THE MYTH OF THE CHILLING EFFECT.

• Author: Bedi, Suneal

TABLE OF CONTENTS I. INTRODUCTION 268 II. THE CURRENT STATE OF THE CHILLING EFFECT 272 A. Historical Background 272 B. Recent Applications to Private Companies 277 C. Existing Empirical Work 281 III. MEASURING THE CHILLING EFFECT 284 A. Overview of the Study 284 B. Design of the Study 285 1. Sample 290 2. Procedure 291 C. Results of the Study 294 IV. THE FUTURE OF THE CHILLING EFFECT 302 A. The Need for More Empirical Work 302 B. The Creation of More Civil Speech 304 C. The Production of More Robust Exchanges 305 V. CONCLUSION 307 I. INTRODUCTION

A guiding legal principle of the First Amendment is the existence of the chilling effect. That is, certain speech regulations may have the "indirect effect of deterring a speaker from exercising her First Amendment rights." (1) Put another way, a regulation may only seek to censor, limit, or restrict certain types of speech, but because citizens fear sanctions, the same regulation ends up censoring, limiting, or restricting other potentially valuable types of speech. (2)

The specter of the principle is often raised in the face of broad and potentially ambiguous restrictions on speech that are argued to be unconstitutional, closely related to constitutional issues of the overbreadth doctrine. (3) The theory is if citizens are not confident of exactly what speech is being limited, they may overregulate their own speech for fear of sanctions, and hence may chill (deter and change) their speech in an unnecessary way. (4)

Although historically associated with constitutional issues surrounding the First Amendment and government censorship, the chilling effect principle has been recently employed against private speech regulations. The debate around limiting hate speech, offensive language, and political conspiracies on social media sites like Facebook and Twitter has reinvigorated discussions of the chilling effect. In January 2021, both Facebook and Twitter blocked users including former President Donald Trump and conspiracy theory supporters from their platforms, and publicly removed posts which were perceived to contribute to propaganda and violence. (5) This led to both criticism and praise, with some arguing that these social media speech restrictions problematically deter and chill otherwise valuable speech, and others arguing that no chilling takes place, and even if it did, it should not matter because the restrictions are important and necessary. (6)

The chilling effect is so frequently used in court cases, scholarly articles, and news outlets that its existence has become self-evident to the legal community--no law school First Amendment course is complete without some detailed discussion of the chilling effect. (7) Yet, surprisingly, little or no empirical scholarship has grappled with whether the chilling effect exists. Does speech in the face of censorship actually get chilled? If it does, in what ways is it chilled?

These questions sit at the core of this common constitutional guiding principle. However, they have effectively gone unanswered for over seventy years: this legal principle is "founded... on nothing more than unpersuasive empirical guesswork." (8) Measuring whether and how speech gets chilled will not only help legal scholars and judges better understand which regulations to question but also will foster more informed decision-making when private companies attempt to censor speech.

Measuring the chilling effect is difficult for various reasons including data collection, data analysis, causal events, and confounding factors. (9) In addition, various types of speech activities (e.g., more factual reporting-based, opinion-based, or a mix of the two) and mediums of speech activity (e.g., social media, in-person protests, traditional news reporting) may manifest the effect differently. A leading scholar once commented that the human behaviors that underly the chilling effect are "most likely unprovable." (10) Therefore, it is not terribly surprising that extant scholarship has neither confirmed nor disproved its existence. (11)

This Article seeks to fill this gap. It is the first empirical study analyzing speech content and measuring whether and how speech is actually chilled. It focuses on social media speech activity--online restaurant reviews--that has both fact and opinion components. In this social media context, this Article empirically showed that the chilling effect created little to no impact on the content of the message; at most, it slightly altered the specific style or tone used.

To do this, the Article uses an actual (not hypothetical) private social media speech restriction and experimentally manipulates how broad or narrow the speech restriction is. It then analyzes the online speech output itself using text analysis and third-party evaluations to show whether and how exactly the speech is changed in the face of a given restriction. The results of this empirical study show that social media user speech is more affected as regulations become broader--which confirms the existence of the chilling effect. But, while users did change their speech, the overall message of the speech did not change at all. That is, users change the tone of what they say (in this case by making it more positive) in the face of restrictions, but ultimately communicate the same message.

This result both partially confirms the chilling effect and simultaneously raises serious issues about how it is has been historically treated as a legal principle. While people do change how they speak in the face of a regulation, restrictions do not seem to have a substantive effect on the message communicated at least in the context studied. The lack of an observable and substantive chilling effect calls into question whether the chilling effect is a principle courts, litigants, governments, and private companies should put so much emphasis on.

In addition to questioning whether the chilling effect should be a guiding principle, the results of this Article may point to the conclusion that the chilling effect can be a good thing in some contexts. By changing only the tone of how individuals speak, rather than the actual content or purpose of a message, the chilling effect can encourage more civil forms of speech that are less offensive. When individuals are less offended, they are more likely to engage in thoughtful exchanges. These exchanges then ultimately promote participation in speech activity rather than upset it. This Article then points to the insight that recent restrictions on social media platforms may ultimately be good policy not in spite of, but because of the chilling effect.

This Article is in five Parts. Part II describes the current state of the legal principle of the chilling effect focusing on its recent use in private social media speech restrictions and extant empirical work. Part III introduces a novel empirical study that measures whether and how speech is chilled in the face of a private social media speech regulation. Part IV discusses the implications of the results for both public and private regulations going forward, focusing on the potential for chilling to actually promote speech rather than frustrate it. Part V concludes.

2. THE CURRENT STATE OF THE CHILLING EFFECT

   1. Historical Background

      The original concept of "chilling" dates back to 1952 with the Supreme Court case Wieman v. Updegraff. (12) In that case, Oklahoma required each state employee to pledge a "loyalty oath" indicating that they currently were not and had not been for five years part of any organization that the Attorney General of the United States deemed to be "subversive." (13) The Court invalidated the statute as it violated the Fourteenth Amendment. Justice Frankfurter in his concurring opinion stated: "Such unwarranted inhibition... has an unmistakable tendency to chill that free play of the spirit which all teachers ought especially to cultivate and practice; it makes for caution and timidity in their associations by potential teachers." (14) Justice Frankfurter's theory is that teachers would not associate freely with various organizations that may be protected and valuable associations because they may fear government sanctions. (15)

      The "chilling effect" later found its way into the speech regulation jurisprudence in 1963 in Gibson v. Florida. (16) Subsequently, the concept of chilling has been extensively applied to invalidate various government regulations, in particular those that implicate the First Amendment. (17) The idea is that the phenomenon of chilling can serve as a weapon to argue against the potential negative externalities associated with various government regulations. (18)

      In terms of a simple, workable definition, leading First Amendment scholar Frederick Schauer defines the chilling effect as something that occurs "when individuals seeking to engage in activity protected by the [F]irst [A]mendment are deterred from so doing by governmental regulation not specifically directed at that protected activity." (19) Take for example a federal law that makes it illegal to say anything vulgar to politicians. (20) In theory, this law would chill speech because individuals may be deterred from saying critical things to politicians, that would otherwise be protected by the First Amendment, due to the fear of being labeled vulgar and hence may change or taper their speech in certain ways.

      There are two important parts to the traditional definition of the chilling effect that are worth discussing further: deterrence and governmental regulation.

      The chilling effect principle rests on the assumption that individuals are frequently deterred in the face of government regulations. (21) But why are people deterred from engaging in an activity by a law that makes it illegal to engage in a different activity? Scholars have recognized that individuals are fearful of criminal and civil sanctions. (22) Ideally, of course, individuals would know exactly what would...