necessary to preserve the institutional function of election administration. Upholding a Tennessee law prohibiting core political speech within 100 feet of polling places, the Court recognized the state’s “compelling interest in protecting voters from confusion and undue influence,” and in protecting “the integrity and reliability of the electoral process itself.” As long as the regulation was “generally applicable and evenhanded” and could be shown “necessary” to protecting the right to vote “freely and effectively,” such a rule could survive First Amendment strict scrutiny. It was thus not surprising to see the Court note in passing in an unrelated case more recently that “[w]e do not doubt that the State may prohibit messages intended to mislead voters about voting requirements and procedures.” Content-based speech regulations surrounding elections may be subject to strict scrutiny for all the same reasons content-based regulations in other contexts are—to ensure the regulation is no more speech-restrictive than necessary to fulfill the government’s democracy-protecting interest—but the danger of democracy harm has surely been great enough to ensure that sufficiently tailored regulations might survive review.

Beyond preserving the functional operation of the institutional organs of constitutional democracy, speech regulation may be permitted in the interest of protecting a nongovernmental but equally indispensable feature of constitutional democracy: a functional “marketplace of ideas,” capable at least at some baseline level of supporting—or not wholly disabling—the public’s role as gatekeeper of democratic decision-making. The recognition that our system of government would depend on the public’s ability to access information necessary to assess the performance of their elected representatives could hardly be more apparent in the framers’ conception. Hamilton, among others, was clear that citizens would require access to “the information of intelligent men” about the conduct of government so they could hold elected officials accountable. In modern jurisprudence, the Court has thus upheld the government’s ability to regulate broadcast and cable media to ensure those media “function consistently with the ends and purposes of the First Amendment,” including the purpose of ensuring that the public has access to competing speech on questions of public debate that is “essence of self-government.” Government can likewise, at least to some extent, compel companies to speak—just as much an infringement on First Amendment rights as prohibiting speech directly—by imposing detailed disclaimer and disclosure requirements on electioneering communications. Such regulations, the Court has explained over decades, equally serve the compelling interest in “help[ing] citizens ‘make informed choices in the political marketplace.’” Whether or not the First Amendment protects any affirmative right of access to such information, the Court at one time even allowed that the government cannot through legal intervention, or through the protection of private monopoly power, leave voters with no more than “a barren marketplace of ideas that had only sellers and no buyers.” As the Court has long understood, “[t]he dissemination of ideas can accomplish nothing if otherwise willing addressees are not free to receive and consider them.”

At farthest remove from the relatively concrete harm caused by speech that impedes the function of democratic institutions is a democracy harm of a third sort: injury to the people’s belief in the existence of a system of constitutional governance. This harm is worth examining closely, for in one sense, the notion that a constitutional democracy might tolerate the regulation of speech in the interest of promoting or securing some (any) perception or belief seems counterintuitive in the extreme. As Justice Jackson put it most famously: “If there is any fixed star in our constitutional constellation, it is that no official, high or petty, can prescribe what shall be orthodox in politics, nationalism, religion, or other matters of opinion.” Indeed, the maintenance of a freedom of belief and opinion is commonly cited as among that which distinguish constitutional democratic governance from more authoritarian forms.

Yet the body of First Amendment case law just canvassed relies heavily for its justification not solely on the need to maintain government’s operational effectiveness (keep the trains running, as it were), but on the recognition that there is a “vital state interest” in maintaining “public confidence in the fairness and integrity” of democratic structures sufficient for government to function. To serve this end, the government can, among other things, regulate the speech of candidates running for judicial election; criminalize the false representation that one is speaking on the government’s behalf; or impose speech regulations on a legislative body’s own elected members to preserve public faith in the body. Just as states bar law enforcement officers from appearing too close to polling places lest they lead voters to perceive there might be coercion in the electoral process, the state may impose limits on direct campaign contributions, disclaimer or disclosure requirements, in order to guard against not just actual corruption but the danger that the public believes its representation to be corrupt. The concept of “public confidence” or constitutional “faith,” the Court has noted, “does not easily reduce to precise definition, nor does it lend itself to proof by documentary record.” Nonetheless, “no one denies that it is genuine and compelling.”

It is here important to caution that the Court has hardly proven willing to uphold all speech regulations aimed at one or another harm to these elements of constitutional democracy. Advocates once defended a state right-of-reply law that required newspapers to provide equal space to political candidates criticized in their pages as advancing just the marketplace-of-ideas interest noted above—ensuring the public has access to a range of views about the relative merits of public officials and their policies. Likewise, the government’s interest in ensuring public confidence in elections (untainted by financial corruption) was centrally behind campaign finance regulations prohibiting corporations from making independent expenditures advocating for the election or defeat of a candidate. Yet the Court struck down both of these regulations as running afoul of First Amendment protections. That democracy harms can justify speech regulation plainly does not mean all speech regulations enacted to serve such a purpose are constitutionally sound.

At the same time, these cases cannot be read as calling into question the constitutional “cognizability” of the democracy-supporting goals government regulators were aiming to advance. On the contrary, neither Miami Herald v. Tornillo (striking down a state right-of-reply law) nor Citizens United v. FEC (striking down federal restrictions on corporate campaign expenditures) calls into question the legitimacy of the democracy interests in promoting a robust marketplace of ideas, or in maintaining public confidence in the integrity of the system of representative government. The problem with the Tornillo and Citizens United regulations, in the Court’s view, was their inability to survive means-ends scrutiny. In the Tornillo Court’s estimation, rather than ensuring a greater diversity of political views reached the public, “the penalties that would accrue to any newspaper that published news or commentary arguably within the reach” of the right-of-reply law would as likely lead editors to avoid publishing political views altogether—producing precisely the opposite of the democracy-enhancing effect the statute aimed to achieve. The Citizens United Court likewise balked at the restriction on corporate campaign expenditures not because it believed that protecting public faith in democracy was a constitutionally uncognizable goal, but rather because the Court was unpersuaded of the regulator’s fundamentally empirical claim that this regulation was necessary: “The appearance of influence or access ... will not cause the electorate to lose faith in our democracy.” (The validity of the basis for these causal findings, and the method by which the Court reached them, is a critical and remarkably underexamined subject to which this essay returns briefly below.)

The Court’s willingness to consider upholding speech regulations in the interest of preserving democratic faith might be one thing if it were limited to regulations protecting judicial legitimacy per se. The Court’s own relative institutional powerlessness has long made it acutely aware that governmental effectiveness depends at least partly, and in the Court’s case, exclusively, on public belief in the legitimacy of its decisions. But the Court’s willingness to entertain speech-regulatory remedies to protect against a loss of public faith extends beyond the courts alone. Even granting that a loss of public faith in structural fairness and legitimacy may cause important harms to the function of constitutional democracy, it is not exactly apparent why it should be within contemplation that speech might be regulated to avoid this end. The next section thus looks more closely at the theoretical basis for recognizing this and other democracy harms as among those capable of justifying the regulation of speech. For present purposes, the point is this: While a great number of the examples just given involve the regulation of false speech, the speech is understood as regulable not solely or especially because it is false, but because its unregulated expression is inconsistent with the maintenance of core structures of constitutional democracy. In that sense, this body of law is easily described as a “category” of “legally cognizable” interests supporting the regulation of speech.

Defining Democracy Harms

If the foregoing cases show anything, it is that there is no clear doctrinal reason why speech likely to incite imminent harm to individuals (for example) counts as a recognized “category” of speech subject to regulation under the First Amendment, but speech that actually harms the functioning of key democratic institutions does not. The cases provide ample support for the existence of some form of exception for both. It is true that the Court has required the categorically excepted kinds of speech to pass only a modest degree of judicial scrutiny, while requiring the government to show for some of the latter that it had a compelling goal in regulation, and that the regulation was tailored as narrowly as possible to serve that goal. But this difference may well be more rhetorical than real. Categorically regulable, but otherwise “high value,” speech tends to incorporate similarly strict limits by definition. For any prosecution for criminal incitement to pass constitutional muster, for example, the government must prove not merely that the speaker said something controversial contributing to an overall greater risk of violence, but that the speaker intended to incite imminent lawless action, and that the regulated speech was likely to incite such action. Put differently, incitement prohibitions must also be narrowly tailored to address only that speech at greatest risk of triggering social violence. Indeed, under a constitution that understands speech as “the matrix, the indispensable condition, of nearly every other form of freedom,” it makes sense that any exception, democracy harms included, must satisfy some measure of detailed and careful justification.

The absence of meaningful doctrinal reasons for viewing certain, even potential, private harms as categorically “cognizable,” and other, even actual, public harms as not necessarily so, thus raises a potentially troubling question: Why should the recognition of structural democracy harms be relegated to this less visible, more ad hoc degree of constitutional consideration? The easiest explanation is perhaps just historical happenstance. Torts like defamation and libel were recognized at common law well before the Constitution, so one might reasonably assume the Constitution’s framers believed such laws would be excepted from coverage by the new Constitution. Harms to the new system of government could not have been, in that sense, grandfathered in. Yet it is hard to imagine that this is the only reason for the noncategorical status of democracy harms. The Court has expressly acknowledged that harms to U.S. national security count as an independent category potentially justifying speech regulation, a concept (U.S. national security) that certainly lacked common law or any other meaning before the formation of the United States. In contrast, apart from the specific cases in which they arise, the Court has never described democracy harms as a category of “legally cognizable” harms justifying the regulation of speech.

A more powerful explanation might thus be historical experience: The ignominious Sedition Act of 1798 and later history of prosecution for seditious libel-like offenses that followed may have made the modern Court rightly wary of overt doctrinal recognition of speech exceptions for harms to democracy. For the same reason it should be exceedingly difficult in a democracy to penalize citizens for mistaken but otherwise good faith criticism of public officials, it should equally be unthinkable to penalize speech merely because it, in some even vaguer way, undermines “public confidence” in democratic governance. Yet the assumption that all democracy harm-based speech regulations necessarily resurrect the specter of seditious libel prosecutions helps expose how undertheorized democracy harms remain in contemporary jurisprudence. For beyond the extraordinary overbreadth and vagueness common to historic seditious libel laws, these laws were constitutionally problematic not to the extent they aimed at preventing speech that undermined the functioning of government, but to the extent they aimed at preventing harm to the reputation of those who governed. As the New York Times v. Sullivan Court explained in what remains the Court’s most thorough repudiation of the constitutionality of the original Sedition Act, civil or criminal liability for false criticism of a public official “transmut[es] criticism of government, however impersonal it may seem on its face, into personal criticism, and hence potential libel.” Where conviction under the most objectionable clauses of the Sedition Act required proof of “intent to defame,” a constitutional democracy-harm regulation would require proof of intent to disrupt some aspect of the function of constitutional governance. One can indeed find evidence of this distinction in the Sedition Act’s statutory descendants, still very much on the books today, including seditious conspiracy, and conspiracy to defraud the United States. These charges—now the subject of pending indictments against individuals accused of conspiring to disrupt congressional proceedings on January 6, 2021, and others accused of conspiring to spread disinformation in the run-up to the presidential election of 2016—have at their heart allegations of a spoken or written agreement and any other step to obstruct the lawful function of democratic government.

Given all this, it seems plausible that at least part of the explanation for democracy harms’ categoryless status is that the category itself lacks sufficiently well-developed theoretical or historical grounding to prove conceptually coherent in practice. True, the cases canvassed above offer a common set of justifications—all supporting the position that whatever the First Amendment means, it must not render unworkable the functional operation and maintenance of constitutional democratic governance, broadly defined. But that may not be sufficient. The definition of “obscenity” has, after all, been notoriously difficult to pin down, raising deep concerns of fundamental fairness in enforcing prohibitions against it, and engaging the Court in the institutionally unstable role of chronic arbiter of social value and public morality. Another undefined, inadequately defined, or even incoherent category of regulable speech serves neither rule-of-law interests in doctrinal predictability, nor structural interests in preserving a constitutionally limited role for the courts. Furthermore, we might fairly worry that highlighting a new, insufficiently defined set of “cognizable” harms makes any exception easier to expand or exploit—makes it more tempting, for example, for government regulators to suppress or penalize any form of disfavored speech. Beyond the Court’s own arbitrary collection of decided cases, the concern might be, there is simply insufficient development of the idea to guide legislatures, prosecutors, or courts in how to address democracy harms without running afoul of core First Amendment principles.

Where, then, to start? Without undertaking to reinvent several centuries of constitutional theory here, one might begin by noting the idea that the First Amendment was adopted in key part to promote democratic self-governance is among the most well-known, arguably least controversial, 20th century theories of freedom of speech. Famously associated with Alexander Meiklejohn, the theory is expressly at the heart of the Court’s decisions recognizing that the First Amendment depends upon a meaningful “marketplace of ideas”—a marketplace said essential to informed democratic decision-making. The idea equally undergirds the doctrine permitting speech regulation in the interest of ensuring the operational effectiveness of governing institutions. Beyond the amendment’s role in ensuring that individual voters would be adequately informed to exercise the right to vote—its protection of the rights of listeners as well as speakers—the amendment was to operate in service of the constitutional democratic plan of self-government, in which “all matters of public policy shall be decided by corporate action, that such decisions shall be equally binding on all citizens, whether they agree with them or not, and that, if need be, they shall, by due legal procedure, be enforced upon anyone who refuses to conform to them.” Institutional governance was part and parcel of the system Meiklejohn understood the amendment to protect. As Nikolas Bowie thus put it much more recently: “The First Amendment, along with the rest of the Constitution, was adopted to create a functional government out of the embers of a failing state.”

One can even find support in this theoretical tradition for the proposition that the government can regulate speech in the interest of protecting against damage to public confidence in the possibility of constitutional governance. In keeping with political scientists’ more contemporary definition of constitutional democracy, Meiklejohn’s successors have explained constitutional democracy as in part a system of governance that “is achieved when those who are subject to law believe that they are also potential authors of law.” As Robert Post put it: “Elections and other mechanisms that we ordinarily associate with democratic decision making are simply institutions designed to maximize the likelihood” that the relationship of government to citizen is like that of character to author—a universe in which citizens understand that they are in control. To be clear, this view does not justify the government in regulating or attempting to control popular belief. That would indeed be contrary not only to the self-government model but to every principled theory spelling out the purpose of the First Amendment. The idea instead is that to preserve conditions necessary for citizens to engage in the practice of self-governance, the government may impose some restrictions on popular expression.

In the modern era, the distinction between the regulation of belief and expression in the interest of protecting a constitutional democratic order is perhaps no more visible than in the body of international human rights law the United States was instrumental in bringing into existence. At the same time that great dissenting opinions of Justices Brandeis and Holmes defending robust speech protections finally began commanding majorities of the Supreme Court in the 1950s and 60s, the United States was actively working to codify these principles into international law as, among other things, an essential antidote to the Cold War authoritarianism that the United States viewed as democracy’s greatest ideological competition. Drafted in the 1950s and early 1960s, and opened for signature in 1966, the International Covenant on Civil and Political Rights (ICCPR) made express several principles the U.S. Supreme Court was itself coming to embrace. For example, where the First Amendment had expressly named only freedoms of speech, religion, press, and assembly—not until later affirming that these words necessarily encompassed freedoms of thought and belief —the ICCPR’s Article 19 made explicit from the outset that human rights included both freedoms of opinion and thought, as well as freedom of expression.

Likewise, where the First Amendment can most easily be read as protecting these freedoms in absolute terms (“Congress shall make no law. …”)—only for the Supreme Court to discover over time the existence of various exceptions—the ICCPR’s Article 19 codified expressly the limited set of exceptions that might justify regulation: “respect of the rights or reputations of others; … the protection of national security or of public order (ordre public); or of public health or morals.” Avoiding the dubious message that rights protections are more subject to exception for private harms than for harms burdening society as a whole, Article 19 recognized that public and private regulation-supporting harms are of a piece, and that “public order,” including the function of governing institutions, may justify the regulation of expression so long as the state can demonstrate it “necessary.” In short, Article 19 made express the principle our Court had come to recognize implicitly: that while freedom of thought, opinion, or belief may never be subject to government restriction, freedom of expression may be subject to narrowly tailored regulation in the interest of preserving a democratic public order.

How much should this treaty matter in helping us understand the meaning of our own Constitution? Purely as evidence of our own government’s understanding of the justifications for regulation of expression, it offers no small measure of insight. Negotiated by U.S. government officials over decades, signed by President Carter, and ratified under President George H.W. Bush, Article 19 was originally part of a proposed convention entirely on the freedom of information, which the United States successfully opposed for containing too many exceptions for the regulation of speech. The final text of the ICCPR’s Article 19 with its named exceptions instead tracks nearly verbatim the formulation the United States introduced and preferred. The U.S. government made plain from the outset that it had no intention of entering into any agreement it thought would impair the freedoms of speech or press guaranteed by our own First Amendment. To make good on that commitment, the United States ultimately withheld its consent to the ICCPR’s Article 20 (prohibiting, among other things, war propaganda). But we agreed without reservation to the terms of Article 19. So long as Article VI of our own Constitution continues to recognize treaties as part of the “supreme law of the land,” this one seems a better historical source than many in helping shed light on our own government’s 20th century understanding of the First Amendment’s meaning.

In more practical terms, our borderless information environment has driven multiple major American communications enterprises to adopt and integrate international human rights law in general, and the ICCPR in particular, into their own rules guiding content moderation choices on their platforms—platforms like Facebook/Meta and Instagram used in the United States and around the world. Whether in facing domestic litigation involving these platforms, or in encountering any new legislation Congress might adopt with a view (again) to harmonizing European and American standards in the interest of doing business, neither U.S. courts nor lawmakers will likely be able to avoid altogether the need to untangle how First Amendment jurisprudence is or is not consistent with the principles of Article 19. To the extent functional considerations have ever driven the Court’s interpretation of the meaning of the U.S. Constitution—and evidence is overwhelming that such considerations matter a lot —it seems more than a little relevant that the ICCPR likewise tolerates at least some degree of “necessary” speech regulation in the interest of protecting the democratic public order.

First Amendment Administration: An election integrity example

All of which brings us back to the question of how we might usefully, and constitutionally, remedy particular democracy-harming effects of our current information ecosystem. In the interest of protecting the functionality of evidentiary rules and judicial processes, would it be constitutional for the federal government to impose criminal penalties on anyone circulating “deepfake” video content not clearly labeled as such? In the interest of preserving a meaningful “marketplace of ideas,” could federal regulation constitutionally require that social media algorithms be designed to present readers news from politically diverse sources? In the interest of protecting voter confidence in elections, would it be constitutional to charge a domestic political organization with criminal fraud for engineering a disinformation campaign regarding vote counts or results?

This last example is perhaps most useful for assessing the impact of a doctrinal approach that takes democracy harms seriously, since it is possible to imagine the government pursuing more aggressive measures to combat this form of disinformation using laws already on the books. To some extent, it already has. Following the 2016 presidential election, criminal investigators succeeded in identifying key individuals responsible for organizing and implementing a multipronged online disinformation campaign designed to, among other things, suppress turnout among African Americans. Multiple defendants have since been charged, seemingly without controversy, with criminal conspiracy to defraud the United States, for their efforts to “sow division and discord in the U.S. political system, including by creating social and political polarization, undermining faith in democratic institutions, and influencing U.S. elections.” The criminal conspiracy charge in these cases requires the government to prove: “(1) the defendant entered into an agreement, (2) to obstruct a lawful function of the Government, (3) by deceitful or dishonest means, and (4) committed at least one overt act in furtherance of the conspiracy.” That the individuals charged in these cases are Russian nationals (still beyond the enforcement jurisdiction of the United States) may well explain the lack of public controversy. But the First Amendment question remains open. If it is constitutional to charge foreign nationals operating in part inside the United States with criminal fraud for mounting an election disinformation campaign, is it constitutional to prosecute American nationals conducting a similarly organized campaign about actual vote results with criminal fraud as well?

A forthright acknowledgment of the extent to which democracy harms count as a “legally cognizable” justification for speech regulation makes quick work of the first step of any First Amendment analysis, even assuming strict scrutiny applies. The United States’ interest in preserving the integrity of its federal elections and voter confidence in them is plainly compelling. True, this regulation does not involve direct operation of polling places, but it surely involves public perceptions of official corruption—both circumstances in which the Court has expressly identified the government’s democracy-protecting interest. To the extent a disinformation campaign could be shown to disrupt a “lawful function of any department of Government” like election administration (as the statute requires), or indeed to undermine “public confidence in the fairness and integrity” of democratic structures, it would seem to fall comfortably within the categorical scope of democracy-harming speech.

Clarity about the doctrinal “cognizability” of democracy harms hardly answers the First Amendment question. But it does help focus the debate past questions of “cognizability” to the far more complex part of the First Amendment inquiry—whether the challenged law is “necessary” or “narrowly tailored” enough to pass constitutional muster. (Even laws regulating some formally “less protected” kinds of speech require the government to show that the law will directly advance the government’s interest and be no more extensive than necessary to serve that interest. ) The goal of protecting against democracy harms cannot justify criminalizing speech, even in the form of organized disinformation campaigns, unless that regulation is “necessary” to relieve an obstruction in governmental functions, or to restore lost “public confidence in the fairness and integrity” of elections caused by the disinformation itself.

Yet given the central importance of this inquiry to the constitutionality of speech regulations, it is remarkable that the Court has never particularly settled what it believes is required to demonstrate “necessity” in the speech context—what kind of evidence one might successfully marshal to do it, or how close the demonstrated causal relationship between ends and means must be. It is clear the Court views the means-ends inquiry as empirically dependent; the Court has said as much expressly more than once. But justices disagree fundamentally on what kind of empirical showing is required. At times, the Court relies heavily on deference to the findings of the legislature; at others, the Court insists relevant expertise is essential (in executive agencies or among witnesses); on still other occasions the justices conduct their own “considerable independent research” to act as fact-finder in the first instance, or simply revert to a default evidentiary presumption that more speech is always the more narrowly tailored option. The type and quality of evidence the Court considers equally varies—from legislative testimony and litigation witnesses, to peer-reviewed studies in the relevant social science, to the Court’s own assessment of logic or historical experience. The Court’s description of how clear the relationship between means and ends must be is especially conflicted. The Alvarez Court insisted “[t]here must be a direct causal link between the restriction imposed and the injury to be prevented.” The Burson Court thought it sufficient to rely on the mere rationality of a “widespread and time-tested consensus” that limiting political speech around polling places was necessary to achieve the government’s concededly compelling interest.

The Court’s apparent reluctance to identify a more consistent standard, or even to inquire too deeply into what its standard is, is perhaps a function of its own discomfort, its embrace of the sensible worry “that it is perilous to permit the state to be the arbiter of truth.” Yet questions requiring state actors to assess empirical reality of one kind or another are chronically before the Court and indeed arise in a host of settings across all branches of U.S. government, every day—in criminal courts, in regulatory adjudication and rulemaking, in legislatures charged with congressional redistricting, and more. In all of these other settings, we have not handled the fallibility of government decision-makers by denying the existence of the decision. We handle it, rather, by guiding and limiting the discretion of the decision-maker through what have become elsewhere a deeply developed array of tools for cabining otherwise untrustworthy government decision-making: by fixing burdens of proof and standards of causation. The Court’s vehement insistence that it would never engage in “free-floating ... balancing” —rather than developing more meaningful guidelines that might limit how that balancing is conducted—has in this respect served neither the interests of free expression nor the interests of constitutional democracy.

But taking that next step reveals just how much work remains to be done, both doctrinal and empirical. In our election integrity example, while research available so far is strongly suggestive that disinformation is at least in part the cause of some democracy harms, the answers of whether criminal prohibitions might be necessary to remedy that harm remain far from clear. There is certainly evidence that suggests organized disinformation campaigns are harmful not only to public confidence in election administration but also the actual function of election administration. At least one important study found that Americans’ sharply polarized beliefs about the dangers of voter fraud in the 2016 election were the result of an organized disinformation campaign, a campaign that capitalized on standard mass media conventions to focus attention on elite opinion, and that was amplified by social media. By the 2020 election, Gallup opinion polling showed that 41 percent of Americans doubted that votes would be cast and counted accurately, a precipitous deterioration in voter confidence compared to the same poll just two years earlier—a decline pollsters believed to be driven largely by widespread but unsubstantiated claims in 2020 that have now left roughly two-thirds of Republicans convinced that President Biden’s election was the product of fraud. Citing these dangers of fraud directly, state legislatures across the country have since enacted laws making it harder for voters to cast a ballot; at least two dozen states have enacted laws that weaken independent election administration by, for example, empowering partisan poll watchers and imposing criminal penalties on workers for minor violations; still other jurisdictions have conducted partisan purges of local election boards. Most acutely, as the establishment of a new Justice Department task force on the matter reflects, election administrators across the country now face chronic, serious threats of violence; many have taken steps to hire personal security, flee their homes, or find counseling for their traumatized children; and growing numbers are leaving election work altogether. As a bipartisan group of 150 of the nation’s leading democracy scholars put it in a public letter urging more federal action to protect elections: “[The] politicization of what has long been trustworthy, non-partisan administration of elections represents a clear and present threat to the future of electoral democracy in the United States.”

Still, even if one can demonstrate that an organized disinformation campaign is something like the but-for cause of these democracy harms, noncriminal, putatively less-speech-restrictive proposals for achieving the same democracy-protecting ends abound: supporting the reestablishment of a more robust traditional press; breaking up monopolistic media enterprises through antitrust laws; requiring that social media companies alter ranking algorithms to deprioritize demonstrably false or unsupported claims of election results; or requiring that such claims be labeled to that effect. The very empirical findings that have undercut the “more speech” approach to bad speech paint a picture that is far from certain. Proposals that would have the effect of adding more channels of information (like antitrust approaches) raise the risk of further information polarization; proposals that require adding additional or corrective information (like labeling requirements) raise the prospect that corrections will only lead to further entrenchment of false beliefs, or that listeners bombarded by too much contradictory information will simply tune out. Proposals aimed solely at social media miss the significant, and perhaps greater, effect of traditional partisan media channels in spreading disinformation. Laws that do aim restrictions directly at social media–like European laws requiring social media companies to take down content amounting to hate speech—have perversely had the effect of dramatically increasing the quantity of content deleted, as platforms operate in “anticipatory obedience” of the dictates of government regulators, pulling down far more content than genuinely poses any harm in an effort to spare themselves the legal risk and operational expense of a more careful system to assess content. Labeling requirements, at least by social media platforms, can produce some effects, but how much and to what extent remains difficult to assess; platforms have no current obligation to disclose publicly whether, how, or to what extent they are implementing their own content moderation rules, and indeed, apart from a few widely reported instances where labels were applied, answers to the effect of many labeling questions remain obscure.

Untangling the actual empirics is hard enough. It is a problem compounded by the Court itself, with an approach that is no better than ad hoc in identifying inferences about the alignment of means with ends on its own. It is entirely appropriate and essential to worry about the competence of government institutions to serve as “arbiter[s] of truth.” But the truth is, government actors are doing that already. Every day. The best way of guarding against the potentially oppressive consequences of such an enterprise is not to deny that the Constitution leaves room for speech regulation in the interest of democracy. The best way is for constitutional lawyers to seek, and would-be reformers to insist, on both more consistent, meaningful doctrinal expectations—and more than theoretical help. In this respect, the most important task for sustaining robust protections for speech in a struggling constitutional democracy is not discerning what the Constitution means, but to demand that regulators understand—and demonstrate—how the information environment the Constitution sustains actually works.

 

Printable PDF

 

© 2022, Deborah Pearlstein.

 

Cite as: Deborah Pearlstein, Democracy Harms and the First Amendment, 22-08 Knight First Amend. Inst. (Oct. 19, 2022), https://knightcolumbia.org/content/democracy-harms-and-the-first-amendment [https://perma.cc/HJ66-S4RA].