Policy Analysis No. 865

Why the Government Should Not Regulate Content Moderation of Social Media

President Trump recently complained that Google searches are biased against Republicans and conservatives. Many conservatives argue that Facebook and Google are monopolies seeking to restrict conservative speech. In contrast, many on the left complain that large social media platforms fostered both Trump’s election in 2016 and violence in Charlottesville in 2017. Many on both sides believe that government should actively regulate the moderation of social media platforms to attain fairness, balance, or other values.

Yet American law and culture strongly circumscribe government power to regulate speech on the internet and elsewhere. Regulations of social media companies might either indirectly restrict individual speech or directly limit a right to curate an internet platform. The First Amendment offers strong protections against such restrictions. Congress has offered additional protections to tech companies by freeing them from most intermediary liability for speech that appears on their platforms. The U.S. Supreme Court has decided that private companies in general are not bound by the First Amendment.

However, some activists support new efforts by the government to regulate social media. Although some platforms are large and dominant, their market power can disintegrate, and alternatives are available for speakers excluded from a platform. The history of broadcast regulation shows that government regulation tends to support rather than mitigate monopolies.

Others worry that social media leads to “filter bubbles” that preclude democratic deliberation. But the evidence for filter bubbles is not strong, and few remedies exist that are compatible with the Constitution.

Speech on social media directly tied to violence—for example, terrorism—may be regulated by government, but more expansive efforts are likely unconstitutional. Concern about “interference” in U.S. elections glosses over the incoherence of current policies. Some foreign speech, online and off, is legal if the relationship of a speaker and a foreign power is disclosed.

Preventing harms caused by “fake news” or “hate speech” lies well beyond the jurisdiction of the government; tech firms appear determined to deal with such harms, leaving little for the government to do.

Continue to full version

Background

In August 2017, a political protest in Charlottesville, Virginia, turned into violent clashes between extremists, leading to one person being killed. In the aftermath, several tech companies denied service to neo-Nazis whose extreme rhetoric was thought to foster that violence. Denied a forum, the extremists retreated from the most widely used part of the internet to the dark web. Matthew Prince, the CEO of Cloudflare, one of the companies that drove the National Socialists out, argued later that businesses lack the legitimacy to govern speech on their forums.1 He suggested that most people see government as the proper authority to suppress speech related to violence.

This policy analysis follows up on Prince’s comments by evaluating the legitimacy of government regulation of speech on the internet. We shall focus primarily on potential policies for the United States.2

Our effort advances in two parts. First, we establish a starting point for our analysis. We show that the values and practices of the public demonstrate a legitimate but quite limited role for government regulation of speech on the internet and elsewhere. The public and policy­makers prefer private governance of speech. Those who wish to introduce new public regulation of social media must overcome this presumption for the private. In the second part, we show that arguments for new public efforts fail to do that. We find that private content moderators have already taken effective and innovative steps to deal with some of these problems. It is private content moderators, not elected or appointed officials, who should have the power to regulate speech on social media.

The Presumption against Public Regulation of Social Media

What are social media? Many experts have offered different definitions. Tom Standage, deputy editor of The Economist, says social media are “two-way, conversational environments in which information passes horizontally from one person to another along social networks, rather than being delivered vertically from an impersonal central source.”3 He identifies aspects of social media that bear on free expression:

[Social media] allow information to be shared along social networks with friends or followers (who may then share items in turn), and they enable discussion to take place around such shared information. Users of such sites do more than just passively consume information, in other words: they can also create it, comment on it, share it, discuss it, and even modify it. The result is a shared social environment and a sense of membership in a distributed community.4

Jonathan A. Obar and Steve Wildman of the Quello Center at Michigan State University add that social media are interactive and “can be characterized as a shift from user as consumer to user as participant.”5 Of course, social media users also consume what others create, but they are no longer primarily consumers of existing material. Thus, content generated by users is the “lifeblood of social media.”

Social media are platforms, not publishers. They provide the means for large numbers of people to produce and consume information. They are open to both producers and consumers. Social media managers regulate the content on a platform, but the platform does not host everything that is posted on it. The regulation is necessarily ex post. The number of users and their expectations of immediate publication preclude ex ante regulation.

In contrast, publishing involved a small number of people communicating information to mass or special audiences. Gatekeeping was inherent in publishing; it was relatively closed to producers but open to consumers. The constrained supply of content enabled ex ante regulation of a publication (i.e., gatekeeping).

Social media are, of course, economic institutions; they need to generate or have a prospect of generating revenue beyond the costs of providing the service. Individuals create a user profile that social media services in turn use to connect individuals to others.6 Social media services often use data gleaned from users to target advertising to them.7

Social media thus comprise four groups of people: users who generate content, users who consume content, users who generate commercial speech (advertising), and social media managers who host speech. Each element involves speech: users generate and consume information, and social media services create the forum in which speech happens. Individual speech is highly protected in the United States. The online activities of social media companies also have considerable protection from government regulation.

The United States highly values individual speech in the public sphere. The Constitution offers strong protections for speech in general and not just for political speech. Similarly, the right to hear the speech of others is pro­tected by the First Amendment.8 American law recognizes a small number of exceptions to these general protections for speech.9 Apart from these exceptions, speech by and for social media users may be presumed to be free of government regulation. For many years, legal experts believed commercial speech had fewer protections from government regulation than political or artistic speech. But this lesser standing has been challenged by the Supreme Court and recent scholarship.10

Economic regulation may also violate the individual’s right to free speech. Social media companies seem to be dependent on ordinary commercial transactions, the regulation of which is presumed constitutional.11 But the exchange underlying social media is not an ordinary commercial transaction. Individuals use social media for speech. They are granted access to social media in exchange for data about themselves. If government blocked (prohibited) that exchange, speech by individuals would be restricted. The prohibition of the economic transaction would be tantamount to prohibiting speech. The validity of less sweeping regulations would involve discerning their effects on speech. However, this exchange is clearly sensitive from a First Amendment standpoint. The exchange underlying social media thus implicates both commerce and fundamental rights. Some part of the protection for social media from government action derives from the protections accorded individual speech.12

Social media may enjoy protections from government independent of their users’ right to free speech. The owners of the companies involved may have First Amendment rights that preclude government requiring a platform to carry speech.13 Publishers have a right to editorial discretion over what to publish.14 Like publishers, platform managers choose what will appear on their platform; after all, not everything sent to the platform stays on it.15 Besides removing content, platform managers also rank content, thereby affecting the likelihood it will be seen by users. Both activities are similar to a publisher’s editorial choices and deserve First Amendment protection.16

Yet social media platforms differ from traditional publishers. The regulation of social media content typically comes ex post after a user is found to have violated a platform’s community standards. One might argue that ex post editorial decisions are less likely than ex ante decisions to involve expression. A company might remove existing content to create a pleasant experience for users (that is, for business reasons). Thus, social media content moderation might differ from editorial discretion in publishing and merit less constitutional protection.

This objection fails on two counts. First, newspapers need to make a profit to remain financially viable and thus might make editorial decisions for business reasons; they would still enjoy the freedom of the press. Moreover, an ex post content removal by social media managers could also be expressive. For example, one could argue that Facebook’s content moderation both expresses and applies a conception of its online community, which in turn favors some values over others.17 The act of moderating speech on social media would therefore be expressive and likely protected by the Constitution.

This similarity to traditional publishers might appear to make social media companies liable for defamation or other legal limits that apply to publishers. However, Congress has exempted social media from defamation standards applicable to traditional publishers. Section 230c(1) of the Communications Decency Act of 1996 says that “no provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.”18 Congress adopted Section 230 for two reasons. First, “Congress wanted to encourage the unfettered and unregulated development of free speech on the Internet.”19 Social media managers who were concerned about liability for user-provided content would tend to remove speech that both did and did not defame others. Second, the section sought “to encourage interactive computer services and users of such services to self-police the Internet for obscenity and other offensive material, so as to aid parents in limiting their children’s access to such material.”20 Social media would here also be free of liability for removing such content. In general, through Section 230, Congress “sought to further First Amendment and e-commerce interests on the Internet while also promoting the protection of minors.”21 Section 230 also frees “online intermediaries from the need to screen every single online post, a need that would render impossible the real-time interactivity that people expect when they engage on social media.”22

Congress might have required social media to police their platforms to enforce accepted public standards for speech (e.g., liability for defamation). But Congress did not do so. At the same time, the section did not protect individuals from public accountability for violating the limited exceptions to freedom of speech, and it did not reduce government authority over those harms. Section 230 neither increased nor decreased government authority over speech on social media. Congress showed in these decisions a preference for private governance of internet speech.

Finally, social media are privately owned forums for speech. The First Amendment protects the freedom of speech from state action. Social media are not government and hence are not constrained by the First Amendment. These platforms are protected by the First Amendment but need not apply it to speech by their users. Social media managers may suppress speech on their privately owned platforms, speech that elected officials could not censor in a public forum.23 Court decisions support this distinction between public and private power.24 But some nuances merit attention.

In older decisions, the Supreme Court said private forums had to respect public limits regulating freedom of speech. In Marsh v. Alabama (1946), the court ruled that a company town, like a government, could not restrict First Amendment rights: “The more an owner, for his advantage, opens up his property for use by the public in general, the more do his rights become circumscribed by the statutory and constitutional rights of those who use it.” The fundamental rights of speech and religion outweighed the rights of private property.25 In 1968, the court also found that a shopping mall was the “functional equivalent” of the company town opened for public use and thus had to respect a right to picket on private property.26

But more recent cases have been friendlier to protecting private forums from public regulation. In Lloyd Corp. v. Tanner (1972), the Supreme Court overturned Logan Valley, saying a shopping mall did not constitute a public forum and thus need not obey the First Amendment.27 Four years later, in Hudgens v. NLRB (1976), a majority noted that

while statutory or common law may in some situations extend protection or provide redress against a private corporation or person who seeks to abridge the free expression of others, no such protection or redress is provided by the Constitution itself. This elementary proposition is little more than a truism.28

Thus, the First Amendment’s freedom of speech clause offered no protection for pickets making their case in a privately owned enclosed shopping mall.

Pruneyard Shopping Center v. Robins (1980) affirmed the proposition that being open to the public did not turn a shopping center into a public forum governed by the First Amendment.29 However, the decision depended on other considerations. The court also concluded that the California Constitution created a right to petition and speech even on private property. The court decided that the state’s rights trumped any claims to protection for private property or the owner’s freedom of speech.30 This extension of free speech rights did not appeal much to other states. Only six states beyond California adopted limited protections for speech on private property.31

Some Americans may believe free speech should trump private property in forums like shopping malls. But national judgments run the other way, supporting private governance of speech. The history of public values and social media suggests a strong presumption against government regulation. The federal government must refrain from abridging the freedom of speech, a constraint that strongly protects a virtual space comprising speech. The government has also generally refrained from forcing owners of private property to abide by the First Amendment. Hence individuals have no expectation that the government will force the owners of social media to refrain from suppressing speech on their platforms (provided the owners do not violate civil rights laws). Those who seek more public control over social media should offer strong arguments to overcome this presumption of private governance. Have they?

Arguments for Public Regulation of Social Media

Since 1934, Congress has required broadcast media to operate in “the public interest.”32 That standard meant more than maximizing the size and satisfactions of an audience. Among other goals, broadcasters were required to carry speech they might not otherwise carry, allegedly in pursuit of the public interest defined as “more news and information programming.”33

Some idea of the public interest often undergirds government actions, including regulation of private firms. In other words, policymakers and others see government vindicating a public interest through regulation. A public interest argument comprises two parts. First, it should establish that government action is needed to secure some widely held value; private activity is assumed, in theory or fact, to be inadequate to achieving that end. Second, it should make the case that government action will achieve the values in question without significant costs to other important values. As we have seen with social media, fundamental values are at stake. The second part of the public interest argument must climb a steep incline.

This section will argue that proposed government regulations of social media fail on one or both criteria. The values pursued by regulation are more important than the restraint of government power. Government action is unlikely to attain the public interest cited by advocates of regulation. Finally, in some cases, the regulation may only attain some value at an unacceptable price in other rights and values.

The Anti-Monopoly Argument

Some critics argue that tech companies are monopolies.34 This claim is often confounded with a different argument. Critics say that tech companies fundamentally practice viewpoint discrimination when managing their platforms. In other words, the companies are said to exclude speakers from their forums because of their views; broadly put, critics argue that liberal employees at tech companies discriminate against conservatives when governing their private forums.35

If these firms are indeed monopolies, there would be a stronger case that their content moderation violates the First Amendment. But if these tech firms are not monopolies, then it matters much less whether their content moderation constitutes a violation of free speech. Private institutions discriminate among viewpoints all the time, and few would wish the government to manage their agendas to assure fairness or balance. Indeed, as we saw, the government may not manage their speech (unless in theory they are broadcast media). But if the monopoly claim is true, then bias in content moderation might matter more. If a private forum such as Facebook owns the only place to speak and to be heard, its discrimination among viewpoints will seem a lot like censorship by the government, notwithstanding its private status. If so, one might think government action (rather than constraint) would serve the cause of speech and debate. More voices might be heard if one company (or a small number) did not govern the private forum in question.

However, the question of viewpoint discrimination would matter a lot less if the dominance of current market leaders were insecure and if users and audiences who were excluded from a platform had alternatives. The discrimination argument also matters less if public regulation (e.g., turning social media into public utilities) seems likely to make matters worse regarding the monopoly question. Each of these contingencies appears true: the dominance of current firms is insecure, alternatives exist, and broad regulation seems likely to make things worse.

The tech companies involved in speech are large and successful. The Wall Street Journal reports that “these companies [Apple, Amazon, Alphabet (Google), Microsoft, and Facebook] are the five most valuable in the U.S. by market capitalization, the first time a single industry has occupied that position in several decades, according to S&P Capital Inc.”36 Their success, however, does not necessarily mean that they enjoy a natural monopoly, a traditional justification for government antitrust action.

The case against the tech companies leans on an older economic theory of network effects. David S. Evans and Richard Schmalensee offer a pithy summary of the theory:

In some cases a service is more valuable if more customers are using it because customers want to interact with each other. Then, if a firm moved fast and got some customers, those customers would attract more customers, which would attract even more. Explosive growth would ensue and result in a single firm owning the market forever. The winner would take all.37

Does this theory comport with what we know from economics and with empirical reality?

The economics of network effects has turned out to be more complicated than the older theory suggests. Internet companies offer multisided platforms whose network effects are indirect between different kinds of customers (say, smartphone users and app developers) rather than direct effects between the same kind of customers (such as telephone callers). These multisided platforms face a much more difficult challenge of attracting customers; they are much more likely to fail during their startup period than a telephone company, which is the model of the older theory of network effects. The new firms also must attend more to attracting the right customers than simply adding customers. Finally, network effects can go in reverse; customers may use multiple platforms and migrate to some of the alternatives. Evans and Schmalensee remark: “This process has happened repeatedly. AOL, MSN Messenger, Friendster, MySpace, and Orkut all rose to great heights and then rapidly declined, while Facebook, Snap, WhatsApp, Line, and others quickly rose.” In general, they find that “systematic research on online platforms … shows considerable churn in leadership for online platforms over periods shorter than a decade.”38

Although a few tech companies dominate some markets, that does not mean these firms can never be displaced.39 A more complex theory of network effects raises doubts about their dominance, while recent history suggests previously dominant firms are declining rather than continuing as monopolies.40 We have reasons to doubt that these firms will continue to dominate their markets.

Do alternatives exist for those excluded from social media platforms? The rapid rise of social media might suggest traditional forums for speech no longer matter, but that is far from true. Traditional public forums continue to exist along with traditional media such as newspapers and television. Such forums are protected, of course, from government censorship. In fact, most people still get most of their news from such sources.41 Fox News should be mentioned, considering that online bias has lately been a concern of conservatives.

Even speakers excluded from major platforms such as Facebook and YouTube can find a home for their speech somewhere else on the internet. LiveLeak, while less reputable precisely because of its willingness to host graphic content, will deliver video to viewers just as effectively as YouTube. Vimeo also exists as an alternative platform, which is owned by IAC, a firm that seemingly specializes in second-tier versions of a host of internet services, and Dailymotion is another option. There are also platforms that are not specifically dedicated to video hosting or sharing but are often used to do so. Snapchat can be used to send self-deleting clips, while file-storage sites like Google Drive or Dropbox are often repurposed as semipublic information repositories.

Not only is it important to recognize that alternatives exist, but that alternatives can continue to come into existence to meet user demand for differing standards of moderation. For several months, YouTube rules concerning videos containing firearms have shifted repeatedly with little transparency. The operators of gun review channels found their videos repeatedly demonetized, and some were banned for running afoul of opaque rulesets. A group of firearm enthusiasts decided to start a YouTube competitor, called Full30, which catered to the tastes of gun owners. Several popular firearms channels on YouTube have moved to the site. While it will never compete with YouTube as a full-spectrum video-hosting platform, it isn’t intended to do so. Instead, it provides a space for shooters to share videos without having to worry about the whims of gun-shy advertisers. In a matter of months, a firearms-friendly video-hosting site had gone from a newly discovered market niche, revealed through a dispute between YouTube and some of its users, to a functional video-hosting platform.42

But even if firms dominate their markets, will government regulation deal with the problem or make it worse? The nation has had experience with similar regulation of communications for similar reasons regarding broadcasting. Such regulation reinforced the market dominance of large firms and threatened freedom of speech.

The federal government claimed control of the broadcasting spectrum in the 1920s. Congress set up the Federal Communications Commission (FCC) to regulate entry into and use of the spectrum. Between 1941 and 1953, the FCC allocated a large part of the spectrum for television broadcasting. To use the spectrum for television broadcasting required a license from the FCC.

According to economist Thomas Hazlett, the agency “sought to carefully plan the structure and form of television service.” It also severely limited the number of competing stations, which drove up the value of the licenses. Such restrictions inevitably constrain competition for the benefit of incumbent firms. The successful licensees did not pay for the permission to use the spectrum, although they did agree to broadcast “in the public interest.” Hazlett notes that “this plan dictated technology and erected insurmountable barriers to entry.” The regulatory effort thus created monopoly power for the regulated.43 Hazlett quotes a later expert assessment: “The effect of this policy has been to create a system of powerful vested interests, which continue to stand in the path of reform and change.”44

Indeed, the Radio Act of 1927 reduced the number of stations by a quarter, affecting “stations owned almost entirely by small busin­esses or nonprofit organizations like schools or labor unions.”45 Moreover, on five occasions from 1962 to 1970 the FCC protected broadcast companies from competition by cable television.46

Eventually, the courts became less tolerant of the FCC’s hostility to competition and the agency itself became more open to markets in the 1980s. Those changes, however, came after decades of the FCC restricting competition in broadcasting markets.

The effects of FCC regulation on freedom of speech may be summarized briefly. The FCC’s Fairness Doctrine required a broad­caster to offer equal time to respond to a position taken on air. The equal time would not be reimbursed, which meant the requirement acted as a tax on the original broadcast speech. Small radio stations criticized the Kennedy administration’s Limited Nuclear Test Ban Treaty with the Soviet Union. Such stations could not afford to supply a great deal of free time. Kennedy operatives arranged for more than 1,000 letters to be written demanding equal time at these stations, leading to 1,678 hours of free broadcasting. This effort to suppress speech was deemed a success by the administration and continued in the Johnson years.47

Richard Nixon’s administration sought to use broadcast regulation against the television networks. Administration officials threatened the local licenses of the networks both publicly and privately, seeking more favorable coverage. The public effort appeared to fail, but Thomas Hazlett has shown that privately network officials were quite compliant with the wishes of the administration.48

The history of broadcast regulation suggests that increasing state control over social media would have a chilling effect on speech. Over time, both political parties might be expected to threaten any speech they find abhorrent.

The monopoly argument for regulating social media has weaknesses. We have reason to think the current market positions of large social media companies may not persist because network effects operate differently than in the past. In any case, speakers have alternatives if they are excluded from a specific platform. Finally, it should not be assumed that government regulation will produce more competition in the online marketplace of ideas. It may simply protect both social media owners and government officials from competition.

Democracy and Deliberation

Consider how political speech works in the world outside the internet. People have views about politics. They associate with others to discuss and perhaps debate those views. Yet associations of the like-minded are no doubt more likely than the debate club. Perhaps they seek out others with similar views because Americans do not like conflict and confrontation.49 The association of the like-minded, now disparaged as a filter bubble, existed long before the internet. Such associations reflected other human failings such as confirmation bias and prejudice. This tendency no doubt did harm to society: debates were less rich and less probing than they otherwise might have been, and citizens were worse off than they might have been if they had learned the errors of their ways through a fuller debate. Yet few called for the government to compel associations to hear speakers with different views.

The internet facilitated the exchange of views about everything, including politics. Once again, groups of the like-minded have formed. Indeed, the cost of speech and association has fallen so fast that we might expect that more people will be more involved in more like-minded groups than ever. One might see this as a tremendous success both in fostering speech and association or in satisfying individual preferences. But some see the association of the like-minded as a danger to democracy and to the individuals who associate this way.

Cass Sunstein, perhaps the most important contemporary critic of speech and association on the internet, argues that “we should evaluate communications technologies and social media by asking how they affect us as citizens, not only by asking how they affect us as consumers.” A central question is whether emerging social practices, including consumption patterns, are “promoting or compromising our own highest aspirations.”50 The government might help them achieve “our” (but not necessarily “their”) highest aspirations.51

Our highest aspirations are said to include meeting the demands of citizenship in a deliberative democracy. Sunstein denies that freedom in general means freedom from coercion by the state. Instead, freedom is understood as the development (including the self-development) of the individual. Freedom for Sunstein is a variation of the freedom of the ancients: it comes from collective action by citizens in a republic, which in turn “requires exposure to a diverse set of topics and opinions.”52

Individuals might be free of state coercion yet unfree because they make choices that preclude their own development. Individuals may prefer, for example, to hear only a subset of all views about a political topic. Indeed, they may prefer to hear little about politics. Sunstein offers an extended argument that people online pursue their own interests to the exclusion of public information and debates. Particularly, they do not come across ideas and arguments they might not seek out. Instead, they form bubbles that filter out opposing views and echo chambers that merely repeat the views already held by the individuals in them. In contrast, more traditional media like newspapers and television “expose people to a range of topics and views at the same time that they provide shared experiences for a heterogeneous public.”53 Ultimately, for Sunstein such exposure and such shared experiences are essential to foster the kind of deliberative democracy sought by the American Founders.54

If people prefer associating with like-minded people on the internet, Sunstein worries that more than political aspirations may be harmed. Associating this way creates “a large risk of group polarization, simply because it makes it so easy for like-minded people to speak with one another—and ultimately move toward extreme and sometimes even violent positions.”55 Obviously violence might follow polarization. It is even possible, Sunstein notes, that social stability could be put at risk.56 This would obviously involve public values on par with freedom of speech. But the real problem seems more prosaic and political: If diverse groups are seeing and hearing quite different points of view or focusing on quite different topics, mutual understanding might be difficult and it might be increasingly hard for people to solve problems that society faces together.57

By enabling and respecting individual choices, the internet complicates and even undermines both the diversity and the unity needed in a deliberative democracy. More diversity of views would improve the disharmony of the internet enclaves, and more unity across enclaves would militate against social and political fragmentation.

Sunstein’s claims about filter bubbles and echo chambers seem plausible. We can imagine people choosing to avoid unpleasant people and views while affirming their prior beliefs. Such choices might be the easiest way forward for them. But the logic of this position does not entail its empirical accuracy. Communications researcher Cristian Vaccari notes that

social media users can make choices as to which sources they follow and engage with. Whether people use these choice affordances solely to flock to content reinforcing their political preferences and prejudices, filtering out or avoiding content that espouses other viewpoints, is, however, an empirical question—not a destiny inscribed in the way social media and their algorithms function.58

In fact, abundant research casts doubt on Sunstein’s claim that individual choices on the internet are turning the nation into a polarized, possibly violent dystopia. For example, several studies published in 2016 and earlier indicate that people using the internet and social media are not shielded from news contravening their prior beliefs or attitudes.59 In 2014, experimental evidence led two scholars to state that “social media should be expected to increase users’ exposure to a variety of news and politically diverse information.” They conclude that “the odds of exposure to counter attitudinal information among partisans and political news among the disaffected strike us as substantially higher than interpersonal discussion or traditional media venues.”60 A 2015 paper found that “most social media users are embedded in ideologically diverse networks, and that exposure to political diversity has a positive effect on political moderation.” Contrary to the received wisdom, this data “provides evidence that social media usage reduces mass political polarization.”61

A broad literature review in 2016 found “no empirical evidence that warrants any strong worries about filter bubbles.”62 Just before the 2016 election, a survey of U.S. adults found that social media users perceive more political disagreement than nonusers, that they perceive more of it on social media than in other media, and that news use on social media is positively associated with perceived disagreement on social media.63

Did the 2016 election change these findings? Several studies suggest doubts about filter bubbles, polarization, and internet use. Three economists found that polarization has advanced most rapidly among demographic groups least likely to use the internet for political news. The cause (internet use) was absent from the effect of interest (increased polarization).64 Three communications scholars examined how people used Facebook news during the 2016 U.S. presidential campaign. They had panel data and thus could examine how internet usage affected the attitudes of the same people over time. The results suggest Sunstein’s concerns are exaggerated. Both internet use and the attitudes of the panel “remained relatively stable.” There was also no evidence for a filter bubble. The people who used Facebook for news were more likely to view news that both affirmed and contravened their prior beliefs. Indeed, people exposed themselves more over time to contrary views, which “was related to a modest … spiral of depolarization.” In contrast, the researchers found no evidence of a filter bubble where exposure to news affirming prior attitudes led to greater polarization.65

Several recent studies have focused on either the United States and other developed nations or just European nations alone. Perhaps data and conclusions from other developed nations do not transfer to the United States. However, cultures and borders notwith­standing, citizens in developed nations are similar in wealth and education. Even if we put less weight on conclusions from Europe, such results bear more than modest consideration.

In 2017, Vaccari surveyed citizens in France, Germany, and the United Kingdom to test the extent of filter bubbles online. He concluded that “social media users are more likely to disagree than agree with the political contents they see on these platforms” and that “citizens are much more likely to encounter disagreeable views on social media than in face-to-face conversations.” His evaluation of Sunstein’s thesis is as follows:

Ideological echo chambers and filter bubbles on social media are the exception, not the norm. Being the exception does not mean being non-existent, of course. Based on these estimates, between one in five and one in eight social media users report being in ideological echo chambers. However, most social media users experience a rather balanced combination of views they agree and disagree with. If anything, the clash of disagreeing opinions is more common on social media than ideological echo chambers.66

Another recent study of the United Kingdom found that most people interested in politics who select diverse sources of information tended to avoid echo chambers. Only about 8 percent of their sample were in an echo chamber. The authors urge us to look more broadly at media and public opinion:

Whatever may be happening on any single social media platform, when we look at the entire media environment, there is little apparent echo chamber. People regularly encounter things that they disagree with. People check multiple sources. People try to confirm information using search. Possibly most important, people discover things that change their political opinions. Looking at the entire multi-media environment, we find little evidence of an echo chamber.67 Finally, another study of multiple countries found that social media was also related to incidental exposure to news, contrary again to Sunstein’s view that older media promoted such unintended exposure while new media do not.68

For Sunstein, the aggregate of individual choices about political speech and engagement on the internet does not serve well the cause of republicanism. A proper culture for deliberative democracy “demands not only a law of free expression but also a culture of free expression, in which people are eager to listen to what their fellow citizens have to say.”69 He also believes that “a democratic polity, acting through democratic organs” may help foster such a culture by creating “a system of communications that promotes exposure to a wide range of issues and views.”70

In this regard, Sunstein follows an older view that sees the First Amendment and the Constitution enabling government to regulate speech to attain a “richer public debate.”71 That older view of activist government called for limits on the autonomy of some speakers to improve public deliberation.72 In contrast, Sunstein does not believe citizens should be “forced to read and view materials that they abhor.”73 Yet he clearly believes that public officials acting at the behest of majorities should have the power to expose individuals to materials they would not choose to see on their own: to nudge, not coerce, Americans for a worthy end that they would not choose on their own.74

Sunstein’s proposed reforms for the internet seem restrained in light of his critique of social media. He remarks, “I will be arguing for the creative use of links on the Internet, although I will not suggest, and do not believe, that the government should require any links.”75 He denies that government should mandate linking to a variety of sites with different opinions to achieve the public good of better debates.76 Sunstein notes the good consequences of ending the Fairness Doctrine and does not advocate restoring it for any media, including the internet, even though he believes its demise probably increased fragmentation and polarization.77

We have seen that Sunstein’s concerns about filter bubbles are open to question. Sunstein might disagree, and perhaps a maturing literature will support regulations to fight such bubbles. But Sunstein proposes restrained efforts to make internet users better citizens. If the literature cited in this report is correct, there is even less reason to regulate social media in the name of democracy. But there may be an inherent problem implicitly recognized in Sunstein’s limited proposals for reform. Forcing people to read and interact with views they dislike or abhor implicates liberal values such as free speech and individual liberty. On the one hand, we may wish that people were more and better informed about politics; on the other hand, we may doubt the wisdom of forcing people to engage in public matters. If filter bubbles threatened popular government, the case for public action might have improved. But studies do not support that proposition.

National Security Concerns

Government traditionally protects the homeland from its enemies. A standard textbook explores the complex meaning of national security:

The term national security refers to the safeguarding of a people, territory, and way of life. It includes protection from physical assault and in that sense is similar to the term defense. … In one definition the phrase is commonly asserted to mean “physical security, defined as the protection against attack on the territory and the people of the United States in order to ensure survival with fundamental values and institutions intact; promotion of values; and economic prosperity.”78

Our concern here is whether the government should increase its power over internet speech to achieve national security. That question inevitably concerns the relationship of speech to violence. But internet speech may involve other aspects of national security.

Terrorism. The clearest example of a threat to national security would be attacks on or occupation of the homeland or its citizens. Terrorism may be defined as “public violence to advance a political, social, or religious cause or ideology.”79 Policymakers thus have reason for concern:

Terrorist groups … use the Internet to disseminate their ideology, to recruit new members, and to take credit for attacks around the world. In addition, some people who are not members of these groups may view this content and could begin to sympathize with or to adhere to the violent philosophies these groups advocate. They might even act on these beliefs.80

Speech is not violence, but a speaker can “intend to incite a violent or lawless action” that might be likely “to imminently occur as a result” of the speech.81 Such speech is not protected by the First Amendment, and the government may act to restrict or punish it. However, the Supreme Court’s leading decision on incitement to violence also states that “the mere abstract teaching … of the moral propriety or even moral necessity for a resort to force and violence is not the same as preparing a group for violent action.” Such speech cannot be punished by the government in a manner consistent with the First Amendment.82 Terrorist speech that seeks to persuade rather than direct would likely escape censorship.

Some have argued for stricter limits on terrorist speech. For example, Eric Posner, a law professor at the University of Chicago Law School, has proposed a law that would make it a crime to visit “websites that glorify, express support for, or provide encouragement for ISIS or support recruitment by ISIS; to distribute links to those websites or videos, images, or text taken from those websites; or to encourage people to access such websites by supplying them with links or instructions.”83 (Presumably the same would apply to other terrorist groups.) He argues that the law would prevent naïve individuals from being drawn into supporting terrorism and thereby preclude deadly attacks. Posner concedes that his proposed law violates the First Amendment under current doctrine. However, he is hopeful that the current war on terror will permit new restrictions on speech that would have been held invalid in less demanding times. In the past, he remarks, war has supported such restrictions.

David Post, another American legal scholar, has noted some problems with Posner’s proposal. He argues that the history of suppressing speech during wartime has often later been judged to be “deeply misguided, counterproductive, and often shameful.” Post suggests that ambiguous terms such as “glorify,” “support,” and “encourage” may be interpreted to suppress legitimate dissenting speech.84 According to Posner, the work of noted First Amendment scholar Geoffrey Stone has established that war and speech suppression go together, but Posner does not mention that Stone believes the government’s actions were almost always unnecessary.

Posner seems concerned about two harms caused by speech that favors terrorism: the harm done to vulnerable individuals who end up being punished for materially supporting terrorism and the mayhem caused by the speech. Liberal governments generally do not protect people from the consequences of their beliefs; however, they do protect other people from those consequences if they are directly related to speech. Hence Posner rightly worries about violence caused by terrorist speech, a concern that informs the incitement exception in First Amendment doctrine. But his example, which is that 300 U.S.-based ISIS symp­athizers were “lured” by Twitter into some affiliation with the group, does not establish that any harm was perpetrated by anyone in the group against other people. There is little doubt they might cause harm in the future, but we have no evidence they have done so because of hearing speech. Posner is proposing a revived “bad tendency” test that weighs free speech against possible harms carried out by beguiled people. One would think we need at least a few cases in which speech probably caused harm to change long-standing doctrine.

Courts have consistently refused to hold social media platforms liable for terrorist acts.85 The most plausible of these attempts, Fields v. Twitter, sought to make use of the civil remedies provision of the Anti-Terrorism Act (ATA), contending that in failing to prevent ISIS from using its platform Twitter knowingly and recklessly provided material support to a terrorist organization, rendering Twitter the proximate cause of harms suffered by ISIS victims. Suits brought under the ATA turn on its proximate cause or “by reason of” requirement. Under this requirement Fields and similar material-support claims falter. While ISIS certainly found value in the ability to tweet, it is unlikely that the organization’s activities would be substantially hampered without access to Twitter. Furthermore, in Fields and similar cases, plaintiffs failed to demonstrate that ISIS’s use of Twitter played an instrumental role in the attacks that victimized them. The District Court for the Northern District of California noted the following in Fields:

The allegations … do not support a plausible inference of proximate causation between Twitter’s provision of accounts to ISIS and the deaths of Fields and Creach. Plaintiffs allege no connection between the shooter, Abu Zaid, and Twitter. There are no facts indicating that Abu Zaid’s attack was in any way impacted, helped by, or the result of ISIS’s presence on the social network.86

Given the plaintiffs’ failure to establish ISIS’s Twitter use as the proximate cause of their harms, the Ninth Circuit rejected Fields’ appeal.

More broadly, any standard of liability that might implicate Twitter in terrorist attacks would also capture transport providers, restaurateurs, and cellular networks. All these services are frequently used by terrorists, though they cannot be seen as uniquely instrumental in the realization of terrorist plots.

A Twitter account is not an unalloyed boon to terrorists. A public social media presence provides opportunities for counterspeech and intelligence gathering. In some cases, state security services have asked social media platforms to refrain from removing terrorist accounts, as they provide valuable information concerning the aims, priorities, and sometimes the locations of terrorist actors.87

As Posner notes, social media platforms have policies against terrorist speech. For example, YouTube’s policy on “terrorist content” states:

We do not permit terrorist organizations to use YouTube for any purpose, including recruitment. YouTube also strictly prohibits content related to terrorism, such as content that promotes terrorist acts, incites violence, or celebrates terrorist attacks.88

Facebook and Twitter have similar policies, though they attempt to limit the subjectivity of terrorism by tying it to violence against civilians.89 Twitter’s policy states:

You may not make specific threats of violence… . This includes, but is not limited to, threatening or promoting terrorism. You also may not affiliate with organizations that—whether by their own statements or activity both on and off the platform—use or promote violence against civilians to further their causes.90

Social media moderation may be more effective than the increases in government power desired by Posner. But that effectiveness may have been acquired by narrowing the kinds of speech heard on those platforms. However one assesses that narrowing, the case for more government power here remains at best unproven.

Electoral Integrity. Many believe that protecting national security also means preventing foreign powers from influencing American elections. In February 2018, Robert S. Mueller III, a special counsel to the U.S. Department of Justice, indicted 13 Russians for intervening in the 2016 U.S. election. The indictment charges that they intervened by purchasing advertising that mostly “focused on issues—like civil rights or immigration—and did not promote specific candidates.”91 In other words, the ads were speech about issues discussed during the campaign. Of course, this was speech by foreign agents, which presumably makes all the difference. But should it? The First Amendment does not refer to speakers but rather to speech, which is protected from government abridgment. It might be assumed that foreign agents seek to do harm to the United States through speech. Indeed, Mueller’s indictment and the law underlying it purport to protect national security from foreign actions during the election.92 But that harm could be ignored or rejected by internet users who are expected under the Constitution to act as a censor of dangerous speech. Lastly, social media users may have a right to receive materials from foreign speakers. Such a right would belong to a reader or listener rather than the speaker; the Russian speakers in this case have no right. In 1965, the Supreme Court invalidated a law requiring readers to sign at the Post Office to receive communist publications. The act of signing chilled a presumed right to receive a publication from abroad.93 In a concurring opinion, Justice William Brennan remarked:

It is true that the First Amendment contains no specific guarantee of access to publications. However, the protection of the Bill of Rights goes beyond the specific guarantees to protect from congressional abridgment those equally fundamental personal rights necessary to make the express guarantees fully meaningful … I think the right to receive publications is such a fundamental right. The dissemination of ideas can accomplish nothing if otherwise willing addressees are not free to receive and consider them. It would be a barren marketplace of ideas that had only sellers and no buyers.94

The Peking Daily was printed on paper; Russian speech appeared online. But the difference does not merit protecting a reader’s right to see the former and not the latter.

In this case, national security seems to have outweighed freedom of speech. But that conclusion is somewhat misleading. In fact, the United States both censors some foreign speech and permits other speech with disclosure of the source. We turn first to the case for censorship.

Why were the Russian efforts indictable? The indicted were supported by the Russian government and are said to have bought ads advocating the election or defeat of candidates for federal office. The relevant law comes from Title 11, Section 110.20(b) of the Code of Federal Regulations:

A foreign national shall not, directly or indirectly, make any expenditure, independent expenditure, or disbursement in connection with any Federal, State, or local election.95

The Federal Election Commission (FEC) concisely explicates the relevant law:

Foreign nationals are prohibited from the following activities: Making any contribution or donation of money or other thing of value, or making any expenditure, independent expenditure, or disbursement in connection with any federal, state or local election in the United States; … [and] making any disbursement for an electioneering communication.96

The ads could have been a “thing of value,” an “independent expenditure,” or “disbursement for an electioneering communication.”

Yet the law may not proscribe all Russian speech concerning American elections. The FEC notes that a federal district court has said that the “foreign national ban ‘does not restrain foreign nationals from speaking out about issues or spending money to advocate their views about issues. It restrains them only from a certain form of expressive activity closely tied to the voting process—providing money for a candidate or political party or spending money in order to expressly advocate for or against the election of a candidate.’”97

Here we discern a descent toward incoherence. On the one hand, a foreign national is prohibited from spending money on American elections, including advertising. On the other hand, that ban extends only to spending on express advocacy for or against a candidate. As noted earlier, most of the spending by the Russians in 2016 involved issue advocacy and not express advocacy. What about internet speech by Russian agents? They were presumably paid to speak, so the same distinction applies; speech about the issues would be permitted. Of course, speech about issues by a foreign national “volunteer” would not be indictable.

The national security interest at stake appears to be the integrity of a fund­amental institution—elections. A foreign national spending money on speech that advocates the election or defeat of a candidate apparently threatens the integrity of elections. A foreign national discussing the issues debated during an election does not pose the same threat. This distinction may belie an assumption that using money to support speech would enable a foreign power to coordinate direct influence over voters and thereby affect the outcome of an election. Observations by random foreign nationals would not likely be effective.

But what’s entailed in that assumption is that without censorship voters and their votes would be affected by the foreign campaign to the detriment of national security. The assumption is paternalistic and contravenes many of the justifications for freedom of speech found in Supreme Court decisions about freedom of speech. At the same time, voters are assumed to be capable of dealing with issue advocacy by foreign nationals. So there is a tension here that reflects badly yet well on the United States.

Speech by foreign nationals is not just a threat to national security. If it were only a threat, that threat would be countered by banning all foreign speech. But speech by foreign nationals also offers benefits to Americans, so banning all foreign speech would involve significant costs. For this reason, foreign speech is often regulated but not prohibited.

The FEC notes that the ban on spending on ads by foreign nationals “was first enacted in 1966 as part of the amendments to the Foreign Agents Registration Act (FARA), an ‘internal security’ statute. The goal of the FARA was to minimize foreign intervention in U.S. elections by establishing a series of limitations on foreign nationals.”98 FARA prohibited some speech, but it also permitted speech by foreigners under certain conditions. FARA required agents of foreign powers to register with the federal government; in short, people who are paid by a foreign government must disclose that relationship.

This more liberal approach to foreign speech may be seen in FARA’s statement of purpose:

To protect the national defense, internal security, and foreign relations of the United States by requiring public disclosure by persons engaging in propaganda activities and other activities for or on behalf of foreign governments, foreign political parties, and other foreign principals so that the Government and the people of the United States may be informed of the identity of such persons and may appraise their statements and actions in the light of their associations and activities.99

The law also requires “that informational materials (formerly propaganda) be labeled with a conspicuous statement that the information is disseminated by the agents on behalf of the foreign principal.” The Department of Justice says disclosure is required of people “attempting to influence U.S. public opinion, policy, and laws.”100

Policy and law are likely the most important contexts for the speech of foreign agents. Foreign governments, acting on behalf of their citizens, need not represent only the interests of foreigners. For example, an exporting nation might wish to make the case against American protectionism. Note that such advocacy might also favor consumers in the United States. Such speech hardly threatens U.S. national security; indeed, it may even serve the general welfare of the nation.101

In other cases, the interests of governments and peoples diverge, and the speech of foreign agents may run counter to the interests of the American people. Even though this speech could be divergent and a potential security threat, it does not require censorship. Diplomacy requires people employed by foreign powers to speak with policy­makers on behalf of their governments. Public officials, including members of Congress and the executive branch, often meet and hear the arguments of foreign agents. Apparently, registering and thereby disclosing such agents sufficiently protects American security in those situations.

Censorship is also apparently not necessary to protect public opinion. The TV channel RT, which is funded by the Russian government, has been required to register as a foreign agent.102 RT offers general news and information to a small number of viewers.103 Presumably RT seeks mostly to influence public opinion, though it might thereby affect policy or law. The content of the speech on RT might be similar to or even the same as an advertisement purchased or speech otherwise uttered by a foreign agent. Even though RT is funded by the Russian government, it was required to register as a foreign agent rather than go silent. Apparently, voters can sort out the propaganda on a television network funded by the Russian government but not the advertising paid for by it.104

In sum, American law permits some speech by foreign nationals during an election. The law may permit issue advocacy by foreign nationals. It does not permit foreign nationals to spend money directly on elections, especially by buying advertising that supports or opposes a candidate.

There is little evidence that the Russian efforts had much effect on the American voters in 2016. Reporting by the New York Times suggests that Russian efforts may have persuaded a few people to show up at a small anti-Muslim rally in Texas.105 Speculation about other effects abounds. But as Brendan Nyhan, a professor of public policy at the University of Michigan, indicates, political science research shows how hard it is to change votes even with significant spending.106 The Russian effort was a minuscule portion of overall spending in 2016.107

Moreover, as Ross Douthat notes, much of the Russian effort “did not introduce anything to the American system that isn’t already present; it just reproduced, often in lousy or ludicrous counterfeits, the arguments and images and rhetorical tropes that we already hurl at one another every day.”108 At the margins, the Russians added “divisive” speech, but the increment was relatively trivial. And divisive speech is not illegal for Americans. In this case, however, the Russian money made the speech illegal.

Federal law seems needlessly incoherent. Allowing foreign nationals to buy ads with disclosure of their participation would vindicate freedom of speech. It might be objected that allowing such spending would permit a hostile foreign power to fund and coordinate a propaganda campaign capable of affecting the outcome of an American election. But allowing foreign nationals to fund lobbying efforts has not subjugated policymaking to foreign interests. Policymakers are assumed to be capable of sorting out arguments and interests. Perhaps voters are not as capable of doing so, although the unpopularity of RT suggests otherwise.

The Mueller indictment may never move to trial because the indicted are unlikely to come to the United States. But even if you believe Americans should be protected from Russian ads, there is little need for federal action.

A Private Alternative

The social media company most affected by the Russian efforts is regulating itself. Mark Zuckerberg, the founder and CEO of Facebook, lists “defending against election interference” as one of the three most important issues facing his company.109 According to Zuckerberg, foreign nations set up fake accounts to enable “coordinated information operations … spreading division and misinformation.” Facebook is using machine learning to identify and remove such accounts. Zuckerberg argues that the accounts are being removed not because of the content of their speech but because they violate Facebook’s Community Standards, which require that an account have an authentic user. He notes that this policy involves both false positives (some accounts with authentic users are taken down) and false negatives (some fake accounts stay up).

Facebook is also enforcing disclosure on ad buyers:

Facebook now has a higher standard of ads transparency than has ever existed with TV or newspaper ads. You can see all the ads an advertiser is running, even if they weren’t shown to you. In addition, all political and issue ads in the US must make clear who paid for them. And all these ads are put into a public archive which anyone can search to see how much was spent on an individual ad and the audience it reached.

The disclosure aims specifically at excluding expenditures by foreign nationals:

We now also require anyone running political or issue ads in the US to verify their identity and location. This prevents someone in Russia, for example, from buying political ads in the United States, and it adds another obstacle for people trying to hide their identity or location using fake accounts.

Facebook appears to be offering a private solution to the perceived threat to the integrity of elections and hence national security. No doubt some Russian accounts will escape the ban on fake accounts. But in this regard Facebook seems much better placed than the federal government to regulate Russian efforts. The private sector is doing what the public sector cannot and should not do.

What about freedom of speech? Facebook’s efforts do not suppress speech on the basis of its content. The rule against accounts that do not identify their owner or location does not implicate the content of speech. Facebook requires disclosure of the funders of all political ads, including ads about issues only. In contrast, federal election law exempts some groups and individuals from disclosing funding for issue ads.110 Facebook’s disclosure rules are thus less liberal than federal election law.

This broad sweep of disclosure may be a response to the content of speech. Zuckerberg writes:

Most of the divisive ads the Internet Research Agency ran in 2016 focused on issues—like civil rights or immigration—and did not promote specific candidates. To catch this behavior, we needed a broad definition of what constitutes an issue ad.

So the broad definition is a response to what? The divisiveness of the speech? Or the source of the speech? To the extent the rule seeks the source, it is roughly similar to federal law governing prohibited sources of funding. But if divisiveness leads to disclosure, Facebook is regulating, though not suppressing, speech based on its content.

But are Facebook’s efforts truly a private decision—thus exempt from the strictures of the First Amendment—or the result of political pressure? Congress has been concerned about Russian internet efforts during the 2016 election. These concerns have led some members to threaten to impose regulation on tech companies.111 Mark Zuckerberg has also emphasized that Facebook is working closely with governments to prepare for elections.112 Whether Facebook’s regulation of speech on its platform is a public-private undertaking and whether such undertakings should be constrained by the First Amendment remain open questions.

Preventing Harms Caused by Speech

Social media comprise speech and little else. For that reason, as noted earlier, social media are largely immune from government regulation; they benefit from the priority given to private judgment in these matters. Despite this protection, it might still be considered valid for the government to manage speech on social media if such regulations were narrowly tailored to achieve a compelling government interest; in other words, the regulation would pass muster under the “strict scrutiny” test the courts apply to restrictions on fundamental rights.113 Here I examine two potential compelling government interests rooted in widely held public values: preventing the harms caused by “fake news” and “hate speech.”114

The cases for regulation of both kinds of speech have a common weakness. If we do not know what a term means, we cannot know how it applies. Thus, vagueness fosters unconstitutionality, as Nadine Strossen explains:

The Supreme Court has held that any law is “unduly vague,” and hence unconstitutional, when people “of common intelligence must necessarily guess at its meaning.” This violates tenets of “due process” or fairness, as well as equality, because such a law is inherently susceptible to arbitrary and discriminatory enforcement. Moreover, when an unduly vague law regulates speech in particular, the law also violates the First Amendment because it inevitably deters people from engaging in constitutionally protected speech for fear that they might run afoul of the law. The Supreme Court has therefore enforced the “void for vagueness” doctrine with special strictness in the context of laws that regulate speech.115

Looked at another way, vagueness would lead government to suppress both prohibited and permitted speech. The “false positives,” which are permitted speech wrongly suppressed, would involve a cost to freedom of speech. Given the importance attached to free speech in the United States, it is unlikely the benefits of suppressing speech would outweigh the costs of those false positives. Such costs would also indicate the chosen means were poorly tailored to the ends sought by the government; vagueness would suggest the government regulation of speech could not pass a strict scrutiny test.

For these reasons, the following analysis pays close attention to the meanings of fake news and hate speech. So far as we find the terms vague, we should have less confidence in calls to suppress such speech, however “fake” and however “hateful.”

What is fake news? The term has been used to refer to “satirical news, hoaxes, news that’s clumsily framed or outright wrong, propa­ganda, lies destined for viral clicks and advertising dollars, politically motivated half-truths, and more.”116 The term fake news has come to public attention relatively recently. The relevant linguistic community might be working toward a clear definition. A recent European Commission Working Paper examines several definitions of fake news.117 Some common elements of these various definitions might suggest a clear definition of the term.

  • “False, often sensational, information disseminated under the guise of news reporting” (Collins Dictionary).
  • “(1) News that is made up or ‘invented’ to make money or discredit others; (2) news that has a basis in fact, but is ‘spun’ to suit a particular agenda; and (3) news that people don’t feel comfortable about or don’t agree with” (Reuters Institute, “Digital News Report 2017”).
  • “All forms of false, inaccurate, or misleading information designed, presented and promoted to intentionally cause public harm or for profit” (European Commission, A Multidimensional Approach to Disinformation).
  • “Verifiably false or misleading information that is created, presented and disseminated for economic gain or to intentionally deceive the public, and in any event to cause public harm” (European Commission, “Tackling Online Disinformation: Commission Proposes An EU-wide Code of Practice”).
  • “Perceived and deliberate distortions of news with the intention to affect the political landscape and to exacerbate divisions in society” (European Commission, “Joint Research Centre Digital Economy Working Paper 2018-02”).

Some elements of these definitions clearly could not pass muster under American constitutional law. Speech that fits a particular agenda, that makes people uncomfortable, or that affects the political landscape would all be protected by American courts. Apart from that, the term fake news appears to comprise three elements: intentionality, falsity, and a public harm. Each of these elements poses serious problems to the First Amendment.

Apparently, only those who deliberately seek to mislead or divide listeners are liable for false or harmful speech, according to those who seek to regulate fake news.

But this intentionality standard itself does not work well with the remaining speech. If the government may not suppress false speech or speech that causes a public harm, then whether the speech is intended to cause a public harm does not matter. So we turn first to whether false and harmful speech may be sanctioned by the U.S. government.

The falsity of speech refers to its content. Generally, governments in the United States may not prohibit or sanction speech because of its content.118 However, some exceptions exist to this general rule: incitement, obscenity, defamation, speech integral to crimes, child pornography, fraud, true threats, “fighting words,” and “speech presenting some grave and imminent threat the government has the power to prevent.”119 These exceptions, the Supreme Court says, have a historical foundation in the court’s free speech tradition.120

In United States v. Alvarez, the court refused to recognize a general exception to the First Amendment for false speech: “The Court has never endorsed the categorical rule the Government advances: that false statements receive no First Amendment protection.”121 In part, false speech was not a traditionally recognized exception. Also, giving the government the power to limit speech on behalf of truth would chill permitted speech, thereby calling into question “a foundation of our freedom.”122 False speech about politics enjoys significant protection under the Constitution.

Courts have long recognized defamation as a general exception to the freedom of speech.123 The government may sanction speech integral to defamation. Individuals defamed by others on social media may seek relief for a tort; the state then enforces the sanction on libelous speech. State sanctions require both harm to reputation and falsity; the exception is not for false speech per se.124 Public figures must also show “actual malice” by defendants to recover compensation.125 Falsity alone is not enough to allow punishment for speech. The standard of actual malice is quite demanding on those seeking relief. Moreover, Section 230 of the Communications Decency Act of 1996 prevents social media platforms from being liable for the torts of their users.126 Defamation will provide a limited public response to fake news.127 It is not broad enough to underpin a substantial government effort to regulate false speech on social media.

What other public harms are said to be caused by fake news? The public believes fake news causes “a great deal of confusion about the basic facts of current issues and events.”128 Such confusion might cause a larger problem. Many consumers now view news online. This change “has lowered costs and expanded market reach for news producers and consumers.” But the shift has also separated news producers (editors) and distributors (curators). Distribution is now being performed by algorithmic advertising and distribution platforms such as search engines, news aggregators, and social media sites. It is argued that editors cared about their reputation for quality news, but now the new distributors seek maximum traffic and advertising revenue instead. A European Commission study group suggests that these developments

may weaken consumer trust and news brand recognition and facilitate the introduction of disinformation and false news into the market. This may contribute to news market failure when it becomes difficult for consumers to distinguish between good quality news and disinformation or fake news.129

News consumers might consume less news to avoid false or misleading information.130 The outcome might be an ill-informed electorate less willing to grant legitimacy to the government.

This market failure argument for regulating fake news lacks empirical support. Scholarly literature notes that social media have offered both costs and benefits to consumers. However, “the impact of all these changes on the welfare of consumers, producers and society as a whole is not so clear. There is very little empirical evidence to date, also because relevant data are often proprietary and not accessible to independent researchers” (emphasis added).131 In light of this, we have little reason to believe that by regulating fake news the government will serve the important interests said to be at stake.

Private content moderators permit false speech. However, they manage such speech much more efficiently than the government. Facebook says:

It’s important to note that whether or not a Facebook post is accurate is not itself a reason to block it. Human rights law extends the same right to expression to those who wish to claim that the world is flat as to those who state that it is round—and so does Facebook.132

Although Facebook does not block false speech, it does make certain categories of false speech more difficult to find and points users toward other presumably more accurate articles about a topic.133 Moreover, sites posting false speech often violate other Facebook rules (e.g., rules against spam, hate speech, or fake accounts) and are suppressed.134 Yet Facebook product manager Tessa Lyons says “we don’t want to be and are not the arbiters of truth.”135 Yet Facebook has delegated the task of determining the factual truth of contested content to a network of third-party fact-checkers.136 While this allows Facebook to avoid the difficult and politically fraught work of distinguishing fact from fiction, Facebook is still held responsible for its selection of fact-checkers and the impact of their decisions. In September, the Weekly Standard, the sole conservative organization admitted to Facebook’s fact-checking program, deemed a ThinkProgress article, or at least its headline, false, limiting its distribution. ThinkProgress took umbrage with the decision and criticized Facebook for granting a conservative publication the ability to downrank its content.137

Facebook appears to want to let a thousand flowers bloom on its platform, yet it employs fact-checking gardeners that cut the false ones. The public values truth, and we hope that conspiracy theories and obvious falsehoods are bad for business. On the other hand, a tech company deciding between the competing truths offered by blue and red speakers invites political attacks against their platform and, over the long-term, sows doubt about the fairness of its content-moderation policies. Tech companies may sanction speech in circumstances where government must remain passive. Yet that empowerment has its own problems, not least of which is deciding between contending armies in an age of cultural wars.

Many nations have undertaken regulation of fake news recently.138 That such illiberal countries as Belarus, China, Cameroon, or Russia (among others) would impose government restrictions on posting or spreading misinformation may not surprise anyone.139 But European nations are more open to actively regulating speech than the United States. In November 2018, France gave authorities the power to “remove fake content spread via social media and even block the sites that publish it.”140 The European Commission has issued an initial report on disinformation that will be followed by a process of oversight and evaluation of online speech.141 For now, the commission is supporting principles and policies that would be enacted by stakeholders including the news media and online companies.142 Does such nudging of private actors constitute political pressure to suppress speech? If disinformation and fake news remain a problem, would the commission directly manage online speech or encourage national governments to take stronger measures to suppress such speech?

The United States regulates speech less than Europe does. Perhaps the European examples about regulating disinformation are not relevant for this nation.143 Yet the debate over fake news has lasted only a couple of years. Little has been said during that debate about the limits of government power over online speech; much has been said about the dangers to democracy of permitting fake news. Should future national elections turn out badly, the United States might be tempted to take a more European attitude and approach to online speech.

We should thus keep in mind that the case for public as opposed to private regulation of fake news online is weak. Fake news has no fixed meaning, and regulations would be unconstitutionally vague. The public values truth, but the search for truth in the United States must abide by the First Amendment, and the courts have held that false speech—the whole of which fake news is a part—also has the protection of the First Amendment. Were this not true, the combination of vagueness and politics in a polarized age would mean virtually anything “the other side” said would be regulated as fake news. But fake news might not be the most likely reason for suppressing online speech.

Hate speech may be defined as “offensive words, about or directed toward historically victimized groups.”144 That definition seems clear enough. But consider the case of The Bell Curve, a 1994 book by Charles Murray and Richard Herrnstein. Among other things, the authors state that the average IQ score of African Americans is one standard deviation below the average score of the population. Many also thought the book argued that nature was far more important than nurture in determining the IQ of individuals and groups, a claim that suggested social reforms would have little effect on individual and group outcomes.145 The Bell Curve offended many people; “historically victimized groups” might well have taken offense. Was The Bell Curve hate speech? If not, where should elected officials draw the line between permitted and prohibited speech?

The Supreme Court has resisted drawing such lines. Even efforts to legislate more common-sense bans on group invective have failed; the court has consistently invalidated laws containing terms such as “contemptuous,” “insulting,” “abusive,” and “outrageous.”146 The U.S. government lacks the power to prohibit “hate speech.”

Yet many nations regulate or prohibit speech offensive to protected groups. They limit freedom of speech to advance other values such as equal dignity. This balancing of values was first developed in Germany and has spread to other jurisdictions in the post-World War II era.147 In Germany, the law punishes “incitement of the people,” which is understood as spurring hatred of protected groups, demanding violent or arbitrary measures against them, or attacking their human dignity. Those convicted of incitement may be jailed for up to five years.148 The United Kingdom also criminalizes the expression of racial hatred.149 In two recent cases, a hate speech conviction led to incarceration.150

The United States has debated regulating hate speech for nearly a century.151 Legal scholar James Weinstein summarizes the outcome of this debate: “The United States is an outlier in the strong protection afforded some of the most noxious forms of extreme speech imaginable.”152 The Supreme Court precludes government from regulating speech because of the message of content-based regulation it conveys. For the court, the worst content-based regulation is “viewpoint discrimination,” which is restrictions based on “the specific motivating ideology or the opinion or perspective of the speaker.”153 This constraint on political power extends to highly offensive speech, which implies, Weinstein remarks, “a complete suspension of civility norms within the realm of public discourse.”154 Government may regulate some speech that is outside public discourse: all unprotected speech involves government activities or commercial advertising.155

The Supreme Court has applied this general framework to protect speech hostile to racial minorities. In their decision in R.A.V. v. City of St. Paul, the court dealt with a Minneapolis ordinance punishing speech that “one knows or has reasonable grounds to know arouses anger, alarm or resentment in others on the basis of race, color, creed, religion or gender.”156 A lower court ruled that the ordinance reached protected as well as unprotected speech and thus was unconstitutionally overbroad. The same court interpreted the ordinance to apply only to “fighting words,” which have been considered outside the protections of the First Amendment. Most of the Supreme Court went further, holding that Minneapolis had engaged in viewpoint discrimination by punishing some but not all “fighting words,” a distinction based on the ideological content of some speech.

In theory, it is possible for the courts to uphold viewpoint discrimination. Such distinctions must pass the strict scrutiny test discussed earlier. To do so, the Minneapolis regulation would have needed to be narrowly drawn to achieve a compelling government interest. The court recognized the importance of protecting minorities. Yet the government had other means to achieve that end, means that were neutral toward the content of the speech.157 Most experts assume R.A.V. v. City of St. Paul precludes government suppression of hate speech. Accordingly, hate speech on social media lies beyond government power.158

In contrast to the government, social media managers may regulate speech by users that is hostile to some groups. Facebook does so extensively. Facebook defines hate speech as “anything that directly attacks people based on what are known as their ‘protected characteristics’—race, ethnicity, national origin, religious affiliation, sexual orientation, sex, gender, gender identity, or serious disability or disease.”159 Facebook is opposed “to hate speech in all its forms”; it is not allowed on their platform as a matter of policy.160 Hate speech is forbidden on Facebook because it causes harm by creating “an environment of intimidation and exclusion and in some cases may have dangerous offline implications.”161 In June 2017, Richard Allan, vice president for public policy at Facebook, said: “Over the last two months, on average, we deleted around 66,000 posts reported as hate speech per week—that’s around 288,000 posts a month globally.”162 However, at that time, Facebook had over two billion active users.163 The number of removed hate speech posts, though very large, is relatively trivial.

Other major platforms have policies that protect people with a similar list of characteristics from hostile speech. Google has a general policy against “incitement to hatred” of a list of groups.164 YouTube, which is owned by Google, does not permit hate speech directed toward seven groups.165 This policy led to videos by Alex Jones being removed.166 Twitter also has a similar policy against hate speech.167

In sum, the First Amendment does not permit government to censor speech to prevent harms to the public apart from known exceptions such as direct incitement to violence. The government may not censor fake news or hate speech. Private regulators are doing what government officials may not do: regulating and suppressing speech believed to cause harm to citizens generally and protected groups specifically. Private action thus weakens the case for moving the United States toward a more European approach to fake news and hate speech.

But such private action presents a mixed picture for supporters of robust protections for speech. The platforms offer less protection for speech than the government does. Social media managers discriminate among speakers according to the content of their speech and the viewpoints expressed. Tech companies have in part applied to speech the proportionality test long-recognized in Europe and rejected in this country. Private content governance of social media poses a quandary, particularly for libertarians and anyone who recognizes that private property implies a strong right for social media managers to control what happens on their internet platforms without government interference. It seems likely that social media managers choose to limit speech in the short term to fulfill their larger goal of building a business for the long term. They may believe that excluding extreme speech is required to sustain and increase the number of users on their platform.

Moreover, we should ask whether these efforts regarding hate speech (along with private suppression of Russian speech, terrorist incitement, or fake news) is truly a private decision and not state action. If Facebook or other platforms remove content to avoid government regulation, is such suppression state action or a hybrid of private choice determined by public threats and offers?

Conclusion

We began with Cloudflare CEO Matthew Prince’s concern about legitimate governance of speech on the internet. Prince’s desire to bring government into online speech controversies is understandable but misplaced. American history and political culture assign priority to the private in governing speech online and particularly on social media. The arguments advanced for a greater scope of government power do not stand up. Granting such power would gravely threaten free speech and the independence of the private sector. We have seen that these tech companies are grappling with many of the problems cited by those calling for public action. The companies are technically sophisticated and thus far more capable of dealing with these issues. Of course, the efforts of the companies may warrant scrutiny and criticisms, now and in the future. But at the moment, a reasonable person can see promise in their efforts, particularly in contrast to the likely dangers posed by government regulation.

Government officials may attempt directly or obliquely to compel tech companies to suppress disfavored speech. The victims of such public-private censorship would have little recourse apart from political struggle. The tech companies, which rank among America’s most innovative and valuable firms, would then be drawn into the swamp of a polarized and polarizing politics. To avoid politicizing tech, it is vital that private content moderators be able to ignore explicit or implicit threats to their independence from government officials.

It is Facebook, Medium, and Pinterest—not Congress or President Trump—that have a presumption of legitimacy to remove the speech of StormFront and similar websites. These firms need to nurture their legitimacy to moderate content. The companies may have to fend off government officials eager to suppress speech in the name of the “public good.” The leaders of these businesses may regret being called to meet this challenge with all its political and social dangers and complexities. But this task cannot be avoided. No one else can or should do the job.

Notes

1. Remarks at presentation at the Cato Institute, November 28, 2017. See also Matthew Prince, “Why We Terminated Daily Stormer,” Cloudflare (blog), August 16, 2017. “Law enforcement, legislators, and courts have the political legitimacy and predictability to make decisions on what content should be restricted. Companies should not.” Thanks to Alissa Starzak for the reference.

2. Social media firms are often obligated to follow laws in nations where they operate. In the future, such laws may create enforceable transnational obligations. For now, however, the debate concerns national audiences and policies. See David R. Johnson and David G. Post, “Law and Borders: The Rise of Law in Cyberspace,” Stanford Law Review 48, no. 5 (1996): 1367.

3. Tom Standage, Writing on the Wall: Social Media—The First 2,000 Years (New York: Bloomsbury, 2013), p. 8. See also Sheryl Sandberg’s definition in testimony before the Senate Intelligence Committee: “Social media enables you to share what you want to share, when you want to share it, without asking permission from anyone, and that’s how we meet our mission, which is giving people a voice.” Sheryl Sandberg, Facebook chief operating officer, and Jack Dorsey, Twitter chief executive officer, Foreign Influence Operations’ Use of Social Media Platforms, Testimony before the Senate Committee on Intelligence, 115th Cong., 2nd sess., September 5, 2018.

4. Standage, p. 13.

5. J. A. Obar and S. Wildman, “Social Media Definition and the Governance Challenge: An Introduction to the Special Issue,” Telecommunications Policy 39, no. 9 (2015): 746.

6. “The backbone of the social media service is the user profile… . The type of identifying information requested, as well as the options for identifying oneself vary considerably from service to service, but often include the option of creating a username, providing contact information and uploading a picture. The reason the profile serves this backbone function is to enable social network connections between user accounts. Without identifying information, finding and connecting to others would be a challenge.” Obar and Wildman, 747.

7. Commercial speech plays a small part in these policy matters. Advertising, as will be shown, does matter. However, the speech carried by ads is political and not commercial. See the discussion of Russian “meddling” in the 2016 U.S. election.

8. Red Lion Broadcasting Co. v. Federal Communications Commission, 395 U.S. 367 (1969).

9. See the website for the Heritage Guide to the Constitution for a concise discussion. Eugene Volokh, “Freedom of Speech and of the Press,” Heritage Guide to the Constitution (website).

10. Sorrell v. IMS Health Inc., 564 U.S. 552 (2011); Martin H. Redish, “Commercial Speech and the Values of Free Expression,” Cato Institute Policy Analysis no. 813, June 19, 2017.

11. United States v. Carolene Products Co., 304 U.S. 144 (1938) 152: “Regulatory legislation affecting ordinary commercial transactions is not to be pronounced unconstitutional unless, in the light of the facts made known or generally assumed, it is of such a character as to preclude the assumption that it rests upon some rational basis within the knowledge and experience of the legislators.”

12. Other reasons might counsel not regulating speech on social media. The costs of such regulation might outweigh the benefits to society. Here I examine only rights that might preclude or weigh heavily against regulation.

13. See the extended discussion in Daphne Keller, “Who Do You Sue? State and Platform Hybrid Power over Online Speech,” Aegis Series Paper no. 1902, Hoover Institution, Stanford, 2019, pp. 17-21.

14. Miami Herald Pub. Co. v. Tornillo, 418 U.S. 241 (1974), 258.

15. Monica Bickert, head of global policy management at Facebook, writes: “First, [social media] generally do not create or choose the content shared on their platform; instead, they provide the virtual space for others to speak.” Lee Bollinger and Geoffrey Stone, eds., “Defining the Boundaries of Free Speech on Social Media,” The Free Speech Century (New York: Oxford University Press, 2018), p. 254. The important word here is “generally.” Relatively speaking, very little content is removed.

16. Eugene Volokh and Donald M. Falk, “Google: First Amendment Protection for Search Engine Search Results,” Journal of Law, Economics, and Policy 8, no. 8.4 (2012): 886-88.

17. See, for example, their CEO’s discussion of content moderation, which takes a stand “against polarization and extremism” while affirming other commitments. Mark Zuckerberg, “A Blueprint for Content Governance and Enforcement,” Facebook, November 15, 2018.

18. Protection for Private Blocking and Screening of Offensive Material, 47 U.S.C. § 230.

19. Batzel v. Smith, 333 F.3d 1018 (Court of Appeals, 9th Circuit 2003), 8443; and Electronic Frontier Foundation (website), “CDA 230: Legislative History.”

20. Batzel v. Smith, 8445.

21. Batzel v. Smith, 8445.

22. Bickert, “Defining,” 254-55.

23. Perry Education Association v. Perry Local Educators’ Association, 460 U.S. 37 (1983).

24. The courts are first among equals in the United States on these matters. “The First Amendment, as interpreted by the courts, provides an anchor for freedom of the press and thus accentuates the difference between publishing and the electronic domain. Because of the unique power of the American courts, the issue in the United States unfolds largely in judicial decisions.” Ithiel de Sola Pool, Technologies of Freedom (Cambridge: Harvard University Press, 1983), p. 8.

25. Marsh v. Alabama, 326 U.S. 501 (1946), 506, 509.

26. Food Employees v. Logan Valley Plaza Inc., 391 U.S. 308 (1968), 324.

27. Lloyd Corp. v. Tanner, 407 U.S. 551 (1972).

28. Hudgens v. NLRB, 424 U.S. 507 (1976), 513.

29. Pruneyard Shopping Ctr. v. Robins, 447 U.S. 74 (1980).

30. Pruneyard Shopping Ctr. v. Robins, 81-82.

31. Dahlia Lithwick, “Why Can Shopping Malls Limit Free Speech?,” Slate, March 10, 2003.

32. Communications Act of 1934, 47 U.S.C. § 151, Pub. L. No. 73-416.

33. Thomas Winslow Hazlett, The Political Spectrum: The Tumultuous Liberation of Wireless Technology, from Herbert Hoover to the Smartphone (New Haven: Yale University Press, 2017), p. 146.

34. The monopoly argument is not limited to one part of the political spectrum. The head of CNN called for investigation of “these monopolies that are Google and Facebook.” See Stewart Clarke, “CNN Boss Jeff Zucker Calls on Regulators to Probe Google, Facebook,” Variety, February 26, 2018; Tim Wu, The Curse of Bigness: Antitrust in the New Gilded Age (New Yorker: Columbia Global Reports, 2018).

35. Trump’s campaign manager Brad Parscale has written, “Google claims to value free expression and a free and open internet, but there is overwhelming evidence that the Big Tech giant wants the internet to be free and open only to political and social ideas of which it approves.” Brad Parscale, “Trump Is Right: More than Facebook and Twitter, Google Threatens Democracy, Online Freedom,” USA Today, September 10, 2018. Parscale’s article offers several examples of putative bias against conservatives. During the Republican primaries in 2016, Facebook employees admitted they “routinely suppressed news stories of interest to conservative readers from the social network’s influential ‘trending’ news section.” Facebook denied the charge. See Michael Nunez, “Former Facebook Workers: We Routinely Suppressed Conservative News,” Gizmodo, May 9, 2016; see also Peter van Buren, “Extend the First Amendment to Twitter and Google Now,” The American Conservative, November 7, 2017. This view appears to be spreading on the right; see Michael M. Grynbaum and John Herrman, “New Foils for the Right: Google and Facebook,” New York Times, March 6, 2018.

36. Laura Stevens, Tripp Mickle, and Jack Nicas, “Tech Giants Power to New Heights,” Wall Street Journal, February 2, 2018.

37. David S. Evans and Richard Schmalensee, “Debunking the ‘Network Effects’ Bogeyman: Policymakers Need to March to the Evidence, Not to Slogans,Regulation 40 (Winter 2017-18): 36.

38. Evans and Schmalensee, “Debunking,” 39.

39. “There are some important industries where ‘winner takes most’ may apply. But even there, victory is likely to be more transient than economists and pundits once thought. In social networking, Friendster lost to MySpace, which lost to Facebook, and, while Facebook seems entrenched, there are many other social networks nipping at its heels.” David S. Evans, Matchmakers: The New Economics of Multisided Platforms (Cambridge: Harvard Business Review Press, 2016), Kindle edition.

40. Facebook “was built on the power of network effects: You joined because everyone else was joining. But network effects can be just as powerful in driving people off a platform. Zuckerberg understands this viscerally.” Nicholas Thompson and Fred Vogelstein, “Inside the Two Years That Shook Facebook—and the World,” Wired, February 12, 2018.

41. Amy Mitchell, Elisa Shearer, Jeffrey Gottfried, and Michael Barthel, “How Americans Get Their News,” Pew Research Center, July 7, 2016.

42. Daniel Trotta, “Shunned by Corporations, U.S. Gun Entrepreneurs Launch Start-Ups,” Reuters, May 6, 2018.

43. Hazlett, The Political Spectrum, pp. 91-92.

44. Bruce M. Owen, Jack H. Beebe, and Willard G. Manning, Television Economics (Lexington, MA: Lexington, 1974), p. 12, quoted in Hazlett, The Political Spectrum, p. 92.

45. Hazlett, The Political Spectrum, pp. 20-21.

46. Hazlett, The Political Spectrum, pp. 14-16.

47. Hazlett, The Political Spectrum, pp. 143-45.

48. For the public failure see John Samples, “Broadcast Localism and the Lessons of the Fairness Doctrine,” Cato Institute Policy Analysis no. 639, May 27, 2009, pp. 7-8; for Hazlett, see The Political Spectrum, pp. 149-52.

49. “A surprising number of people it seems dislike being exposed to the processes endemic to democratic government. People profess a devotion to democracy in the abstract but have little or no appreciation for what a practicing democracy invariably brings with it… . People do not wish to see uncertainty, conflicting options, long debate, competing interests, confusion, bargaining, and compromised, imperfect solutions.” John R. Hibbing and Elizabeth Theiss-Morse, Congress as Public Enemy: Public Attitudes Toward American Political Institutions (Cambridge: Cambridge University Press, 1995), p. 147.

50. Cass R. Sunstein, #Republic: Divided Democracy in the Age of Social Media (Princeton: Princeton University Press, 2018), p. 157.

51. One way to deal with this conflict between “our” aspirations and the revealed preferences of individuals has been to insist that revealed preferences actually contravene the true interests of individuals, whereas our aspirations represent a truth to be honored by government action. Equating revealed preferences with false consciousness is one version of this argument. As we shall see, Sunstein does not go far down the path toward imposing our aspirations.

52. Sunstein, #Republic, p. 260.

53. Sunstein, #Republic, p. 43. He appeals to the spirit if not the letter of constitutional doctrine: “On the speakers’ side, the public forum doctrine thus creates a right of general access to heterogeneous citizens. On the listeners’ side, the public forum creates not exactly a right but rather an opportunity, if perhaps an unwelcome one: shared exposure to diverse speakers with diverse views and complaints.” Sunstein, #Republic, p. 38.

54. Sunstein, #Republic, p. 49.

55. Sunstein, #Republic, p. 71, 259. Former president Barack Obama has said that “essentially we now have entirely different realities that are being created with not just different opinions, but now different facts. And this isn’t just by the way Russian inspired bots and fake news. This is Fox News vs. The New York Times editorial page. If you look at these different sources of information, they do not describe the same thing. In some cases, they don’t even talk about the same thing. And so it is very difficult to figure out how democracy works over the long term in those circumstances.” He added that government should put “basic rules of the road in place that create level playing fields.” Robby Soave, “5 Things Barack Obama Said in His Weirdly Off-the-Record MIT Speech,” Hit and Run (blog), Reason, February 26, 2018.

56. Sunstein, #Republic, p. 255.

57. Sunstein, #Republic, p. 67.

58. Cristian Vaccari, “How Prevalent Are Filter Bubbles and Echo Chambers on Social Media? Not as Much as Conventional Wisdom Has It,” Cristian Vaccari (blog), February 13, 2018.

59. See the studies cited in Michael A. Beam, Myiah J. Hutchens, and Jay D. Hmielowski, “Facebook News and (de)Polarization: Reinforcing Spirals in the 2016 US Election,” Information, Communication and Society 21, no. 7 (July 3, 2018): 4.

60. Solomon Messing and Sean J. Westwood, “Selective Exposure in the Age of Social Media: Endorsements Trump Partisan Source Affiliation When Selecting News Online,” Communication Research 41, no. 8 (December 2014): 1042-63.

61. Pablo Barberá, “How Social Media Reduces Mass Political Polarization. Evidence from Germany, Spain, and the U.S.,” working paper, New York University, 2014. Paper prepared for the 2015 APSA Conference.

62. Frederik J. Zuiderveen Borgesius, Damian Trilling, Judith Möller, Balázs Bodó, Claes H. de Vreese, and Natali Helberger, “Should We Worry about Filter Bubbles?,” Internet Policy Review 5, no. 1 (2016): 10.

63. Matthew Barnidge, “Exposure to Political Disagreement in Social Media Versus Face-to-Face and Anonymous Online Settings,” Political Communication 34, no. 2 (2016): 302-21.

64. Levi Boxell, Matthew Gentzkow, and Jesse M. Shapiro, “The Internet, Political Polarization, and the 2016 Election,” Cato Institute Research Brief in Economic Policy no. 88, November 1, 2017.

65. Beam et al., “Facebook News and (de)Polarization,” 1.

66. Cristian Vaccari, “How Prevalent Are Filter Bubbles and Echo Chambers on Social Media? Not as Much as Conventional Wisdom Has It,” Cristian Vaccari (blog), February 13, 2018.

67. Elizabeth Dubois and Grant Blank, “The Echo Chamber Is Overstated: The Moderating Effect of Political Interest and Diverse Media,” Information, Communication and Society 21, no. 5 (2018): 729-45.

68. Richard Fletcher and Rasmus Kleis Nielsen, “Are People Incidentally Exposed to News on Social Media? A Comparative Analysis,” New Media and Society 20, no. 7 (July 2018): 2450-68.

69. Sunstein, #Republic, p. 262.

70. Sunstein, #Republic, p. 260; see also p. 158.

71. Owen Fiss, “Free Speech and Social Structure,” Iowa Law Review 71 (1986): 1405-25.

72. Owen Fiss, The Irony of Free Speech (Cambridge, MA: Harvard University Press, 1996).

73. Sunstein, #Republic, p. 260.

74. Sunstein has proposed “nudging” people to make better decisions by altering their choice architecture; see Richard H. Thaler and Cass R. Sunstein, Nudge: Improving Decisions about Health, Wealth, and Happiness (New Haven: Yale University Press, 2008).

75. Sunstein, #Republic, p. 215.

76. Sunstein, #Republic, p. 226. “I certainly do not suggest or believe that government should require anything of this kind (i.e., mandatory linking to opposing views). Some constitutional questions are hard, but this one is easy: any such requirements would violate the First Amendment.” Sunstein, p. 231.

77. Sunstein, #Republic, pp. 84-85, 221.

78. Amos A. Jordan et al., American National Security (Baltimore: Johns Hopkins University Press, 2011), pp. 3-4.

79. J. M. Berger, “The Difference between a Killer and a Terrorist,” The Atlantic, April 26, 2018.

80. Kathleen Ann Ruane, “The Advocacy of Terrorism on the Internet: Freedom of Speech Issues and the Material Support Statutes,” Congressional Research Service, September 8, 2016, p. 1.

81. Brandenburg v. Ohio, summarized in Ruane, “The Advocacy of Terrorism on the Internet,” p. 5.

82. Brandenburg v. Ohio, 395 U.S. (1969): 448; and Ruane, “The Advocacy of Terrorism on the Internet,” p. 4.

83. Eric Posner, “ISIS Gives Us No Choice but to Consider Limits on Speech,” Slate, December 15, 2015.

84. David G. Post, “Protecting the First Amendment in the Internet Age,” Washington Post, December 21, 2015.

85. See Pennie v. Twitter Inc., 2017 WL 5992143 (N.D. Cal. Dec. 4, 2017); Force v. Facebook Inc., 2018 WL 472807 (E.D.N.Y. Jan. 18, 2018); Crosby v. Twitter Inc., 303 F. Supp. 3d 564 (E.D. Mich. April 2, 2018); Gonzalez v. Google Inc., 2018 WL 3872781 (N.D. Cal. Aug. 15, 2018); Cain v. Twitter Inc., 2018 WL 4657275 (N.D. Cal. Sept. 24, 2018); Cohen v. Facebook Inc., 2017 WL 2192621 (E.D.N.Y. May 18, 2017); and Fields v. Twitter Inc., 2018 WL 626800 (9th Cir. Jan. 31, 2018).

86. Fields v. Twitter Inc., 2018 WL 626800 (9th Cir. Jan. 31, 2018).

87. Zann Isacson, “Combating Terrorism Online: Possible Actors and Their Roles,” Lawfare, September 2, 2018; Matt Egan, “Does Twitter Have a Terrorism Problem?,” Fox Business (website), October 9, 2013.

88. “Violent or Graphic Content Policies,” YouTube Help.

89. “Dangerous Individuals and Organizations,” Facebook Community Standards.

90. “The Twitter Rules,” Twitter Help Center.

91. Mark Zuckerberg, “Preparing for Elections,” Facebook, September 13, 2018.

92. Ithiel de Sola Pool believed national security issues would become more important in the electronics age: “Censorship is often imposed for reasons of national security, cultural protection, and trade advantage. These issues, which have not been central in past First Amendment controversies, are likely to be of growing importance in the electronic era.” Ithiel de Sola Pool, Technologies of Freedom (Cambridge, MA: Harvard University Press, 1983), p. 9.

93. Lamont v. Postmaster General, 381 U.S. 301 (1965).

94. Lamont v. Postmaster General, 307.

95. 11 CFR 110.20(f).

96. FEC.gov, FEC Record: Outreach, “Foreign Nationals.”

97. FEC.gov, “Foreign Nationals,” citing Bluman v. FEC, 800 F. Supp. 2d 281, 290 (D.D.C. 2011), affirmed, 132 Supreme Court 1087 (2012).

98. “Foreign Nationals Brochure,” Federal Election Commission, July 2003.

99. 22 U.S.C. § 611. The Mueller indictment notes that this disclosure informs “the people of the United States … of the source of information and the identity of persons attempting to influence U.S. public opinion, policy, and law.” This information in turn allows Americans to “evaluate the statements and activities of such persons in light of their function as foreign agents.” Indictment at 11, U.S. v. Viktor Borisovich Netyksho et al., Case 1:18-cr-00032-DLF (D.D.C. filed Feb. 16, 2018).

100. “General FARA Frequently Asked Questions,” Department of Justice, August 21, 2017.

101. Lobbying on behalf of commercial interests makes up a significant part of foreign lobbying of the U.S. government, see Holly Brasher, Vital Statistics on Interest Groups and Lobbying (Thousand Oaks, CA: SAGE Publications, 2014), pp. 136-44.

102. Jack Stubbs and Ginger Gibson, “Russia’s RT America Registers as ‘Foreign Agent’ in U.S.,” Reuters, November 13, 2017; James Kirchik, “Why Russia’s RT Should Register as an Agent of a Foreign Government,” Brookings (blog), September 22, 2017.

103. RT claims to have eight million weekly U.S. viewers, though the real numbers are likely far smaller. See Amanda Erickson, “If Russia Today Is Moscow’s Propaganda Arm, It’s Not Very Good at Its Job,” Washington Post, January 12, 2017.

104. The Russian ads would have still been illegal even if the funder had been disclosed.

105. Scott Shane, “How Unwitting Americans Encountered Russian Operatives Online,” New York Times, February 19, 2018.

106. Brendan Nyhan, “Fake News and Bots May Be Worrisome, but Their Political Power Is Overblown,” New York Times, February 19, 2018.

107. Byron York, “A Non-alarmist Reading of the Mueller Russia Indictment,” Washington Examiner, February 18, 2018.

108. Ross Douthat, “The Trolling of the American Mind,” New York Times, February 21, 2018.

109. Mark Zuckerberg, “Preparing for Elections,” Facebook, September 13, 2018. Subsequent quotations from Zuckerberg will refer to this source.

110. “What Super Pacs, Non-profits, and Other Groups Spending Outside Money Must Disclose about the Source and Use of Their Funds,” OpenSecrets.

111. Speaking during a 2017 congressional investigation of Russian efforts during the 2016 election, Sen. Diane Feinstein (D-CA) said to tech leaders: “You’ve created these platforms, and now they are being misused, and you have to be the ones to do something about it. Or we will.” Byron Tau, Georgia Wells, and Deepa Seetharaman, “Lawmakers Warn Tech Executives More Regulation May Be Coming for Social Media,” Wall Street Journal, November 1, 2017.

112. Zuckerberg, “Preparing for Elections.”

113. Stephen A. Siegel, “The Origin of the Compelling State Interest Test and Strict Scrutiny,” American Journal of Legal History 48, no. 4 (2006): 355-407.

114. We treat here “misinformation” or “disinformation” as a subset of fake news and more generally as a kind of false speech. Misinformation may be speech that is intentionally false. For First Amendment purposes, it would be difficult to distinguish unintentionally false speech from intentionally false speech. If that distinction cannot be made, the analysis that applies to false speech also includes misinformation, keeping in mind the focus here will be on public values.

115. Nadine Strossen, HATE: Why We Should Resist It with Free Speech, Not Censorship (New York: Oxford University Press, 2018), pp. 69-70.

116. Brooke Borel, “Fact-Checking Won’t Save Us from Fake News,” FiveThirtyEight, January 4, 2017.

117. Bertin Martens, Luis Aguiar, Estrella Gomez-Herrera, and Frank Mueller-Langer, “The Digital Transformation of News Media and the Rise of Disinformation and Fake News—An Economic Perspective,” Digital Economy Working Paper 2018-02, JRC Technical Reports, pp. 8-11.

118. See Alvarez, p. 4, quoting Ashcroft, “The First Amendment means that government has no power to restrict expression because of its message, its ideas, its subject matter, or its content.” United States v. Alvarez, 567 U.S. 709 (2012); and Ashcroft v. American Civil Liberties Union, 535 U.S. 564, 573 (2002).

119. Alvarez, p. 4.

120. Alvarez, p. 4.

121. Alvarez, p. 7.

122. Alvarez, p. 11.

123. See New York Times Co. v. Sullivan, 376 U.S. 254 (1964), and Alvarez.

124. United States v. Alvarez, 567 U.S. 709 (2012).

125. New York Times Co. v. Sullivan, 376 U.S. 254 (1964).

126. Protection for Private Blocking and Screening of Offensive Material, 47 U.S. Code § 230.

127. David French, “A Better Way to Ban Alex Jones,” New York Times, August 7, 2018.

128. Michael Barthel, Amy Mitchell, and Jesse Holcomb, “Many Americans Believe Fake News Is Sowing Confusion,” Pew Research Center, December 15, 2016.

129. Bertin Martens et al., “The Digital Transformation of News Media and the Rise of Disinformation and Fake News—An Economic Perspective,” Digital Economy Working Paper 2018-02, JRC Technical Reports, pp. 40-47.

130. George A. Akerlof, “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism,” Quarterly Journal of Economics 84, no. 3 (1970): 488-500.

131. Martens et al., “Digital Transformation,” pp. 51-52.

132. Richard Allan, “Hard Questions: Where Do We Draw the Line on Free Expression?,” Facebook Newsroom, August 9, 2018.

133. Allan, “Where Do We Draw,” which states, “And rather than blocking content for being untrue, we demote posts in the News Feed when rated false by fact-checkers and also point people to accurate articles on the same subject.”

134. Tessa Lyons, Facebook product manager, “Hard Questions: What’s Facebook’s Strategy for Stopping False News?,” Facebook Newsroom, May 23, 2018.

135. Laura Hazard Owen, “Facebook Is Paying Its Fact-Checking Partners Now (and Giving Them a Lot More Work to Do),” Nieman Lab.

136. “How Is Facebook Addressing False News through Third-Party Fact-Checkers?,” Facebook Help Center, Facebook.

137. Casey Newton, “A Partisan War over Fact-Checking Is Putting Pressure on Facebook,” The Verge (website), September 12, 2018.

138. See the updated database about fake news regulation throughout the world prepared by the Poynter Institute. Daniel Funke, “A Guide to Anti-misinformation Actions around the World,” Poynter Institute, Poynter.org.

139. The reputations of China, Russia, and Belarus are well known in this regard. Cameroon is less infamous, but its problems are summarized by a recent headline in The Guardian, “Cameroon Arrests Opposition Leader Who Claims He Won 2018 Election.”

140. Funke, “Guide,” Poynter.org. See the entry for France.

141. European Union (website), Publications Office of the EU, A Multidimensional Approach to Disinformation: Report of the Independent High Level Group on Fake News and Online Disinformation, March 30, 2018. For the process going forward, see p. 33.

142. European Commission, Multidimensional, pp. 35-38.

143. According to the Poynter Institute, neither Congress nor the states has tried to suppress fake news. The California legislature did pass a bill setting up an advisory commission “to monitor information posted and spread on social media.” The governor vetoed the bill. See “Governor Brown Vetoes Fake News Advisory Group Bill, Calls It ‘Not Necessary’,” CBS Sacramento (website), September 27, 2018.

144. Samuel Walker, Hate Speech: The History of an American Controversy (Lincoln: University of Nebraska Press), p. 1.

145. The authors appear somewhat skeptical about the effects of nature vs. nurture on IQ. See Richard J. Herrnstein and Charles Murray, The Bell Curve: Intelligence and Class Structure in American Life (New York: Free Press, 1996), p. 131. For the average IQ claim, see pp. 276-7.

146. Strossen, HATE, p. 71.

147. Dieter Grimm, “Freedom of Speech in a Globalized World,” in Extreme Speech and Democracy, ed. Ivan Hare and James Weinstein (New York: Oxford University Press, 2009), p. 13.

148. See Strafgesetzbuch (StGB), § 130 Volksverhetzung, 1-2.

149. Public Order Act, 1986, Part III.

150. Britain First is a “nationalistic, authoritarian, … nativist, ethnocentric and xenophobic” group hostile to Muslim immigrants in the United Kingdom. They are active online with significant consequences for their leaders if not for British elections. The leading and perhaps only scholarly study of the group is Chris Allen, “Britain First: The ‘Frontline Resistance’ to the Islamification of Britain,” Political Quarterly 85, no. 3 (July-September 2014): 354-61. See also the report by the organization Hope not Hate, “Britain First: Army of the Right,” November 2017. The leaders of Britain First, Paul Golding and Jayda Fransen, were incarcerated for distributing leaflets and posting online videos that reflected their extreme antipathy to Muslims. Fransen received a 36-week sentence and Golding an 18-week sentence. Kevin Rawlinson, “Britain First Leaders Jailed over Anti-Muslim Hate Crimes,” The Guardian, March 7, 2018.

151. For the origins of the debate, see Walker, Hate Speech, pp. 17-37.

152. James Weinstein, “An Overview of American Free Speech Doctrine and Its Application to Extreme Speech,” in Extreme Speech and Democracy, eds. Ivan Hare and James Weinstein (New York: Oxford University Press, 2009), p. 81.

153. Weinstein, “Overview,” pp. 81-82, quoting Rosenberger v. Rector and Visitors of University of Virginia, 515 U.S. 819, 829 (1995).

154. Weinstein, “Overview,” p. 82.

155. Weinstein gives examples of such settings: government workplace, state school classroom, and the courtroom. See “Overview,” p. 83.

156. Weinstein, p. 85, n. 34.

157. R.A.V. v. City of St. Paul, 505 U.S. 377 (1992). In the past, the Supreme Court upheld a group libel law, Beauharnais v. Illinois 343 U.S. 250 (1952). It is generally assumed that while the court has not formally overruled the precedent, it would not validate a group libel law today. See Weinstein, “Overview,” p. 88, and n. 52.

158. This assumes hate speech does not fall into a category of speech recognized as unprotected by the First Amendment (e.g., a “true threat”).

159. Richard Allan, “Hard Questions: Who Should Decide What Is Hate Speech in an Online Global Community?,” Facebook Newsroom, June 27, 2017. Allan is currently Facebook’s vice president for policy.

160. Allan, “Who Should Decide?”

161. Richard Allan, “Hard Questions: Where Do We Draw the Line on Free Expression?,” Facebook Newsroom, August 9, 2018.

162. Allan, “Who Should Decide?”

163. Maddy Osman, “28 Powerful Facebook Stats Your Brand Can’t Ignore in 2018,” Sprout Social (website).

164. “Prohibited Content,” AdSense Help, Google Support; “Hate Speech | Inappropriate Content | Restricted Content,” Developer Policy Center, Google Play.

165. “Violent or Graphic Content Policies,” YouTube Help, Google Support.

166. Catherine Shu, “YouTube Punishes Alex Jones’ Channel for Breaking Policies against Hate Speech and Child Endangerment,” Techcrunch, July 2018.

167. “Hateful Conduct Policy,” Twitter.

John Samples is a vice president of the Cato Institute and author of The Fallacy of Campaign Finance Reform (University of Chicago Press).