| Original Full Text | William & Mary Bill of Rights Journal Volume 32 (2023-2024) Issue 3 Article 2 3-2024 The Private Abridgment of Free Speech Erin L. Miller Follow this and additional works at: https://scholarship.law.wm.edu/wmborj Part of the Constitutional Law Commons, and the First Amendment Commons Repository Citation Erin L. Miller, The Private Abridgment of Free Speech, 32 Wm. & Mary Bill Rts. J. 615 (2024), https://scholarship.law.wm.edu/wmborj/vol32/iss3/2 Copyright c 2024 by the authors. This article is brought to you by the William & Mary Law School Scholarship Repository. https://scholarship.law.wm.edu/wmborj THE PRIVATE ABRIDGMENT OF FREE SPEECHErin L. Miller*ABSTRACTThis Article challenges the orthodoxy that First Amendment speech rights canbind only the state. I argue that the primary justification for the freedom of speech isto protect fundamental interests like autonomy, democracy, and knowledge from thekind of extraordinary power over speech available to the state. If so, this justificationapplies with nearly equal force to any private agents with power over speech rival-ing that of the state. Such a class of private agents, which I call quasi-state agents,turns out to be a live possibility once we recognize that state power is more limitedthan it seems and can be broken down into multiple, equally threatening parts. Quasi-state agents might include a limited set of corporations, from the largest social mediaplatforms to powerful private employers. However, because quasi-state agents arenot exactly like state agents but pursue important private aims that the state cannot,I argue that the First Amendment might bind them slightly differently (and less de-mandingly) than it does the state. Drawing on examples from American state andcomparative constitutional law, I offer several analytical models for understandingthis differential application.INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 616I. HOW THE FIRST AMENDMENT CURRENTLY BINDS (FEW) PRIVATE AGENTS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 620A. State Action Doctrine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6211. The Power Approach. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6222. The Modern Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 626B. Editorial Rights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 628II. THE THEORETICAL CASE FOR BINDING STATE-LIKE PRIVATE AGENTS. . . . . 633A. Why Bind the State by Constitutional Rights? . . . . . . . . . . . . . . . . . . . 634B. Why Bind the State by Free Speech Rights? . . . . . . . . . . . . . . . . . . . . 637C. Is the State’s Power Over Speech Unique? . . . . . . . . . . . . . . . . . . . . . 6391. Dense-and-Strong-Power Quasi-State Agents. . . . . . . . . . . . . . . . 6422. Broad-and-Strong-Power Quasi-State Agents. . . . . . . . . . . . . . . . 646D. State Motives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 652* Erin L. Miller, Assistant Professor of Law & Philosophy, University of SouthernCalifornia Gould School of Law. The author is indebted to feedback from Scott Altman,Elettra Bietti, Rebecca Brown, Ángel Díaz, Danieli Evans, Jonathan Gingerich, FelipeJiménez, Mugambi Jouet, Gregory Keating, Mathis Koschel, Jeesoo Nam, and MarcelaPrieto. Research assistance for German case law was kindly provided by Dominic Schabert.615616 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615III. REASONS AGAINST BINDING STATE-LIKE PRIVATE AGENTS . . . . . . . . . . . 653A. Personal Liberty. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6551. Expressive and Associative Liberty . . . . . . . . . . . . . . . . . . . . . . . 6552. Property Rights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 658B. Societal Harms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 660IV. DOCTRINAL MODELS FOR BINDING STATE-LIKE PRIVATE AGENTS . . . . . . 662A. Expanding State Action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 664B. Direct and Identical Duties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 665C. Direct and Lesser Duties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 667D. Indirect Duties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6701. Judicial Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6702. Legislative and Regulatory Definition . . . . . . . . . . . . . . . . . . . . . 672E. Distrust of the State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 673CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 674Private property owners who have so transformed the life of societyfor their profit (and in the process, so diminished its free speech)must be held to have relinquished a part of their right of freespeech. They have relinquished that part which they would now useto defeat the real and substantial need of society for free speech.—N.J. Coal. Against War in the Middle E. v. J.M.B. Realty Corp.1INTRODUCTIONIn his 1909 short story, “The Machine Stops,” E.M. Forster describes a worldin which the human population has, due to some calamity, lost the ability to surviveon the surface of the earth. Each person lives in an isolated chamber underground.There appear to be no common spaces, and people do not congregate together. Travelto visit other chambers is possible but arduous and unpopular. Humans communi-cate, between their chambers, using a form of video messaging. All systems belowthe surface, including the video messaging, are provided by a powerful technologyknown as the Machine, which is thought of as a sort of deity. It is unclear how theMachine originated or who, if anyone, influences its operations now.Imagine a world just like this one, but with two modifications: the Machine isnot a deity, but a powerful technology corporation; and the only technology itprovides is video messaging. The Machine would control nearly all communicationin this world, excepting only the occasional in-person conversations that peoplehave when they travel. Now imagine further that, one day, the Machine decides tocut off all outgoing video messaging services for approximately half of the world’spopulation, based on criticisms they have made of the Machine.1 650 A.2d 757, 780 (N.J. 1994).2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 617If this happened in the United States, it would not seem to offend any constitu-tional right. First Amendment doctrine treats public and private agents as categoricallydifferent with respect to the free speech guarantee. For all public (i.e., government)agents, that guarantee imposes a duty: whenever they act, they are bound to respectsubjects’ freedom of speech. For practically no private (i.e., nongovernment) agent,however, does the guarantee create any duty. The only regular exception is when aprivate agent’s conduct is causally linked with the state itself, such as when the stateencourages or facilitates it.2 This sharp public-private distinction is usually justassumed by courts and seldom challenged.The justification for the distinction might seem obvious: it is the state alone thatpossesses the extraordinary power to silence speakers and distort public discourse,whereas it is private agents who are vulnerable to that power. The typical private agentcan do exceedingly little to actually silence another’s speech or to influence the overallcourse of public discussion. At the same time, if private agents bore a duty not to treatother people differently based on their views and utterances, the enforcement of sucha duty would threaten interests in privacy, private property, freedom of association,and even freedom of expression itself. In other words, the distinction between publicand private is supposed to track the distinction between the powerful and the power-less, between those who need to be constrained and those who need to be protected.Yet this distinction begins to crack when applied to Forster’s Machine, or evento today’s non-fictional mammoth media corporations like Meta, Twitter, and Google.The critical communication channels that the latter own—from social media tosearch engines—are so widely used, especially for discussions of political issues,that scholars refer to them as the “public infrastructure” of communication, and theSupreme Court itself has described them as “the modern public square.”3 Theirowners—if they wanted—could drastically and immediately alter the topics andbalance of nationwide public discourse just by tweaking their speech-filtering al-gorithms or “deplatforming” specific speakers. In other words, they could funda-mentally transform our political order and thwart many of the lofty values associatedwith the First Amendment, from the openness of political discourse to the pursuitof knowledge and autonomy.And yet the First Amendment, as currently interpreted, is powerless to place anyobstacle in their path if they choose to do so. Indeed, the Amendment may guaranteetheir power if courts conceive of the exercise of that power as a matter of the corpora-tions’ own freedom of expression.4 Scholars alarmed by this power have sought to2 See, e.g., Evans v. Newton, 382 U.S. 296, 299 (1966) (stating private conduct that istoo entwined with the government will no longer be considered private conduct); MooseLodge No. 107 v. Irvis, 407 U.S. 163, 173 (1972) (explaining that private conduct that re-ceives a benefit from the state will be subject to constitutional protections). I discuss anotherrarely applied exception in Section I.A, infra.3 Packingham v. North Carolina, 582 U.S. 98, 107 (2017).4 The Court will likely decide this issue in Moody v. NetChoice, LLC, 144 S. Ct. 478(2023) (mem.).618 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615legally constrain it primarily through constitutional evasion: by denying that thecompanies’ editorial choices about the speech on their platforms are the sorts ofspeech acts protected by the First Amendment.5 These proposals, however, alsoevade the point that these choices are at the very core of First Amendment concern.This Article articulates a more muscular vision of the First Amendment, one thatconstrains power, whether it is private or public. I argue that the Amendment im-poses free speech duties on a limited set of private agents with extraordinary, state-like power over speakers or the primary channels of speech amplification. I callthese quasi-state agents. Their free speech duties, for reasons articulated in theforegoing, also override any prima facie First Amendment rights that these agentsmay have possessed.The argument runs as follows. The First Amendment explicitly forbids only“Congress” from abridging the freedom of speech.6 But judicial interpretation hasimplicitly expanded the category of duty holders to all state agents. Their justifica-tion is that these agents, like Congress, have power to impinge on the core FirstAmendment values, including the integrity of the democratic process, the pursuit ofcollective knowledge, and the autonomy of individual speakers.7 Therefore, for thetheoretical coherence of the doctrine, a further expansion is required, to bind allagents—public or private—with similar powers to impinge on these values. Indeed,the Supreme Court seemed to recognize the same in a series of cases from the 1940s,1950s, and 1960s—including the famous Marsh v. Alabama,8 which is gaining re-newed attention in scholarship9—before veering course.What sort of power impinges on First Amendment values? The answer requirestwo new conceptual distinctions. First, free speech values can be individual or collec-tive. Individual values are those that are largely realized by individuals, such asautonomy or participation in the democratic process; collective values are those thatare largely realized by society as a whole, and benefit individuals only indirectly,5 See, e.g., Pauline Trouillard, Social Media Platforms Are Not Speakers. Why AreFacebook and Twitter Devoid of First Amendment Rights?, 19 OHIO ST. TECH. L.J. 257,265–66 (2023) (arguing that content moderation is not speech); Eugene Volokh, TreatingSocial Media Platforms Like Common Carriers?, 1 J. FREE SPEECH L. 377, 408–09 (2021)(arguing that social media can be regulated as common carriers, without First Amendmentconcern, with respect to their function of “hosting” speech).6 See U.S. CONST. amend. I.7 See, e.g., Columbia Broad. Sys., Inc. v. Democratic Nat’l Comm., 412 U.S. 94 (1973)(Douglas, J., concurring) (explaining that Congress cannot cause a government agency toviolate the First Amendment). See generally Daniel J. Hemel, Executive Action and the FirstAmendment’s First Word, 40 PEPP. L. REV. 601 (2013).8 326 U.S. 501 (1946).9 See Genevieve Lakier & Nelson Tebbe, After the “Great Deplatforming”: Reconsider-ing the Shape of the First Amendment, LPE PROJECT (Mar. 1, 2021), https://lpeproject.org/blog/after-the-great-deplatforming-reconsidering-the-shape-of-the-first-amendment/ [https://perma.cc/4YZG-R46D].2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 619such as the integrity of the democratic process or the advancement of collectiveknowledge. Second, power over speech—that is, the power to control what is saidand heard—is not monolithic but has at least three different dimensions: strength, thelikelihood that exerting the power will produce the acts of speaking or hearingfavored by the agent with the power; density, the number and importance of oppor-tunities for speaking or listening that are under the agent’s power; and breadth, thenumber of individuals whose speaking or listening can be affected by the power.Equipped with these distinctions, we can see that, to threaten a given type offree speech value, an agent only needs to have two of the three dimensions of power.To threaten individual values, an agent needs to be able to reliably control what oneindividual says or hears across nearly all channels of communication. That requirespower that is high in strength (reliable control) and density (over many channels),but not in breadth. To threaten collective values, an agent only needs to reliablycontrol at least one channel of speech amplification that is very broadly used acrossa population. Such an agent has power that is high in strength (reliable control) andbreadth (over many speakers/listeners), but not in density.Identifying the exact degree of strength, density, or breadth of power overspeech that poses an intolerable threat to First Amendment values would be difficultfor three reasons: because drawing lines on spectrums is always challenging; becausejudges and constitutional theorists—those whose writings might help to establishsuch a line—have always just presumed, arguendo, that the state is the only relevantthreat; and because different free speech theories might yield different answers. So,this Article argues by analogy: because the state’s power over speech poses intolera-ble threats to free speech values, private agents with power over speech comparableto the state’s—in the sense of posing a similar threat to a core free speech value—should be similarly constrained by free speech rights.Two types of currently existing corporations are likely to fit the bill, eachpossessing great power over speech along different dimensions. The first are thecorporations mentioned above that own communications services used regularly bynearly a majority of the population, such as the largest social media platforms andsearch engines (think Facebook, YouTube, and Google).10 These corporations havevery strong and broad power over speech, as described above. Corporations in thesecond category are less often in the limelight: private employers in labor marketswith a low number of jobs and a high number of workers available to fill them.These corporations can have very strong and dense power over speech: the abilityto make certain individuals afraid to speak up about public issues, potentially foryears, for fear of losing their jobs.10 See Belle Wong, Top Social Media Statistics and Trends of 2023, FORBES ADVISOR(May 18, 2023, 2:09 PM), https://www.forbes.com/advisor/business/social-media-statistics/[https://perma.cc/54HJ-PSWS] (reporting that in 2023 Facebook and YouTube each had 2.9million and 2.5 million monthly active users across the world respectively, making them themost used social media platforms in the world).620 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615None of this is to say that the duties imposed on any quasi-state agent wouldlook exactly like those typically imposed on the state. To the extent that the agentsdiffer from the state, their duties might be different or even less demanding. Doctrin-ally, this might work in two ways. First, the doctrine might be altered such that theduties of the state would apply to private quasi-state agents in a diluted or otherwisemodified fashion. Second, the differences might be accommodated within existingdoctrine. The First Amendment is already more context-sensitive than one mightthink; even the state possesses modified (and in a sense, less demanding) dutieswhen it is acting in certain managerial roles such as educator or employer. Perhapsoperating certain private organizations is simply a distinct managerial role callingfor distinct duties.Models for adopting constitutional law to private parties are more easily foundthan American constitutional lawyers might suppose. High courts in both Americanstates and foreign nations have interpreted their constitutions to provide speech rightsagainst certain private entities. This Article, in its final lap, draws on those bodiesof case law to develop options for doctrinally implementing private free speechduties. I offer three options: quasi-state agents could have duties directly under theFirst Amendment that are formally (though perhaps not substantively) identical tothose of the state; they could have duties directly under the First Amendment thatare less demanding than those of the state; or they could have duties indirectly underthe First Amendment, to be enacted by government officials via legislation, regula-tion, or the common law that reflect constitutional principles.The Article proceeds as follows. Part I introduces the historical Supreme Courtcases mentioned above that, for a time, recognized First Amendment duties for cer-tain powerful private agents that cut against two traditional doctrines: the stateaction doctrine and media’s editorial rights. Part II explains that these deviant casesreflect a principle underlying First Amendment law and theory: that the state isbound by free speech duties because of its extraordinary power over speech and thusits ability to impinge on free speech values. I then argue that this same principlejustifies an extension of at least some First Amendment duties to the class of quasi-state agents. Part III considers several reasons we might nonetheless have for di-luting the free speech duties that we impose on quasi-state agents, relative to thoseof the state. These reasons pertain to the agents’ own liberty claims, and to the socialvalue of their pursuit of private aims. Finally, Part IV draws on comparative law andU.S. state judicial opinions to present the three possibilities for how quasi-stateagents might be held to have free speech duties.I. HOW THE FIRST AMENDMENT CURRENTLY BINDS (FEW) PRIVATE AGENTSThe proposition that some private agents, acting entirely independent of govern-ment, must respect free speech duties appears to run against the current of precedent.2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 621Countless cases assume that constitutional rights—with one exception11—create dutiesonly for government agents. But my argument is not an interpretation of the case lawthat seeks consistency with every case outcome. Rather, it aims to articulate a princi-ple that represents a deeper undercurrent in the doctrine: that what the First Amend-ment constrains is extraordinary power—power over the channels of communication.This underlying principle has even—on occasion—surfaced to determine caseoutcomes. For fleeting moments during the mid-twentieth century, the Supreme Courtextended free speech duties to private agents. It did so precisely in cases in whichprivate power over speech was at its zenith: when a big company had a firm gripover critical channels of communication. Unsurprisingly, the opinions do not painttheir holdings as revolutionary; and those holdings were closely confined by subse-quent decisions. The cases nonetheless demonstrate the influence of the principledundercurrent; highlight areas of doctrine that need to flex to overtly accommodateit; and offer a glimpse of what such an accommodation might look like in practice.This Part presents these cases, organized around the two First Amendmentdoctrines that they appeared to flout. The first and most obvious of these doctrinesis the state action doctrine, which establishes that parties can only raise constitu-tional rights claims against the government, not against private actors. The secondis the “editorial rights” doctrine, according to which media entities have rights tochoose exactly what and how to publish of others’ speech—even if that means thatthey deny other private parties the opportunity to speak. This doctrine implies thatmedia entities lack conflicting duties like the ones advocated in this Article.Here, the discussion will be purely descriptive. In later sections of the Article,I will defend the theoretical merits of these cases and explain how their insightscould be incorporated more systematically into free speech jurisprudence.A. State Action DoctrineThe rule that no private agent has constitutional free speech duties is mostclearly enforced in constitutional law by the state action doctrine. This doctrinerequires that, for a plaintiff to bring a claim that her constitutional right was vio-lated, she must prove that the action was taken by a state official or institution.12 Itis often viewed as a threshold requirement, such that, unless it is met, a court cannoteven hear the merits of the claim. The requirement applies not just to free speechrights but to every individual right in the Constitution, aside from the right againstslavery and involuntary servitude.13 It is religiously applied, perhaps especially with11 The Thirteenth Amendment secures the right against enslavement and involuntary servi-tude against everyone, not just the government. U.S. CONST. amend. XIII.12 See Lugar v. Edmondson Oil Co., 457 U.S. 922, 936 (1982) (outlining the state-actionrequirement).13 See U.S. CONST. amend. XIII.622 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615respect to free speech. Courts have repeatedly thrown out frivolous lawsuits claim-ing that private actors like Facebook or Amazon have violated someone’s freedomof speech.14A limited number of exceptions to the state action doctrine exist. The primaryset of exceptions allows an action to be counted as state action when it was taken bya private agent at the behest of15 or with the approval of the state,16 or by a privateagent institutionally entwined with the state.17 These “exceptions” could also be char-acterized as relatively straightforward extensions of the state action doctrine, insofaras they involve causal involvement on the part of the state.But the final and most controversial exception applies to agents that are fully,structurally and causally, independent of the state. Under this exception, a privateagent may be bound by constitutional rights, including free speech rights, when itexercises a “public function.”18 Below, I explain how the public function exception,when it was first introduced in the First Amendment context, rested on a recognitionof the state-like power of the entities to which it applied. However, this functionalistpower analysis quickly gave way to a more formalistic inquiry into whether theagent was like the state in all respects or served a function that had only ever beenprovided by the state—shrinking the number of exceptional cases recognized effec-tively to zero.1. The Power ApproachThe Supreme Court first recognized the public function exception for a “com-pany town” in the 1946 case Marsh v. Alabama.19 While the opinion in this case didnot explicitly claim to create any exception to the state action doctrine, the Courtlater read it as doing so20—and for good reason.2114 Cf. Parler LLC v. Amazon Web Servs., LLC, 514 F. Supp. 3d 1261, 1264 (W.D. Wash.2021) (affirming that the First Amendment only applies to the government and not privateactors).15 See Adickes v. S.H. Kress & Co., 398 U.S. 144, 170 (1970) (discussing how a state isresponsible for a private party’s action when state law has compelled the act).16 See Moose Lodge No. 107 v. Irvis, 407 U.S. 163, 178–79 (1972) (holding that statesanctions enforcing adherence by private parties to rules that violate constitutional rights areunconstitutional).17 See Evans v. Newton, 382 U.S. 296, 299 (1966) (“Conduct that is formally ‘private’may become so entwined with governmental policies . . . as to become subject to the con-stitutional limitations placed upon state action.”).18 See Manhattan Cmty. Access Corp. v. Halleck, 139 S. Ct. 1921, 1928–30 (2019).19 326 U.S. 501, 509 (1946).20 See, e.g., Halleck, 139 S. Ct. at 1942 n.11 (Sotomayor, J., dissenting) (describing Marshas extending First Amendment liability to a company town); Lugar v. Edmondson Oil Co., 457U.S. 922, 939 (1982) (describing Marsh as articulating a “public function” test for state action).21 One might alternatively read Marsh as simply concluding that the state cannot enforce2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 623Chickasaw, Alabama, was a densely populated town owned in its entirety—buildings, streets, sidewalks, sewers, and all—by Gulf Shipbuilding Corporation,a private company.22 Most of Chickasaw’s residents were employees of Gulf.23 Thetown connected to a public highway, and non-resident members of the public couldfreely enter it.24 The town had a commercial area in which many companies hadrented out stores, and Gulf had posted in the windows of all of these stores the samenotice: “This Is Private Property, and Without Written Permission, No Street, orHouse Vendor, Agent or Solicitation of Any Kind Will Be Permitted.”25 Grace Marsh,a Jehovah’s Witness who lived outside of Chickasaw, attempted to distribute reli-gious literature on the sidewalk outside one of the stores and was asked to leave.26When she refused, she was arrested for trespassing.27The Court reversed Marsh’s resulting conviction as a violation of her FirstAmendment rights by Gulf.28 The corporation could not be allowed “to govern acommunity of citizens so as to restrict their fundamental liberties . . . .”29 JusticeHugo Black’s opinion explicitly compares the power that Gulf had over speechwithin Chickasaw to the power a municipal government would have over speechwithin its jurisdiction, describing both as inherently constrained.30 While a munici-pality would have the physical power to ban the distribution of religious literatureon its streets, it would not have the authoritative power to do so given the FirstAmendment. Black says the same of Gulf: he denies that “the mere fact that all theproperty interests in the town are held by a single company is enough to give that com-pany power, enforceable by a state statute, to abridge these freedoms [of speech].”31In what sense does a private corporation’s power need to be state-like in orderto be bound by the First Amendment? The opinion is hardly a model of clarity, but itgives two major clues. The first is its emphasis that Gulf was acting like a companyits trespassing laws in ways that violate its own duties; in a sense, the private act becomesan act of state through enforcement. See, e.g., Frederick Schauer, Hudgens v. NLRB and theProblem of State Action in First Amendment Adjudication, 61 MINN. L. REV. 433, 435–36(1977) (offering this as one possible interpretation). However, if the interpretation rests onthe premise that the state acts whenever it enforces its laws relating to private property, thenit would implode the state action doctrine: private agents would be effectively bound by allconstitutional rights if they ever want their property rights enforced. While I have some sym-pathies for such a reading, as I explain in Part IV, my reading is equally plausible (as discussedin this section) and has far less radical implications. See discussion infra Section IV.A.22 See Marsh, 326 U.S. at 502–03.23 Cf. id. at 502.24 Id. at 503.25 Id.26 Id.27 Id. at 503–04.28 See id. at 508–10.29 Id. at 509.30 Id. at 504–06.31 Id. at 505.624 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615that provides critical public infrastructure in a monopolistic or oligopolistic market—an entity today often called a “common carrier.”32 Black developed a lengthy anal-ogy between Gulf’s company town and typical common carriers, “privately heldbridges, ferries, turnpikes and railroads . . . .”33 He describes these enterprises asdistinctive for two reasons: they “open[] up [their] property for use by the public ingeneral” and provide a “public function,” seemingly a service vital to the publicinterest.34 Such common carriers, Black explicitly observes, are bound by the Con-stitution on a sliding scale: “The more an owner, for his advantage, opens up hisproperty for use by the public in general, the more do his rights become circum-scribed by the statutory and constitutional rights of those who use it.”35Relevant to the First Amendment context of Marsh, Gulf was serving the “pub-lic function” of operating certain channels of communication. As Black states it,“[w]hether a corporation or a municipality owns or possesses the town the public ineither case has an identical interest in the functioning of the community in such man-ner that the channels of communication remain free.”36 But it was surely not enoughfor Gulf to control just any channel of communication. Black explains further:Many people in the United States live in company-owned towns.These people, just as residents of municipalities, are free citizensof their State and country. Just as all other citizens they mustmake decisions which affect the welfare of community andnation. To act as good citizens they must be informed. In orderto enable them to be properly informed their information mustbe uncensored.37Notice first the passive language—“their information must be uncensored”—confirming that, for First Amendment purposes, what matters is that the interest ininformation is infringed, irrespective of who does the infringing. The language alsosuggests that Gulf controlled channels of communication vital or numerous enoughto actually “censor” information necessary for democratic participation. Gulf could32 See Christopher S. Yoo, The First Amendment, Common Carriers, and Public Accom-modations: Net Neutrality, Digital Platforms, and Privacy, 1 J. FREE SPEECH L. 463, 465–75(2021) (describing contested definitions of a common carrier).33 Marsh, 326 U.S. at 506–09.34 Id. at 506. The idea of treating corporations as bound by constitutional rights whenthey performed public functions can be found as far back as Justice Harlan’s dissent in theCivil Rights Cases. 109 U.S. 3, 48 (1883) (“It is fundamental in American citizenship that,in respect of [civil] rights, there shall be no discrimination by the State, or its officers, or byindividuals, or corporations exercising public functions or authority, against any citizenbecause of his race or previous condition of servitude.”) (emphasis added).35 Marsh, 326 U.S. at 506.36 Id. at 507.37 Id. at 508.2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 625not effectively ban its citizens from acquiring information anywhere. But it con-trolled a central distribution point for information in Chickasaw: the streets in thebusiness district, and indeed all the streets.From the common carrier and censorship discussions in Marsh, we can gatherthat the private agents that must respect free speech rights include those that controlcertain channels of communication on which the public in general—especially intheir capacity as voters—depend for information. Given the Court’s examples, thechannels may need to function as public forums (i.e., central spaces for the discus-sion of public issues) or be otherwise critical for democracy (e.g., channels on whichmany people depend). Either way, the Court recognized that a corporation capableof infringing on vital free speech interests, on a scale that the state could, was boundby the First Amendment.After a long period of disuse, Marsh’s public function exception was revived inthe late 1960s—again, in a First Amendment case. Amalgamated Food EmployeesUnion Local 590 v. Logan Valley Plaza arose after members of a local union beganpicketing outside a store that had hired nonunion employees.38 Logan Valley Plaza,the large shopping center within which the store was located, secured an injunctionagainst picketing on the center’s grounds.39 The Court lifted the injunction, applyingthe public function exception and citing Marsh.40 Justice Marshall, writing for themajority, analogized the shopping mall to the business district of Chickasaw.41 Heemphasized the Marsh opinion was reinforced by the economic development of theUnited States that had led to a “large-scale movement of this country’s populationfrom the cities to the suburbs . . .” and the accompanying “advent of the suburbanshopping center . . .” which he seemed to anticipate might rival or even replace citystreets and parks—the original public forums—as the center of public life.42 In otherwords, he emphasized the shopping mall’s control over a space—a functional busi-ness district—that was critically important for realizing free speech interests.43Logan Valley arguably expanded the public function exception beyond Marsh.Marshall acknowledged that the shopping center’s total power was much less thanGulf’s; the center’s entire power was comparable only to Gulf’s power over the com-mercial area of Chickasaw.44 He nonetheless suggested that no more than this powerwas at issue in Marsh, because there was “no showing made there that the corporateowner would have sought to prevent the distribution of leaflets in the residential areasof the town.”45 At the very least, Logan Valley clarified that a corporation could bebound by First Amendment duties even if it lacked total power like the state’s.38 391 U.S. 308, 311 (1968).39 See id. at 313.40 See id. at 325 (citing Marsh, 326 U.S. at 506).41 See id. at 318.42 Id. at 324.43 Id. at 319–20.44 Id. at 318.45 Id. (emphasis added).626 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615Marshall also hinted at limitations on the public function exception. He ac-knowledged that “it may well be that respondents’ ownership of the property herein question gives them various rights, under the laws of Pennsylvania, to limit theuse of that property by members of the public in a manner that would not be permis-sible were the property owned by a municipality.”46 The shopping center, he ac-knowledged, might not be “without power to make reasonable regulations governingthe exercise of First Amendment rights on their property. Certainly their rights tomake such regulations are at the very least co-extensive with the powers possessedby States and municipalities . . . .”47The effects of Logan Valley, at least for a time, radiated into state constitutionallaw. Over about the next decade, several state supreme courts extended the reach oftheir state free speech clauses to some powerful private agents, primarily large shop-ping complexes and universities. Most notably, in 1980 in State v. Schmid, the Su-preme Court of New Jersey deemed Princeton University bound by state constitutionalfree speech duties.48 The court went so far as to conclude that New Jersey constitu-tional rights are subject to no state action doctrine at all.49 California’s Supreme Courtinterpreted its constitution to provide a right of “speech and petitioning, reasonablyexercised, in shopping centers even when the centers are privately owned” becausethose centers “provide an essential and invaluable forum for exercising those rights.”50The constitutions of Massachusetts, Colorado, and Washington were also read to granta right—either under free speech provisions or other democratic process provisions—to collect signatures for ballot access or initiatives at regional shopping malls.512. The Modern ApproachThe public function exception has not exactly flourished since Marsh. LoganValley’s extension of the exception beyond the company town was first confined toits facts in a case called Lloyd Corp. v. Tanner, and eventually altogether over-ruled.52 In limiting Marsh’s application, Lloyd explained that Marsh had “involved46 Id. at 319.47 Id. at 320.48 423 A.2d 615, 632–33 (N.J. 1980).49 Id. at 628.50 Robins v. Pruneyard Shopping Ctr., 592 P.2d 341, 347 (Cal. 1979).51 Batchelder v. Allied Stores Int’l, Inc., 445 N.E.2d 590, 597 (Mass. 1983); Bock v.Westminster Mall Co., 819 P.2d 55, 61 & n.7 (Colo. 1991) (en banc); Alderwood Assocs.v. Wash. Env’t Council, 635 P.2d 108, 117 (Wash. 1981) (en banc).52 Lloyd Corp. v. Tanner, 407 U.S. 551, 570 (1972) (holding that a shopping center hadnot infringed First Amendment rights when it told distributors of anti-war handbills on itspremises to leave); see also Hudgens v. NLRB, 424 U.S. 507, 518 (1976) (formally over-ruling Logan Valley, 391 U.S. 308).2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 627the assumption by a private enterprise of all of the attributes of a state-createdmunicipality and the exercise by that enterprise of semi-official municipal functionsas a delegate of the State.”53 This time, the Court picked apart differences betweenthe company town and the involved shopping center, such as that Gulf owned landbeyond the shopping district of Chickasaw whereas the Lloyd Corporation did not.54Marsh thus survived, but only as a formalistic test for whether a private agent walksand talks exactly like a state.In the years that followed, the Court emphasized that “very few” functionsqualify as public in Marsh’s sense.55 The Court applied the classification to only oneother setting, primary elections, and excluded from it a variety of activities, includ-ing “running sports associations and leagues, administering insurance payments,operating nursing homes, providing special education, representing indigent crimi-nal defendants, resolving private disputes, and supplying electricity.”56 It appearsthat the Marsh doctrine is now only applicable to two relics of history: companytowns and all-white primaries.57The Court ultimately clarified that the public function exception applies onlywhere the private agent has “powers traditionally exclusively reserved to the State.”58For instance, three years ago in Manhattan Community Access Corp. v. Halleck, aradio channel creator claimed that Manhattan Neighborhood Network, a non-profitthat runs New York City’s public access channels, had violated her First Amend-ment rights by barring her from all public channels because of a show she producedthat was critical of the network.59 The Court dismissed Halleck’s claim, and empha-sized the history of public access channels run by private agents to prove that thestate did not exclusively engage in this activity.60 The Court also described thefunction of Manhattan Neighborhood Network more generally as providing a forumfor speech—an activity that, obviously, is not an exclusively government function.61Indeed, it is hard to imagine any function involving communication that is exclu-sively a government function. Never mind that the running of a public forum was theoriginal public function recognized by the Court in Marsh and Logan Valley.The state cases applying free speech duties against private agents had somewhatlonger-lasting effects than their Supreme Court counterparts. The Logan Valley53 Lloyd Corp., 407 U.S. at 569 (emphasis added).54 Id. at 562–64.55 Flagg Bros. v. Brooks, 436 U.S. 149, 158 (1978).56 Manhattan Cmty. Access Corp. v. Halleck, 139 S. Ct. 1921, 1929 (2019) (collectingcases).57 See Terry v. Adams, 345 U.S. 461, 468–70 (1953) (applying the Fifteenth Amendmentin the context of primary elections); Smith v. Allwright, 321 U.S. 649, 662–66 (1944) (same).58 Halleck, 139 S. Ct. at 1928 (citing Jackson v. Metro. Edison Co., 419 U.S. 345, 352(1974)).59 Id. at 1927.60 Id. at 1929–30.61 Id. at 1930.628 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615spinoffs recognizing state constitutional rights to gather signatures in shoppingcenters were never officially revoked. Most were narrowly cabined by the 1990s,with the partial exceptions of New Jersey and Massachusetts.62 Indeed, the freespeech clause of New Jersey remains without any state action requirement at all, andthe question remains open in Massachusetts.63B. Editorial RightsAs explained above, courts have held that, under the First Amendment, mediaentities have robust “editorial rights” to decide which speech to publish, or not. Courtshave applied these rights to newspapers, cable television companies, and (so far) so-cial media companies.64 If a company has such an enforceable right, then this impliesthat they cannot have an enforceable duty, like the state, to respect free speechrights. They are allowed to discriminate finely based on the content and viewpointof the speech they are considering for publication.Yet a small number of cases from the mid-to-late twentieth century, the sameperiod as Marsh and Logan Valley, found that certain very powerful media compa-nies lack editorial rights and could be forced, by Congress or federal agencies, torespect free speech values. The rationale for treating these media companies differ-ently lies, again, in the nature of the companies’ power to undermine free speechinterests. The key cases are Associated Press v. United States, Red Lion Broadcast-ing Co. v. FCC, and Turner Broadcasting System, Inc. v. FCC (Turner I).6562 Compare Southcenter Joint Venture v. Nat’l Democratic Pol’y Comm., 780 P.2d 1282,1283 (Wash. 1989) (en banc) (rejecting a right to solicit contributions and distribute literaturein a shopping mall), and Golden Gateway Ctr. v. Golden Gateway Tenants Ass’n, 29 P.3d797, 810 (Cal. 2001) (finding no right to distribute leaflets in an apartment complex), Comm.for a Better Twin Rivers v. Twin Rivers Homeowners’ Ass’n, 929 A.2d 1060, 1072, 1074(N.J. 2007) (finding that a restriction on expressive activities in a “common-interest [residen-tial] community” was not “unreasonable or oppressive,” and therefore did not violate speechrights), with Dublirer v. 2000 Linwood Ave. Owners, Inc., 103 A.3d 249, 260 (N.J. 2014)(holding that a restriction on posting notices and distributing campaign materials inside aprivate cooperative apartment building was unreasonable and thus violated the residents’speech rights), and Glovsky v. Roche Bros. Supermarkets, Inc., 17 N.E.3d 1026, 1035(Mass. 2014) (applying the right to gather signatures only to area outside a supermarket).63 See generally Roman v. Trs. of Tufts Coll., 964 N.E.2d 331 (Mass. 2012) (decliningto find a speech rights violation on private college property but only because a state itselfwould not have violated rights under the circumstances).64 Mia. Herald Publ’g Co. v. Tornillo, 418 U.S. 241, 258 (1974) (newspapers); TurnerBroad. Sys., Inc. v. FCC (Turner I), 512 U.S. 622, 636 (1994) (cable); NetChoice, LLC v.Att’y Gen., Fla., 34 F.4th 1196, 1203 (11th Cir. 2022) (holding social media platforms’content moderation does fall within protected editorial judgment).65 See generally Associated Press v. United States, 326 U.S. 1 (1945); Red Lion Broad.Co. v. FCC, 395 U.S. 367 (1969); Turner I, 512 U.S. 622.2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 629The earliest case, decided in 1945, involved the Associated Press (AP), a co-operative association of newspapers, whose employees collected and wrote up newsand then distributed that news for a fee to the association’s members.66 The AP gaveits members the ability to block non-member competitors from membership (andthus from access to AP stories).67 The United States brought an antitrust lawsuitagainst the AP for this policy.68 One of the AP’s defenses was that enforcing anti-trust law against it would abridge the freedom of the press by compelling the membersof the AP to share the product of their ingenuity with others before publishing itthemselves.69 The Court rejected this argument.70 The justices’ brief explanation wastelling, and worth quoting in full.It would be strange indeed . . . if the grave concern for freedomof the press which prompted adoption of the First Amendmentshould be read as a command that the government was withoutpower to protect that freedom. The First Amendment, far fromproviding an argument against application of the Sherman Act,here provides powerful reasons to the contrary. That Amendmentrests on the assumption that the widest possible dissemination ofinformation from diverse and antagonistic sources is essential tothe welfare of the public, that a free press is a condition of a freesociety. Surely a command that the government itself shall notimpede the free flow of ideas does not afford non-governmentalcombinations a refuge if they impose restraints upon that consti-tutionally guaranteed freedom. Freedom to publish means free-dom for all and not for some. Freedom to publish is guaranteedby the Constitution, but freedom to combine to keep others frompublishing is not. Freedom of the press from governmental in-terference under the First Amendment does not sanction repres-sion of that freedom by private interests.71One can scarcely find in a judicial opinion clearer language acknowledging that theFirst Amendment protects a freestanding right to free speech—that it is not only arestraint on government, but on all agents powerful enough to repress that right.While the free press claims of the AP members were perhaps not especially strongin this case, the Court sets the stage for its future approval of much more reachinggovernment regulation on large media companies.66 Associated Press, 326 U.S. at 4.67 Id. at 4–9.68 Id. at 4.69 Id. at 14–15.70 Id. at 15–16.71 Id. at 20 (emphasis added).630 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615Two decades later, the Court in Red Lion Broadcasting Co. v. FCC, allowed theFederal Communications Commission (FCC) to place far more stringent restrictionson the speech of broadly powerful media companies.72 The Court’s First Amend-ment cases are extremely skeptical of state regulations of media that interfere withthe content of what media may publish. However, in Red Lion, Justice White’s opin-ion upheld the FCC’s rule, called the fairness doctrine, which required that radio andtelevision broadcasters to discuss public issues and give at least some coverage toeach side of those issues in order to maintain their federal licenses.73 Effectively, theFCC was telling broadcasters to cover some universe of content and, given thefinitude of the airwaves under their control, not some other universe of content.74In response to the broadcasters’ claim that the fairness doctrine abridged theirFirst Amendment rights, the Court maintained that broadcasters did not have the samefreedom of “refusing in [their] speech or other utterances to give equal weight to theviews of [their] opponents” as did other speakers or media actors.75 In other words, theCourt declared that broadcasters lacked at least some of the standard editorial rights.This holding is generally read as resting on a unique feature of the broadcastmedium, namely, the physical scarcity of the spectrum. Only a finite number ofradio and television frequencies exist, such that it is not even possible for everyoneto speak over them; even if there were enough airwaves for everyone to speak, thespeech would, due to signal interference, not amount to “intelligible communica-tion.”76 It is under these conditions that broadcasters lack editorial rights. The Courtstates: “[w]here there are substantially more individuals who want to broadcast thanthere are frequencies to allocate, it is idle to posit an unabridgeable First Amend-ment right to broadcast comparable to the right of every individual to speak, write,or publish.”77 The Court makes a point of observing that the broadcast spectrum is“increasingly congested” and that, while the spectrum is expanding, the uses for thatspectrum are also growing apace over time.78But I suggest that the feature of broadcasting that most strongly underwrote theCourt’s opinion was not physical scarcity, but a form of economically inducedscarcity: concentrated market power. What mattered to the Court was not just thelimited number of frequencies that existed, but the limited number of frequenciesthat in fact reached a large audience. The fact that such a small number of radio andtelevision stations were able to dominate viewership was due to the lack of a suf-ficiently competitive marketplace for these media. There are several reasons to72 395 U.S. 367, 400–01 (1969).73 Id. at 369–75.74 See id. at 375–78.75 Id. at 386–87.76 Id. at 388.77 Id.78 Id. at 396–97.2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 631support this reading. First, the Court repeatedly implies that “the right of the publicto receive suitable access to social, political, esthetic, moral, and other ideas andexperiences . . .” requires access to a diversity of such material.79 It repeatedlyworries about the limited number of views on the air, stating that without the fairnessdoctrine, the “station owners and a few networks would have unfettered power tomake time available only to the highest bidders, to communicate only their ownviews on public issues, . . . and to permit on the air only those with whom theyagreed.”80 The Court here even mentions “private censorship,” as it did in Marsh.81Second, the Court explicitly invokes the idea of an economic monopoly no fewerthan seven times, even though the respondent broadcasters—including CBC andNBC in addition to Red Lion Broadcasting—were clearly not monopolies. At onepoint, the Court declares: “It is the purpose of the First Amendment to preserve anuninhibited marketplace of ideas in which truth will ultimately prevail, rather thanto countenance monopolization of that market, whether it be by the Governmentitself or a private licensee.”82 Third, the Court mentions that the top broadcastersnow (e.g., CBC and NBC) had gained an edge in the market from the FCC’s de-cision to grant them licenses at the advent of broadcasting.83 Their licenses gave theman advantage over new entrants to the market and solidified their dominance.84 Thesehistorical observations were irrelevant to the physical scarcity of the broadcast spec-trum, but were crucial to an account of concentration in the broadcasting market.Red Lion, read as an opinion about concentrated media markets rather thanphysical scarcity, implies that media companies lose their editorial rights preciselywhen they are powerful enough to affect whether or not the general public is hearinga diversity of viewpoints. When this happens, their would-be editorial discretionconflicts with (and is trumped or even negated by) the First Amendment right of thepublic to be informed. As the Court says: “[T]he people as a whole retain theirinterest in free speech by radio and their collective right to have the medium func-tion consistently with the ends and purposes of the First Amendment. It is the rightof the viewers and listeners, not the right of the broadcasters, which is paramount.”85This makes legal space for Congress or the FCC to sweep in to force the broad-casters to respect the right of the public. Nothing the Court says settles the questionof whether the broadcasters might also have a freestanding constitutional duty to dothe same, but nothing it says is incompatible with this line of thought, either. I returnto this distinction in Part IV.79 Id. at 390.80 Id. at 392.81 Id.82 Id. at 390.83 Id. at 400.84 Id.85 Id. at 390.632 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615Nonetheless, the Court is clear that the broadcasters’ lack of editorial rights didnot entail regulatory carte blanche for the FCC. The agency was not free to “vindi-cate its own idiosyncratic conception of the public interest or of the requirements offree speech.”86 It was therefore critical that the Court was offered no evidence of theCommission’s:[R]efusal to permit the broadcaster to carry a particular programor to public his own views; of a discriminatory refusal to requirethe licensee to broadcast certain views which have been deniedaccess to the airwaves; of government censorship of a particularprogram contrary to [the Telecommunications Act]; or of theofficial government view dominating public broadcasting. Suchquestions would raise more serious First Amendment issues.87In other words, the FCC was fine so long as they simply required coverage of certainissues from all sides without discriminating against any of those sides.In subsequent cases, however, the Court declined to extend the holding of RedLion beyond broadcast media, on the grounds that these other media sources werenot physically scarce. The Court never again analyzed concentration in a mediamarket as a basis for the lifting of editorial rights. Just four years after Red Lion inMiami Herald Publishing Co. v. Tornillo, the Court held that Florida was blockedfrom applying a fairness doctrine-like policy to any newspapers, because of newspa-pers’ editorial rights.88 Despite Red Lion, in Turner I two decades later, the Courtrecognized the editorial rights of cable television companies because cable lackedthe “unique physical limitations of the broadcast medium.”89 Moreover, the Courtin that case was unequivocal that Red Lion was decided based on those physicalfactors and not, as the FCC had argued, on what it called “market dysfunction” ina speech market.90 While lower courts have held that internet service providers, likebroadcasters, do not have editorial discretion rights, the conclusion is based not ona market concentration rationale but on the assumption that these companies aremerely conduits for the speech of others that do not actually engage in speech.91Marsh and Logan Valley represent the Court’s (fleeting) willingness to impose con-stitutional free speech duties on private corporations that controlled vital opportunities86 Id. at 395.87 Id. at 396.88 418 U.S. 241, 258 (1974).89 Turner I, 512 U.S. 622, 637 (1994).90 Id. at 639; see also id. at 640 (“[T]he special physical characteristics of broadcasttransmission, not the economic characteristics of the broadcast market, are what underliesour broadcast jurisprudence.”).91 See, e.g., Mozilla Corp. v. FCC, 940 F.3d 1, 65–66 (D.C. Cir. 2019).2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 633for engaging in public speech within a community. Red Lion stands for the Court’s(again, fleeting) willingness to step back and allow Congress to impose constitutional-like free speech duties on private corporations that served as gatekeepers over someof the highest-visibility sources of news in the nation—even when those dutiesappeared to conflict with those corporations’ own constitutional free speech rights.All of these are exceptional cases, and not just in the sense that their precedential sunhas set. They all represent situations in which private corporations posed genuinestate-like threats against free speech interests. The fact that, even in a few cases, theCourt interpreted constitutional rights and duties to protect against those threats,whomever posed them, suggests at the very least that the First Amendment orderrests not exclusively on a concern about states as such but on a concern about state-like power over speech. Understandably, these concerns ordinarily line up. But whenthey come apart, courts are faced with a genuine, internal First Amendment conflict.The remainder of this Article defends more systematically orienting free speechdoctrine around the concern for state-like power and offers a glimpse of how it couldbe done. Unfortunately, the deviant cases from this Part are too few and too ab-stractly written to give us a unified account of state-like power over speech, muchless of the sort of duties that might come with that power. The next two Parts do thatdifficult theoretical work.II. THE THEORETICAL CASE FOR BINDING STATE-LIKE PRIVATE AGENTSThis Part argues that we have no principled reason for binding only state agentsby First Amendment speech rights.92 To be clear, I am not arguing that we have nomore reason to bind the state than any private agent. I contend only that the rigidcontemporary First Amendment public-private line over which no free speech claimsmay cross is indefensible in principle.I am, somewhat unusually in an article on law, excluding stare decisis as a prin-cipled reason, i.e., that we have always done it this way. (I will even here set asidethe argument that we in fact have not always done it this way, as Part I illustrated.)My inquiry concerns whether we have a justification for our practice.My contention is that the only principled reason we have for binding stateagents at all requires binding more than state agents. I first make the case that theonly affirmative justification we have for binding the state itself by any constitu-tional right is the state’s power over specific fundamental interests that the right issupposed to serve. I dismiss as implausible justifications based on the text, originalintent, and the nature of a constitution. I then explain that, in the context of speechrights, our justification for binding the state—its specific power over the realization92 I deliberately set aside practical reasons. I doubt that it is seriously impracticable tobreach the public-private barrier because it has been done before in other jurisdictionswithout any social calamity, but I am analyzing the principles that motivate our doctrine.634 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615of the free speech interests—does not justify binding it alone. Indeed, two types ofprivate agents might, at least in principle, have structurally similar power over theseinterests. These are the promised quasi-state agents. They will in certain respectsresemble Gulf Shipbuilding Corporation or CBS in the 1960s—but perhaps witheven more power.This Part thus establishes the prima facie case for imposing some sort of FirstAmendment duties on quasi-state agents. But the case is only prima facie: that is, Ioffer an affirmative justification for imposing these duties. Yet I do not claim, norcould I, that in all respects state agents and quasi-state agents are identically situ-ated. Thus, there might still be countervailing reasons for not applying these dutiesto quasi-state agents in the usual way, to the usual degree, or even at all. I postponean exploration of these countervailing reasons until Part III.A. Why Bind the State by Constitutional Rights?I have by now thoroughly set up the consensus assumption that individual rightsin the Constitution generally create duties for government agents only. Going for-ward, I will call this principle the state-only-duties principle. Can the principle bejustified?A justification cannot be found in the text, which barely hints at whom it binds.The constitutional Preamble, the introductory statement of the document’s purpose,makes no mention of any duty holder.93 Most of the rights provisions themselvessimply grant a “right of the people” (e.g., the “right of the people to keep and beararms” or “the right of the people . . . against unreasonable searches and seizures”)—without specifying who must respect the right.94 When a rights provision does specifya duty bearer, that bearer is a government agent. The First Amendment names Con-gress as the agent with the duty to respect the liberties of speech and religion.95 TheFourteenth Amendment, too, binds a specified set of agents: the states.96 But these ex-ceptions are far from conclusive, especially because they specify different state agents.The state-only-duties principle matches the proposition, occasionally expressedin American judicial opinions, that a constitution, by its nature, limits governmentaland not private power.97 In one document, on this account of constitutions, thePeople both delegate power to the government and limit it. As a historical matter,93 U.S. CONST. pmbl.94 The bulk of the exceptions come after the Bill of Rights were passed, with the Four-teenth Amendment and the various voting rights amendments. Several provisions in the Billof Rights concerning the criminal process could not be read as binding anyone other than thestate insofar as the state is the only entity that conducts any criminal process. See U.S.CONST. amends. II, IV.95 U.S. CONST. amend. I.96 U.S. CONST. amend. XIV.97 See, e.g., Southcenter Joint Venture v. Nat’l Democratic Pol’y Comm., 780 P.2d 1282,1287 (Wash. 1989).2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 635it is certainly true that the primary subject of all constitutions is governmentalpower. Yet no justification is obvious for why a constitution could not also grant orlimit power to private agents. Other peoples have done so.98 Indeed, as previouslyobserved, the American people99 themselves enshrined the right against enslavementand involuntary servitude against all actors, state or private, back in 1865.100 It is alsounclear that in the eighteenth century a long enough tradition of constitution-writingexisted for anyone to have a clear conception of what one—in general—was. Whycouldn’t a constitution be conceived more simply as a set of fundamental, difficult-to-alter rules, institutions, and values that the People choose to govern their society?Perhaps the best answer is that the American People at the Founding intendedto limit the rights in their Constitution to rights against the government. History makesclear that the Founders feared, above all else, a centralized national government.101When they spoke of the Bill of Rights, they described them as binding a potentiallytyrannical state.102 Yet this is decisive only on a rather cramped view of originalism,sometimes known as application originalism.103 On this view, we should think oforiginal intent as the Founders’ concrete “‘assumptions and expectation[s] about thecorrect application’ of their principles . . . .”104 It is much harder to know how theFounders’ broader principles as originally construed would apply to our modernmedia reality.105We could alternatively, consistently with everything just said, think that theFounders intended to bind the federal government just because it happened at the timeto be the most powerful agent around—the agent most in need of constraint. Indeed,the U.S. constitutional tradition adapted over time in this direction. Initially, the people98 See, e.g., India Const. art. 24 (“No child below the age of fourteen years shall be em-ployed in any factory or mine or engaged in any other hazardous employment.”); CONSTITUI-ÇÂO FEDERAL [C.F.] [CONSTITUTION] art. 40, § XVI (Braz.) (providing that National Congressshall have exclusive powers “to authorize, in Indian lands, the exploitation and use of hydricresources and the prospecting and mining of mineral resources.”).99 Or rather, the constitutional proxy for the people: the state legislatures elected by thepeople at that time eligible to vote.100 U.S. CONST. amend. XIII.101 See, e.g., Jud Campbell, Natural Rights and the First Amendment, 127 YALE L.J. 246,266–67 (2017).102 Cf. id. (citing the Founders’ call for a bill of rights so that a “Check will be placed onthe Exercise of . . . the powers granted”).103 See Mitchell N. Berman, Originalism Is Bunk, 84 N.Y.U. L. REV. 1, 28 (2009) (notingthat “the only commentators who take [this originalism variant] seriously are those aimingto attack it”).104 Michael W. McConnell, The Importance of Humility in Judicial Review: A Commenton Ronald Dworkin’s “Moral Reading” of the Constitution, 65 FORDHAM L. REV. 1269,1284 (1997).105 See generally, e.g., JACK M. BALKIN, LIVING ORIGINALISM (2011) (arguing that fidelityto original meaning still leaves enough flexibility to confront technological developmentsfrom after the founding).636 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615feared the power of a federal government over individuals (and the new states). Butover time, the threat that state governments could pose to individual rights becameclearer and was curbed in 1868 by the ratification of the Fourteenth Amendment,which ensured that state governments, and not just the federal government, guaran-tee equal protection, due process, and the privileges or immunities of citizens.106 Yet,what if we have now reached an era in which a new type of private agent can posethe same threat to individual rights that a state government could? Might the draftersand ratifiers of the Bill of Rights or Fourteenth Amendment think that the constitu-tional rights they developed were meant to be used against such a threat?107But perhaps the justification for binding the government by constitutional rightsis its extraordinary power. The central state at the Founding had the power to passlegislation binding across the nation, enforce that legislation against citizens, tax, andraise armies.108 The state is only more powerful today, with its massive armies, tanks,and missiles.109 We might finally have arrived at a justification for binding only thestate: for a state, to use the classical definition, has a monopoly on the legitimate useof force within its jurisdiction.110 It is always going to be the (by far) most powerfulagent around.Yet the rights provisions in the Constitution do not bluntly dilute, impede, andslow down power, the way the structural provisions do. They are more scalpel-like,limiting the exercise of power in specific ways. The guaranteed rights, it is generallyaccepted, are singled out among all other possible rights based on their importance tocertain fundamental interests of individuals and society. Arguably they are meant tolimit government’s power to impinge on those specific interests. For instance, the rightagainst unreasonable searches in the Fourth Amendment is justified as ensuring pri-vacy; the right to bear arms is justified as ensuring self-defense.111 The power the statewould use to impinge on those interests might be significantly less than its total power.In order to determine if any other constitutional right could apply to privateagents, therefore, one would need to conduct an inquiry for each right. One wouldfirst need to identify the fundamental interests that the right is supposed to serve andthe sort of threat the government poses to those interests. For some rights, like the106 U.S. CONST. amend. XIV, § 1.107 See, e.g., John Adams, A Dissertation on the Canon and the Feudal Law, BOS.GAZETTE, May 21, 1765 (describing the people’s natural right to knowledge and the utmostimportance of the “preservation of the means of knowledge”); Campbell, supra note 101, at268–70 (describing the Founding-era belief in a natural right to freedom of expression).108 U.S. CONST. art. I, § 1; id. art. I, § 7, cls. 1, 11–15.109 Cf. MAX WEBER, POLITICS AS A VOCATION 2 (1919) (explaining that a state’s abilityto enforce laws within its territory is contingent on its legitimate and exclusive ability to useviolence).110 See id.111 Cf. Aharon Barak, Constitutional Human Rights and Private Law, 3 REV. CONST.STUD. 218, 229 (1996) (explaining the argument that human rights protect dignity, anddignity can be just as easily harmed by private as by public agents).2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 637trial rights in the Fourth, Fifth, and Sixth Amendments, it is readily apparent that thestate is the only agent that poses the relevant threat. For others, the answer may beless clear. Below, I sketch the inquiry for free speech and press rights.B. Why Bind the State by Free Speech Rights?On the traditional theories, the First Amendment’s Speech Clause is justified aspromoting one of three fundamental interests: autonomy, democracy, or knowledge(often described as truth112).113 Under autonomy theories, free speech—encompassing,whenever I use the phrase, the ability to both speak freely and to hear what othershave freely to say—serves individuals’ interests in expressing themselves or inexploring and developing their intellect and character.114 Under democracy theories,free speech serves individuals’ interests in political participation or in living in ademocratic society where voters are informed and elections are legitimate.115 Underknowledge theories, free speech allows for truth to be tested and sorted fromfalsehood, which serves individuals’ interests in having true justified beliefs and inliving in a society where knowledge is growing and widely disseminated. Mostcontemporary free speech scholars are pluralists who believe that First Amendmentspeech rights serve all of these interests, and possibly additional ones.116But where is the justification for protecting speech rights against the state inparticular? So far, each of these theories only asserts that a certain general free flowof speech promotes autonomy, democracy, and knowledge. After all, the flow ofspeech can be constrained by all sorts of interventions. When someone tells me to“Be quiet!,” this may, depending on the speaker and context, stem the flow of myspeech and may, depending on what I have to say, undermine my autonomy, thedemocratic process, and the dissemination of knowledge. Yet few would argue thatthis intervention unacceptably impinges free speech interests. What is it about gov-ernment interventions with speech, in particular, that are unacceptable?112 See, e.g., Joseph Blocher, Free Speech and Justified True Belief, 133 HARV. L. REV.439, 441 (2019).113 This is a simplification, but in line with the ordinary way of talking about free speechtheory.114 See generally, e.g., Seana Valentine Shiffrin, A Thinker-Based Approach to Freedomof Speech, 27 CONST. COMMENT. 283 (2011); C. EDWIN BAKER, HUMAN LIBERTY ANDFREEDOM OF SPEECH (1989); Martin H. Redish, Self-Realization, Democracy, and Freedomof Expression: A Reply to Professor Baker, 130 U. PA. L. REV. 678 (1982); Thomas Scanlon,A Theory of Freedom of Expression, 1 PHIL. & PUB. AFFS. 204 (1972).115 See ALEXANDER MEIKLEJOHN, POLITICAL FREEDOM: THE CONSTITUTIONAL POWERSOF THE PEOPLE 10–14 (1960); Robert Post, Participatory Democracy and Free Speech, 97VA. L. REV. 477, 479–82 (2011).116 See, e.g., Leslie Kendrick, Are Speech Rights for Speakers?, 103 VA. L. REV. 1767, 1788(2017); FREDERICK SCHAUER, FREE SPEECH: A PHILOSOPHICAL ENQUIRY chs. 2–6 (1982);Joseph Blocher, Nonsense and the Freedom of Speech: What Meaning Means for the FirstAmendment, 63 DUKE L.J. 1423, 1441–56 (2014).638 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615Free speech theorists are relatively silent on this question, perhaps because it issimply assumed that the only duty holder with respect to free speech rights is thegovernment. But the answer seems straightforward: state officials’ unusual powerover the channels of communication allows them to staunch the flow of speech toan exceptionally high degree, and their motives to maintain their ruling status mayencourage abuses of that power. The state can harm the flow of speech in two waysthat are critical to the triad of free speech values: (1) it can cut off most of an individ-ual’s avenues for speaking or listening, such that it is extremely difficult for anindividual to express themselves or be heard by an audience aside from their closestassociates, or (2) it can advantage or disadvantage the dissemination of certain ideas,information, or views at such scale so as to distort public discussion across societyand, ultimately, public opinion. The first way of staunching the flow hurts any of themore individualistic framings of the free speech values: the individual’s ability toform themselves, the individual’s participation in the democratic process, or theindividual’s ability to form justified true beliefs. The second way of staunching theflow hurts the more consequentialist framings of the free speech values: society’sculture of free inquiry and tolerance; democracy’s legitimacy, or the informed natureof its decision-making; or the building and widespread dissemination of a societalbody of knowledge. Theorists might disagree about exactly which degree of en-croachment is unacceptable, but nearly all can agree that the degree to which thestate is capable is.So, we have our more narrowly drawn justification for binding the state by freespeech rights. Is it unique to the state? It might be, if state agents are the only sortthat pose this extreme threat to free speech interests. The state certainly seemsuniquely able to control the channels of communication, with its riot police andtanks. But if it turns out that some private agent can threaten free speech interestsas effectively, or nearly as effectively, as the state does, then the case for the state-only-duties principle falters. Indeed, failing to apply the First Amendment againstthis private agent might actually thwart the purposes of that Amendment. Accordingto the free speech theories mentioned above, the First Amendment plays a pivotalrole in noble projects such as the development of the self, the legitimacy of elec-tions, and the education of citizens. But if a private party controls these processesand their outcomes as tightly as the state could, then, no matter how far the statesteers clear of speech, these projects would crumble. The First Amendment wouldshield the freedom of speech from one attacker, only to step back and allow theattacker standing in the wings to step in.Below I examine whether the state truly is unique in its power over speech andits motives to abuse that power. Before I do, I want to acknowledge that my pursuitof state-like private power involves a proxy. If what the First Amendment is aboutis achieving certain free speech goals, then there is no reason to think that the powerthat threatens those goals rises even close to the power of the state, whatever thestate’s power happens to be. The threatening level of power might be significantly less2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 639than that of the state.117 But articulating an independent yardstick for measuringexactly when power triggers constitutional duties would be extraordinarily difficult,not least because it might vary based on one’s preferred free speech theory. Reasoningby analogy therefore offers the advantages of being relatively clear and ecumenical—as well as capturing the summits of private power, such as the Machine. As the Su-preme Court of New Jersey said in applying its free speech clause against a regionalshopping center, we “cannot determine precisely the extent of damage to free speechthat will call forth our constitutional provision to prevent it, but precision is notrequired in this case: the damage is massive.”118An alternative model might link private agents’ constitutional duties to respectfree speech to a continuum of power: the more power an agent has, the more strin-gent its free speech duties.119 But such an approach would be much harder toimplement than the one suggested here. Courts would have to determine a widerange of duties for private agents, based on their exact level of power. I expect thatit is similarly for the practicalities of implementation that courts have always appliedfree speech duties equally to all state agents, irrespective of their exact degree ofpower (for surely governmental entities and officials vary in this respect). Moreover,a continuum model, insofar as it did impose some duties on private agents withrelatively low magnitudes of power, might contribute greatly to hardship and un-certainty for individuals, as I detail in Part III.C. Is the State’s Power Over Speech Unique?No private agent could match a functioning state’s total power over the channelsof communication in a society, given the latter’s police power. At least any privateagent other than the science-fictional Machine. But certain rare private agents mighthave power that poses a comparable threat to free speech. This possibility is gradu-ally made more realistic by a closer examination of state power over speech, whichturns out to be both: (1) excessive and (2) less than it appeared. First, I show that thestate may be able to achieve grave impediments to free speech interests while onlyutilizing part of its power. In other words, the state has excess power with respect117 For instance, one might believe that the enshrinement of certain constitutional rightsmeant that they were guaranteed, simpliciter, not just that the government could not infringethem. Cf. The Civil Rights Cases, 109 U.S. 3, 46–49 (1883) (Harlan, J., dissenting) (arguingthat the Privileges or Immunities Clause was meant to grant constitutional rights to everycitizen and not only to prohibit states from infringing those rights).118 N.J. Coal. Against War in the Middle E. v. J.M.B. Realty Corp., 650 A.2d 757, 779(N.J. 1994).119 I take Genevieve Lakier and Nelson Tebbe to have suggested something analogous tothis model in a recent online essay. See Lakier & Tebbe, supra note 9. Chemerinsky alsoproposed a similar model in the 1980s for private duties to respect all constitutional rights,and not just speech rights. See generally Erwin Chemerinsky, Rethinking State Action, 80NW. U. L. REV. 503 (1985).640 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615to speech. Second, I contend that the state’s power is subject to many unique con-straints and limitations. When we couple the excess nature of the state’s power withthe constraints on its power, it becomes less far-fetched that a private agent couldachieve similar results.To begin, the state’s power over speech is not a monolith, but, like all power,has different discernable dimensions. I discuss three. The first, and perhaps mostobvious, is the power’s strength, or the likelihood that an exertion of power overspeech will in fact achieve the desired results: that is, that the speaker(s) will speakwhat the state wishes (perhaps silence) and the listener(s) will hear what the statewishes (again, perhaps silence). Persuasive abilities might afford very weak powerover speech; a gun, or blackmail, might afford very strong power over speech. Thedensity of power over speech is the range of potential speech acts to which the powerextends. Say that a gunman follows a speaker around and threatens them each timethey try to speak their message, whispering it, shouting it, typing it, broadcasting it,painting it onto a canvas. Such a gunman has power of not only great strength, butalso great density. Lastly, the breadth of power over speech is the number of speakersand listeners subject to the power. So, the gunman following just one speaker aroundmight have a relatively low breadth of power—unless that speaker is routinelytalking to vast crowds of people. Each of these three dimensions of power contrib-utes to achieving the powerful agent’s desired results with respect to speech.The state’s power over speech is extraordinary in terms of its strength, density,and breadth. To begin, the strength of the state’s power (over any action, includingspeech) is unparalleled because of its control of physical force, which in turn grantsit nearly endless resources and a massive network of agents for executing its will.By using or threatening the use of force, the state is able to coerce people into speak-ing or not speaking, hearing or not hearing. It can use its police power to shut downwebsites, presses, or public squares. Or it can deter people from using those meansof communication by attaching to their use fines or prison sentences. That deterrenceis reinforced by the state’s surveillance apparatus, made possible by the resourcesmentioned above.For similar reasons, the state’s density of power over speech is also extraordi-nary because it may deploy force over almost any of the channels of communicationwithin its jurisdiction. If protestors stand on one corner, then the state may run themoff that corner—and the next, and the next, and the next. Due to the state’s surveil-lance capacities, it might then, like the gunman described above, effectively followthem around and censor them over phones, over the internet, or on private property.In other words, the state can police and obstruct almost any channel of communica-tion that a speaker (or listener) might want to utilize.The breadth of the state’s power is no less extraordinary. A state can use forcethroughout its territory, and therefore, over the potentially huge number of speakersand listeners located there. The state can, for instance, credibly forbid all people inits territory from speaking a certain message. Even if the state cannot enforce this2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 641prohibition against everyone, the fear that it will often suffices to achieve very highcompliance across a wide swath of its population. The state could also affect mul-titudes of speakers and listeners simply by closing off a widely used means ofcommunication, like a popular television station or telephone service. India, forexample, frequently shuts down internet access to prevent mass protests.120Recall above that I mentioned two specific threats that the state poses to freespeech: (1) it can cut off most of an individual’s avenues for speaking or listeningor (2) it can advantage or disadvantage the society-wide dissemination of certainideas, information, or views so as to tip the balance of public opinion. The first sortof threat matters most for individualistic free speech theories, like autonomy-basedones. The second threat matters more for theories, like those that are democracy-based and truth-based, that foreground larger-scale societal consequences.121 Asmentioned above, contemporary free speech scholars are usually pluralists aboutfree speech values and see all three of the standard values as playing a role in whywe protect free speech.122 I therefore submit that we would have reason to constrainthe state if it was powerful enough to pose either of these threats.Yet the state may be able to pose each of these types of threats with power thatis high along only two of the three dimensions identified: strength, plus one of theothers.123 With respect to thwarting individual’s speech opportunities, the state mightsingle out a small minority and cut off all avenues of communication for them andonly them. The state would be exercising power of a high strength and density, butnot a high breadth. Yet they would be clearly posing the first grave threat to freespeech. With respect to the second grave threat, the state might tilt public opinion inits favor just by forcing the top one or two media outlets to suppress anti-governmentmessages. It would thereby exercise power of a high strength and breadth, but nota high density, because it would only be affecting a couple of channels of communi-cation. While the state would increase its total impact on public opinion by clampingdown on additional media outlets and preventing people from gathering on the streetsfor anti-government protests, these steps might not be necessary to tip opinion in itsfavor. In other words, the state has excess power relative to the free speech dangersit poses, because it does not need all dimensions of its power to pose them.On the other hand, the state faces constraints and limitations on its use of force,personnel, and resources to manipulate speech, particularly in a society like the120 See Anand Katakam, India Leads the World in Internet Shutdowns, REUTERS (Dec. 20,2019, 11:28 AM), https://www.reuters.com/article/india-citizenship-internet/india-leads-the-world-in-internet-shutdowns-idINKBN1YO1WR [https://perma.cc/WF6N-AWTD].121 Both threats are of course overlapping, insofar as interferences with an individualspeaker’s and listener’s opportunities will affect the flow of ideas, and vice versa.122 See sources cited supra note 114.123 An agent might also trade off among these dimensions of power. One might make up forhaving slightly less strength of power than the state by having more breadth or more density.642 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615United States with a strong democratic and free press culture. There are severalreasons for this. The first is the state’s highly imperfect detection of noncomplianceat scale. While the state might be able to single out and deter a particular individualfrom speaking anywhere, it would struggle to similarly deter the entire population.It only has so many personnel available to assist in enforcement, and only so manyinstallations of surveillance technology. Even authoritarian countries cannot, despitetheir best efforts, fully contain dissent. The second is that state actions to silence,twist, or compel speech can predictably produce public backlashes and, in democra-cies, electoral rebukes that deter the state from acting in the first place. The state isan extremely, and perhaps uniquely, transparent agent, especially in a highly digi-tized society. Its most forceful interventions in speech are widely noted and hard tohide. Throwing many people, or even a small number of high-profile people, in jailfor speech tends to get noticed, and to cause an uproar. The third is the law, whichmay explicitly limit the state’s power. For instance, the Fourth Amendment wouldlimit American government agents’ ability to follow people into their homes andlisten to their speech.124 The fourth is that many states are not univocal actors: onepart of the government may prevent another part from acting. In the United States,for instance, checks and balances might deter any severe abuses of speech by asingle power-hungry branch.Given that the state’s power over speech is both excessive and limited relativeto the free speech threats it poses, it seems more plausible that a private agent mightpose at least a comparable threat. Indeed, I argue that there are two categories of suchspeech-threatening private agents. These quasi-state agents have either (a) highstrength and high density of power over speech; or (b) high strength and high breadthof power over speech. Each type of quasi-state agent will pose a different type ofthreat to free speech. In the case of high density-but-not-breadth of power actors, theprimary threat is to individuals’ speech opportunities. In the case of high breadth-but-not-density of power, the primary threat is to the larger system of public dis-course (and elections). Sometimes quasi-state agents of either type will actually posea greater threat to free speech than can the state.1. Dense-and-Strong-Power Quasi-State AgentsThe government has power over speech that is both exceptionally strong andexceptionally dense: it can altogether shut off many avenues of communication, andit can do this for most if not all avenues of communication. A private agent withsimilar strength and density of power over speech across an entire population is per-haps science fiction. But a private agent might gain power of comparable strengthand density over a smaller number of speakers or listeners. This may be because theagent controls a group or institution to which members are tightly bound, perhaps124 See, e.g., Silverman v. United States, 365 U.S. 505, 511–12 (1961).2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 643because the group or institution provides resources crucial for life. I have in mindcases such as certain private employers, nursing homes, or universities.Consider first the strength of power. Even though private agents lack the state’sability to employ physical force, some will be able to coerce individuals with a highrate of success by threatening deprivation of a vital benefit that they control. Ofcourse, the strength of power thereby achieved is not as high as the state’s—a privateagent cannot physically shut off means of communication or guarantee their aban-donment. But, if they can effectively determine—via robust incentives or disincen-tives—how an individual uses an avenue of communication, then their strength ofpower over that avenue is comparable to the state’s. Indeed, as explained above, thestate may have excess power for this purpose.Many private employers, for instance, could easily hit this strength benchmark.Employers can threaten penalties for breaking speech rules that, for most people,carry extraordinary motivational force: demotion, or even job loss. A threat oftermination may be nearly as effective as a threat of physical violence or jail time,insofar as the employee’s general long-term well-being (and perhaps that of herfamily) largely depends on her job. Indeed, if she is ill or disabled, her very healthand life may depend on the continued medical insurance coverage that her job pro-vides. One theorist of the workplace describes the penalty of job loss as “exile.”125 Ofcourse, even job loss may not be a strong enough motivation to guarantee the em-ployee’s compliance if she can easily find another job. But in industries with nounions and a high supply of laborers but few jobs in the labor market, employees mayhave no realistic option to quit and look for another job and potential employers mayanyway have little incentive to woo new employees with reasonable speech policies.The fact that the state provides a “higher authority” above a powerful privateemployer provides little consolation where there is no law. Only about half ofAmerican workers in the private sector are protected from any employer controlover their speech off the job through state and local legislation,126 and many of theseprotections apply only to narrow classes of speech.127 While federal statutes shieldsome private employee speech as well, they are also restricted to a few types ofspeech such as whistle-blowing and union activity.128 Indeed, property law enforced125 See, e.g., ELIZABETH ANDERSON, PRIVATE GOVERNMENT: HOW EMPLOYERS RULE OURLIVES 38–39 (2017); Hélène Landemore & Isabelle Ferreras, In Defense of WorkplaceDemocracy: Towards a Justification of the Firm-State Analogy, 44 POL. THEORY 53, 55 (2016).126 Eugene Volokh, Private Employees Speech and Political Activity: Statutory ProtectionAgainst Employer Retaliation, 16 TEX. REV. L. & POL. 295, 297 (2012).127 Id.; see also Genevieve Lakier, The Non–First Amendment Law of Freedom of Speech,134 HARV. L. REV. 2299, 2357–58 (describing various state protections for a narrow rangeof employees’ expressive acts, such as political speech, political activity such as voting orrunning for office, and union activity); cf. ANDERSON, supra note 125, at 53 (observing thatin the U.S. private employers have sweeping authority over employees’ off-duty conduct,with limited exceptions such as labor union activity).128 See Cynthia Estlund, Can Employees Have Free Speech Rights Without Due ProcessRights (in the Private Sector Workplace)?, 2 J. FREE SPEECH L. 259, 260–63 (2022).644 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615by the state would back up employers in disputes with employees: if an employeefired for her unprotected speech shows up at work, she could be removed by thepolice for trespassing, at the employer’s behest.Private entities with such strength of power over the speech of their dependentsmay also, sometimes, have high density of power, too—if they can surveil many oftheir dependents’ means of communication, both inside and outside the areas theycontrol. Gulf Shipbuilding is an exemplar, because it owned both the workplace andthe homes of its employees in Chickasaw.129 But even a more typical employer canmonitor employees’ speech across the workday, by installing recording devices, moni-toring electronic communications that use the firm’s software, or even relying on thereports of coworkers (under the influence of a similar power).130 Beyond the work-place’s walls, an employer can also monitor employee posts on social media andkeep an eye out for employee names in high-visibility media.131 While informal re-search suggests that the cause of most firings for social media activity are racist andother discriminatory posts, there are also numerous examples of firings for expressingone’s sexual orientation or participating in artistic or political movements.132Virtually no employer will have the motivation, much less resources, to monitoremployees’ use of every means of communication. Phone calls, digital instantmessaging, lower-visibility public spaces or media, and the home will presumablybe beyond their reach. These entities may nonetheless make it perilous and exhaust-ing for their employees to express themselves as they search out safe means ofcommunication. Moreover, if these private methods of communication were the onlyones available to an employee, they would be altogether inadequate for certain typesof speech, such as advocating for union organization, protesting employment con-ditions, or any form of political speech designed to reach large audiences. Yet some129 See Marsh v. Alabama, 326 U.S. 501, 502 (1946).130 See The Daily, The Rise of Workplace Surveillance, N.Y. TIMES (Aug. 24, 2022), https://www.nytimes.com/2022/08/24/podcasts/the-daily/workplace-surveillance-productivity-tracking.html [https://perma.cc/DP9M-RQ78]; cf. ANDERSON, supra note 125, at 39–40.131 See, e.g., Joe Gagliese, Your Employees Are On Social Media, You’re Right To BeWorried: 5 Worst-Case Scenarios, FORBES (Nov. 2, 2022, 7:00 AM), https://www.forbes.com/sites/theyec/2022/11/02/your-employees-are-on-social-media-youre-right-to-be-worried-5-worst-case-scenarios/?sh=1a7e55be1752 [https://perma.cc/8LCJ-VP2E]; From Insta-gram to Insta-Fired: 86% of Canadian Companies Would Fire Employees for InappropriateSocial Media Posts, FIN. POST (Jan. 11, 2023), https://financialpost.com/globe-newswire/from-instagram-to-insta-fired-86-of-canadian-companies-would-fire-employees-for-inappropriate-social-media-posts [https://perma.cc/W2WX-LUYW].132 Brady Robards & Darren Graf, Who Really Gets Fired Over Social Media Posts? WeStudied Hundreds of Cases to Find Out, THE CONVERSATION (June 15, 2022, 10:24 PM),https://theconversation.com/who-really-gets-fired-over-social-media-posts-we-studied-hundreds-of-cases-to-find-out-182424 [https://perma.cc/NV2S-LHKN]; see also Joseph Goldstein,NYU Langone Fired Him for His Posts on the Mideast War. He’s Suing., N.Y. TIMES(Nov. 24, 2023), https://www.nytimes.com/2023/11/24/nyregion/nyu-langone-cancer-doctor-fired-lawsuit.html [https://perma.cc/WR93-HKXM].2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 645theorists describe this sort of public speech as the most valuable speech under theFirst Amendment.133One can imagine, for instance, a secretary who posts on social media about acontroversial political topic. She may be friends with several of her colleagues in theoffice, including her boss, and her post might be noticed. Some of those colleaguesmay then start to complain about working with her because of her political views,though those views have nothing to do with her ability to carry out her job responsi-bilities, including her ability to interact respectfully with other members of the of-fice. She may then be fired to placate other workers, or because the boss’s views donot match her own. In the actual world, employees are not uncommonly pressuredto refrain from—or even to engage in—certain types of political or religious speech.134On the listener’s side, an employer may also have significant opportunities tomonopolize the information that their employees receive about policy questions andelections. Often employers will circulate emails to their employees recommendingparticular candidates for election, for the good of the business.135 If an employee isgenerally politically apathetic, these emails can carry great weight. In addition, theymay chill employee speech that runs counter to the employer’s expressed views.Private entities may have even more objectionable power over individuals’speech than governments do, because the former lack both the pressures towardtransparency and the democratic accountability to which the latter is subject.136 Atthe same time, constraints on their power are not altogether absent. A firm’s treat-ment of its employees, for instance, may affect the willingness of other firms, orconsumers, to do business with them. However, unlike government, private agents—even large companies—are often able to act under the radar. Individual firings andundignified work conditions are rarely picked up by the media unless they involvehigh-profile firms; the threat of firing will be covered much less often still. While133 See e.g., Robert Post, Meiklejohn’s Mistake: Individual Autonomy and the Reform ofPublic Discourse, 64 U. COLO. L. REV. 1109, 1109 (1993).134 See, e.g., Charlotte Garden, Was It Something I Said? Legal Protections for EmployeeSpeech, ECON. POL’Y INST. (May 5, 2022), https://www.epi.org/unequalpower/publications/free-speech-in-the-workplace/ [https://perma.cc/VMM3-MVTL] (describing cases, inter alia,in which workers were pressured to join a Trump rally, fired for having a Kerry-Edwardsbumper sticker on their car, or raising concerns about COVID-19 protections in the work-place); Christopher Spata, Can You Be Fired for Protesting? In Florida, You Can, TAMPABAY TIMES (June 11, 2020), https://www.tampabay.com/news/2020/06/11/can-you-be-fired-for-protesting-in-florida-you-can/ [https://perma.cc/ZTN3-XB4M] (relating (sometimes con-tested) claims by workers to be fired for participating in or posting about (positively ornegatively) the 2020 racial justice protests).135 See, e.g., Allie Robbins, Note, Captive Audience Meetings: Employer Speech vs. Em-ployee Choice, 36 OHIO N. U. L. REV. 591, 591 (2010).136 See ANDERSON, supra note 125, at 45 (“Private government is government that hasarbitrary, unaccountable power over those it governs. This is of course a matter of degree.Its powers may be checked in certain ways by other governments, by social norms, and byother pressures.”).646 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615one might argue that social media is slowly shrinking the transparency gap betweenthe public and private spheres, the gap remains.137One can offer similar analyses for powerful organizations that dominate thelives or livelihoods of their members or residents, like, as mentioned above, nursinghomes and universities. Such groups or institutions may be hard to exit because theyprovide crucial resources for life; their leaders may actually find it easier to monitormembers than the state does because detection is easier on a smaller scale or withaccess to personal spaces. The more dependent the individual is on the institution,the more the institution looks functionally like the state. Yet the Supreme Court hasregularly insisted that such entities are not bound by the First Amendment becausethey do not serve functions traditionally exclusively performed by the state.1382. Broad-and-Strong-Power Quasi-State AgentsAnother set of private agents that might rival state power over speech are thosewith high strength and breadth of power over speech. These agents have power overthe way in which a massive quantity of speech is distributed across an entire juris-diction. It is this breadth of power that allows some quasi-state agents to mold andmanipulate public discourse and ultimately public opinion, just as the state—absentfree speech protections—could.139It might be tempting to think that certain politicians, celebrities, and influencershave very broad power over speech—relatively and absolutely—because they haveenormous standing audiences. Some of these individuals have millions of followerson social media who read their every post; when they are interviewed on nationaltelevision, millions of people tune in. If we think of a speech event as a speech actby a speaker that is heard by a listener, then these individuals control millions ofspeech events by simply deciding whether and how to speak, thereby changing whatmillions of people hear. I would not consider these individuals to be quasi-stateagents, however, because the density of their power over speech is typically trivial.These individuals control no one’s speech but their own. Put another way, they onlycontrol speech events involving their own speech. This matters for their impact onpublic discourse because they are just one (even famous) voice in a sea of other(often famous) voices and will struggle to truly manipulate public discourse unlessthey have something truly extraordinary to say. Considered holistically, then, this137 As noted above, federal and state laws also constrain some employer control ofemployee speech, but these restrictions are far from expansive. See supra note 127.138 See Manhattan Cmty. Access Network v. Halleck, 139 S. Ct. 1921, 1926 (2019).139 This Article focuses on private agents that have both relatively and absolutely broadpower over speech. That is, they have the power to reach many speakers and listeners withina large jurisdiction, like the United States. It may nonetheless be worthwhile considering, infuture work, quasi-state agents with broad power relative to smaller jurisdictions, such asGulf Shipbuilding or the only local newspaper in a small town.2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 647makes their power—broad as it is—no match for the state’s. A broad-power quasi-state agent need not have high density of power, but it needs more than trivialdensity. Typically, a broad-power quasi-state agent will control at least one whole,widely used channel of communication.140The only entities likely to achieve broad-power status are therefore owners oftruly mass media and telecommunications—those who serve up information to anexceptionally large segment (perhaps a majority) of the population. Mass mediacompanies specialize in one-to-many communication channels. They offer a smallnumber of people the opportunity to amplify their speech to massive audiences. Butthe companies remain the gatekeepers, deciding who those amplified speakers willbe. Consider a television or radio station with a massive audience, like CBS in the1960s when 29 million Americans regularly tuned into the nightly news with WalterCronkite.141 (By contrast, the top American television news programs today reachonly one or two million listeners.142) Mass telecommunications companies specializein many-to-one communication, or interpersonal communication at scale. They offera massive number of people the opportunity to connect with a small number ofassociates. Consider a telephone or internet company. Both mass media and tele-communications companies can affect a massive number of speech events each dayand therefore have the ability to manipulate public discourse at scale. Indeed, statesoften use the largest media and telecommunications companies within their territo-ries in order to achieve their own power over speech.143 They threaten and cajole theowners to shut down or manipulate their essential channels of communication inways that the state prefers.144These companies naturally lack the government’s density of power over speech.They can only affect speech conveyed over the channels of communication that theycontrol. A television station, for instance, cannot control anything that is communi-cated on other television stations, on radio stations, in newspapers, during in-personconversations, etc. But their density of power is still not trivial, insofar as theycontrol all the speech that is communicated over their channel.140 It is for similar reasons that most civic organizations, like churches, lack state-likebreadth of power over speech. These organizations may issue statements, and perhaps controlthe speech of their leaders, but they cannot control the speech of still others.141 See Karlyn Bowman, The Decline of the Major Networks, FORBES (July 27, 2009,12:01 AM), https://www.forbes.com/2009/07/25/media-network-news-audience-opinions-columnists-walter-cronkite.html?sh=2901235147a5 [https://perma.cc/377E-E8VX].142 See Cable News Fact Sheet, PEW RSCH. CTR. (Sept. 14, 2023), https://www.pewresearch.org/journalism/fact-sheet/cable-news/ [https://perma.cc/MR6R-J84Y].143 See, e.g., Daniela Stockmann & Mary E. Gallagher, Remote Control: How the MediaSustain Authoritarian Rule in China, 44 COMPAR. POL. STUDS. 436, 441–42 (2011) (describ-ing the use of media by the Chinese government to achieve regime stability); Katakam, supranote 120.144 Id.648 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615At exactly what point a mass communications company acquires state-likebreadth of power may be hard to pinpoint. But a company that is regularly able tocontrol speech events involving a majority of a population, as speakers and listeners,seems a good candidate. Such a company may be able to affect broad patterns inpublic discourse: which topics fade or surge in discussion, which evidence is thoughtto be trustworthy, which views are thought to be well-accepted.145 It will, in ameaningful sense, control public attention. It might even, by tilting speech in favorof some views and against others, be able to change public opinion at the margins.In the United States, no media organization is as poised to manipulate publicopinion as the owners of the major social media platforms and search engines. Thesecompanies’ power over what is said and heard is unparalleled in history in its abso-lute scale. One of the reasons is the sheer number of individuals who use their mediaservices. While perhaps no private agent can reach quite as many speakers andlisteners within a state as the state can, many big media companies still touch a veryhigh percentage of the total population. When the United States was founded, nosingle newspaper was able to reach a majority of residents.146 Today, many mediacompanies come close, and Meta and Google surely exceed the target. Nearly seven-tenths of the American population has Facebook accounts;147 40% has Instagramaccounts.148 Google, through YouTube, reaches 81% of the population.149 Its searchengine is used by 86%.150 These modern companies control the means of communi-cation that are used both to speak and to consume speech. Indeed, they are frequentlydescribed—arguably even by the Supreme Court itself—as the new public forums.151The reach of these platforms extends not only to those who directly use them butalso to those who hear second-hand of speech circulated on them. Even newspaperstories now frequently include screenshots of posts and comments from the topsocial media platforms.What all of this means is that the platforms can, often with a metaphorical flickof a switch, change a national conversation. Often, they can do this by using theiralgorithms simply to demote content. Take, for instance, the news story about Hunter145 Erin Miller, Media Power Through Epistemic Funnels, 20 GEO. J.L. & PUB. POL’Y 873,877–79 (2022).146 Charles G. Steffen, Newspapers for Free: The Economies of Newspaper Circulationin the Early Republic, 23 J. EARLY REPUBLIC 381, 381–82 (2003) (discussing the reach andgrowth of newspapers in early nineteenth century America).147 Brooke Auxier & Monica Anderson, Social Media Use in 2021, PEW RSCH. CTR.(Apr. 7, 2021), https://www.pewresearch.org/internet/2021/04/07/social-media-use-in-2021/[https://perma.cc/W2W4-YALU].148 Id.149 Id.150 Alexander Kunst, Most Used Search Engines by Brand in the U.S. as of June 2023,STATISTA (Aug. 25, 2023), https://www.statista.com/forecasts/997254/most-used-search-engines-by-brand-in-the-us [https://perma.cc/7RJK-FWVY].151 See Packingham v. North Carolina, 582 U.S. 98, 104 (2017).2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 649Biden’s abandoned laptop. Twenty days prior to the 2020 election, the New York Postran a story claiming to reveal contents from Joe Biden’s son’s laptop, includingemails that supposedly revealed corruption by the presidential candidate.152 Facebookand Twitter quickly disabled the option for their users to share it, ultimately citingconcerns that the story was based on hacked (and therefore unreliable) material.153These actions effectively suppressed the story not just on those platforms, but inpublic discourse more broadly. While the suppression was certainly no great loss topublic edification given that the claims of corruption were baseless, it raised thespecter that a snap judgment about journalistic “reliability” by the social media giantscould have had an impact on a rather close election. Indeed, hacking was not, in theend, the source of the story’s unreliability.154 Even the former CEO of Twitter, JackDorsey, criticized the decision.155These “big tech” companies also have blunter methods to change which topicsare publicly discussed, and when: they can outright block speech and speakers. Mostfamously, Meta, followed shortly after by Twitter, chose in 2021 to suspend a sittingAmerican president from its platforms after the January 6 riot at the Capitol.156Whether one thought that move justified to prevent violence or not, it exhibited anextraordinary power over public discourse. Nor is Trump’s the only case. Socialmedia platforms routinely make controversial decisions to deplatform or “blacklist”other individuals and groups.157 (While most of the largest social media companiesdo not appear to deplatform out of mere personal dislike, “X”’s owner Elon Muskhas shown the possibility of even this abuse of power.158) The platforms have,152 See Lauren Feiner, FEC Says Twitter Acted Lawfully in Restricting the New YorkPost’s Article on Hunter Biden, CNBC (Sept. 15, 2021, 5:06 PM), https://www.cnbc.com/2021/09/15/twitter-acted-lawfully-in-restricting-nypost-hunter-biden-article-fec.html[https://perma.cc/7AJJ-3SBM].153 Andrew Rice & Olivia Nuzzi, The Sordid Saga of Hunter Biden’s Laptop: The MostInvasive Data Breach Imaginable Is a Political Scandal Democrats Can’t Just Wish Away,N.Y. MAG. (Sept. 12, 2022), https://nymag.com/intelligencer/article/hunter-biden-laptop-investigation.html [https://perma.cc/KRX6-PX9P].154 Id.155 See Feiner, supra note 152.156 Brian Fung, Facebook Bans Trump from Posting for Remainder of His Term in Office,CNN (Jan. 7, 2021, 7:06 AM), https://www.cnn.com/2021/01/07/tech/facebook-trump-restrictions/index.html [https://perma.cc/D878-7GLD].157 Sam Biddle, Revealed: Facebook’s Secret Blacklist of “Dangerous Individuals andOrganizations,” THE INTERCEPT (Oct. 21, 2021, 1:16 PM), https://theintercept.com/2021/10/12/facebook-secret-blacklist-dangerous/ [https://perma.cc/NKD5-2KV3] (relating scholars’worries that white militant groups are treated more leniently, or less often listed at all, thanother violence-linked groups composed primarily of Muslims or people of color).158 See A. Martinez & Bobby Allyn, Twitter Owner Elon Musk Suspends the Accounts ofSeveral High-Profile Journalists, NPR (Dec. 16, 2022), https://www.npr.org/2022/12/16/1143330589/twitter-owner-elon-musk-suspends-the-accounts-of-several-high-profile-journalist[https://perma.cc/G2JZ-45K4].650 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615conversely, made decisions to treat certain “VIP” accounts—usually those of poli-ticians and celebrities—with kid gloves.159The concern is that these methods of influencing public discussion at scale canaffect not just what we read but also what we think, how we act, and how we vote.Research suggests that social media as it is currently operated has had impacts onrates of, among other things, political polarization, extremist radicalization, and evenbullying among youth.160 Any reasonably aware observer over the past decade canalso see how social media has changed our habits of attention and informationconsumption more generally.161 Definitive studies of the platforms’ influence overelections are scarce, owing in part to the difficulty of any such causal analysis. Yeta Facebook executive in 2020 opined in a leaked internal email that the companyhad “tools available” to them to change the outcome of the upcoming presidentialelection.162 And indeed Facebook did, in its own internal study back in 2010,demonstrate that, by manipulating their site design, it can drastically change voterturnout in a congressional election.163 In a country with a history of close elections,this should raise much more serious alarm than it has.Another reason why social media platforms have such breadth of power overspeech is that they control so much of not just which speech is heard but also which159 See Salvador Rodriguez, Facebook Shields Millions of VIP Users from StandardModeration Protocols, Per Report, CNBC (Sept. 13, 2021, 12:31 PM), https://www.cnbc.com/2021/09/13/facebook-shields-millions-of-vip-users-from-moderation-protocols.html[https://perma.cc/8ME6-2JTA].160 See, e.g., Paul Barrett et al., How Tech Platforms Fuel U.S. Political Polarization andWhat Government Can Do About It, BROOKINGS (Sept. 27, 2021), https://www.brookings.edu/articles/how-tech-platforms-fuel-u-s-political-polarization-and-what-government-can-do-about-it/ [https://perma.cc/SJ4M-ZNAP]; Tanya Basu, YouTube’s Algorithm Seems toBe Funneling People to Alt-Right Videos, MIT TECH. REV. (Jan. 29, 2020), https://www.technologyreview.com/2020/01/29/276000/a-study-of-youtube-comments-shows-how-its-turning-people-onto-the-alt-right/ [https://perma.cc/C2AA-32BN]; Reginald H. Gonzales, SocialMedia as a Channel and Its Implications on Cyber Bullying, DLSU RESEARCH CONGRESS2014, https://www.dlsu.edu.ph/wp-content/uploads/pdf/conferences/research-congress-proceedings/2014/LCCS/LCCS-I-009-FT.pdf [https://perma.cc/FHM3-9MTH]; see also ChadDe Guzman, Meta’s Facebook Algorithms ‘Proactively’ Promoted Violence Against theRohingya, New Amnesty International Report Asserts, TIME (Sept. 28, 2022, 9:13 PM),https://time.com/6217730/myanmar-meta-rohingya-facebook/ [https://perma.cc/HSY6-9KZZ].161 See, e.g., Elettra Bietti, The Data-Attention Imperative (Feb. 16, 2024) (unpublishedmanuscript) (available at SSRN), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4729500 [https://perma.cc/5BVA-LYVQ].162 Kevin Roose et al., Don’t Tilt Scales Against Trump, Facebook Executive Warns, N.Y.TIMES (June 30, 2020), https://www.nytimes.com/2020/01/07/technology/facebook-trump-2020.html [https://perma.cc/PS7G-YNFZ]. The executive, Andrew Bosworth, also statedthat while it was “tempting” to use these tools, and he had been “desperately wanting to pullany lever at [his] disposal to avoid” a Trump victory, he firmly believed that the companyshould not in fact do so. Id.163 See generally Robert M. Bond et al., A 61-Million-Person Experiment in Social Influenceand Political Mobilization, 489 NATURE 295 (2012) (reporting the results of the internal study).2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 651speech is uttered and amplified in the first place. A television station will providea platform for the speech of a small number of people. A social media platform willprovide a platform for a vast number of people. This increases the total number ofspeech events occurring on the platform, and the number of people seeking out theplatform (either for self-publication or consumption).Even if what distinguishes the class of agents that we are characterizing is thebreadth of their power, they must also have strength of power to qualify as quasi-state agents. But the strength of their power will manifest in a different way than thedense-power quasi-state agents discussed above. Because broad-power agents di-rectly control (because they own) means of communication, they will be able to fullyclose off those means to certain speakers and listeners. They might reject a requestfor publication or withdraw already-published speech. Or, in the case of a socialmedia platform, they might deny a speaker or (less likely) listener access to theplatform altogether.In some cases, these quasi-state agents’ power might even be stronger thangovernment power, for several reasons. First, their power usually cannot be effec-tively resisted. It would be extremely challenging for a speaker to force themselvesonto a media platform if denied access by the owner. By contrast, since the govern-ment does not directly control many means of communication, its influence overthose means must take the form of coaxing of the owners that theoretically could berefused. In the possible if extremely unlikely case that a speaker was able to forcethemselves onto a speech platform, perhaps by hacking, the owner might call on thestate itself, via its criminal and/or property law, to dispel them.Second, the checks on the power of broad quasi-state agents may be weaker thanthe democratic checks on the state’s power.164 Of course, media corporations are, likeemployers, subject to market pressures. Whenever a social media company, for ex-ample, suppresses controversial speech, it is likely to lose some customers who dis-liked the suppression. At the same time, a decision that alienates some customers canbring in or retain others. Likely only decisions to take down very popular speech aresufficiently deterred by market considerations. This is presumably partly why, at leastuntil Fox News introduced a new business model, the top news companies presentedonly mainstream views and excluded outlier views.165 Moreover, even a consumerwho is disconcerted by social media’s manipulative decisions might find it harderto “exit” the company than a citizen finds it to vote a manipulative politician out of164 See, e.g., Lakier, supra note 127, at 2320–23 (presenting historical testimony thatmonopolistic telecommunications corporations were feared to be less constrained than thegovernment).165 Miller, supra note 145, at 876; Bruce Bartlett, How Fox News Changed AmericanMedia and Political Dynamics, THE BIG PICTURE (May 21, 2015, 9:00 AM), https://ritholtz.com/2015/05/how-fox-news-changed-american-media-and-political-dynamics/ [https://perma.cc/NRT7-G7L3] (discussing how through ideology and economics media maintained a lib-eral character until Fox News upended the model).652 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615office. The social media market is oligopolistic and beset by “network effects”—thatis, when consumers gain benefits from using the same service that everyone else isusing.166 These can leave consumers with few viable options to exit to.Finally, like dense-power quasi-state agents, broad-power quasi-state agentsmay be able to act less transparently than the state can. Rather than twist the armsof publishers like the state must, broad-power quasi-state agents are the publishersand can simply decline to publish on their own. More subtly, they can manipulateeven speech that they publish so as to minimize its impact—diminishing its visibil-ity, juxtaposing it with counter-speech, or attaching warnings and disclaimers toit.167 Social media platforms can be all the subtler at such manipulations of speechbecause of their use of algorithms to display different speech to different users; theycan manipulate speech for only some users, or manipulate speech differently foreach user, such that patterns are hard to track.168D. State MotivesOne might say that the state poses a special threat to speech not only because ithas extraordinary power, but also because it has a distinctive and potent reason toabuse that power with respect to speech. Officials of the state generally want to con-tinue to rule. For that, they may find it helpful to silence anti-government speech,much as Vladimir Putin has done in Russia during his invasion of Ukraine.169 Theymay also find it helpful to ingratiate themselves with the population’s majority, andthereby to suppress speech that the majority would prefer not to hear. This motiveis all the stronger in democratic countries like the United States. To put all of thestate’s power over speech in the hands of one agent is additionally concerning whenthat agent has strong temptations to use it to their advantage to manipulate thepolitical process.Yet private corporations can have similarly bad motives to use their power forself-entrenchment. They have motives to silence voices favoring their competitorsor criticizing themselves (including their abuses of power). Research shows thatcompanies like Meta and Google favor their own products in search results.170166 Miller, supra note 145, at 893–94; What Is the Network Effect?, WHARTON ONLINE(Jan. 17, 2023), https://online.wharton.upenn.edu/blog/what-is-the-network-effect/ [https://perma.cc/3R67-9LM3].167 See Filippo Menczer, Facebook’s Algorithms Fueled Massive Foreign PropagandaCampaigns During the 2020 Election—Here’s How Algorithms Can Manipulate You, THECONVERSATION (Sept. 20, 2021, 8:31 AM), https://theconversation.com/facebooks-algorithms-fueled-massive-foreign-propaganda-campaigns-during-the-2020-election-heres-how-algorithms-can-manipulate-you-168229 [https://perma.cc/524S-CE2Q].168 See id.169 Anton Troianvoski & Valeriya Safronova, Russia Takes Censorship to New Extremes,Stifling War Coverage, N.Y. TIMES (May 18, 2022), https://www.nytimes.com/2022/03/04/world/europe/russia-censorship-media-crackdown.html [https://perma.cc/RGL5-ERA6].170 See, e.g., Orla Lynskey, The Power of Providence, in DIGITAL DOMINANCE: THE POWER2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 653Employers often have motives to prevent union organizing that might give employ-ees bargaining power. But corporations can also have motives specifically to affectthe balance of political power in their jurisdiction. Most will want to ensure that thejurisdiction passes laws cementing their own economic power. For instance, U.S.corporations might favor Republican candidates who will deregulate, reduce taxes,and discourage antitrust action.171 But a smaller number of corporations may havepurely political agendas. For example, Rupert Murdoch, owner of Fox News, re-portedly seeks to influence global political patterns in a more conservative andspecifically business-friendly direction.172 And Elon Musk, the new owner of Twitter,often appears to suppress speech based purely on personal preference.173III. REASONS AGAINST BINDING STATE-LIKE PRIVATE AGENTSThe last Part argued that our justification for binding state agents by FirstAmendment free speech rights also justifies binding certain highly powerful privateagents as well. But even assuming that argument is successful, it establishes no morethan a prima facie case for actually imposing any First Amendment duties on quasi-state agents. We can always decline to impose a duty we have reason to imposewhere there are persuasive countervailing reasons against doing so. For instance,one might be concerned (i) that the potential bearer has their own competing rightsclaims, or would not be able to meet the duty without undue cost; (ii) that theconsequences of enforcement for society would be unacceptable; or (iii) that theduty would not be practically enforceable against the bearer.174 Since my interest inthis Article is primarily about issues of principle, I focus on the first two.175To begin, what sorts of free speech duties could quasi-state agents be subject to?Under the First Amendment’s Free Speech and Press Clauses, the primary duty toOF GOOGLE, AMAZON, FACEBOOK, AND APPLE 176, 183–87 (Damian Tambini & MartinMoore eds., 2018).171 Harry Stevens, First of its Kind Study Shows CEO Political Donations Favor GOP,AXIOS (Mar. 31, 2019), https://www.axios.com/2019/03/31/ceo-political-giving-republicans[https://perma.cc/N5JE-92NN].172 See Jane Mayer, The Making of the Fox News White House, THE NEW YORKER(Mar. 4, 2019), https://www.newyorker.com/magazine/2019/03/11/the-making-of-the-fox-news-white-house [https://perma.cc/98G5-5KF2].173 See Robert A. George, Elon Musk Sabotages His Defense of Free Speech, BLOOMBERG(Nov. 4, 2022, 9:00 AM), https://www.bloomberg.com/opinion/articles/2022-11-04/elon-musk-sabotages-his-defense-of-free-speech?embedded-checkout=true [https://perma.cc/PM6R-8KZV].174 See generally, e.g., IRIS MARION YOUNG, RESPONSIBILITY FOR JUSTICE (2011); T.M.Scanlon, Rights and Interests, in 1 ARGUMENTS FOR A BETTER WORLD: ESSAYS IN HONOROF AMARTYA SEN (Kaushik Basu & Ravi Kanbur eds., 2008).175 I also doubt that the practical costs of implementation would be unbearable, par-ticularly because quasi-state agents will likely be (1) a small number of (2) corporations withgreat resources.654 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615which the state has been held is what we might call a duty of impartiality. While thestate can say whatever it wants in its own speech, it cannot interfere with others’speech by suppressing, burdening, or twisting it based on its content or viewpoint.There are limited exceptions for when the state is regulating speech in specialmanagerial zones (such as an office, school, or military base)—but even these regu-lations must have some motivation other than censorship. The same duty of impar-tiality, if applied to private agents, would seem to demand that they not discriminate,in distributing benefits and burdens under their control, on the basis of the recipients’speech. For instance, a private employer likely could not hire, fire, promote, ordemote employees based on viewpoints expressed in their speech. Or a newspapereditor likely would not be able to reject letters to the editor based on their viewpointor correctness.Certainly, imposing such a duty on a private individual would set off most, ifnot all, of the three alarm bells above. Completely dissolving the boundary betweenpublic and private actors for First Amendment purposes could have devastatingconsequences for both individuals and society as a whole. However, I argue that theseconcerns are much weaker if the duty of impartiality were to be applied only againstquasi-state agents—particularly because those agents are likely to be large corpora-tions distributing benefits and burdens in commercial exchanges. That said, I ac-knowledge that some of the concerns may justify imposing on quasi-state agentsonly diluted versions of First Amendment rights. I explain how that dilution couldwork in Part IV.Before discussing these countervailing concerns, let me explain one crucialanalytical point. I have maintained that free speech duties can “override” free speechrights. I say this metaphorically. What I mean, more directly, is that duties and rightsmust cohere or correlate (on Hohfeld’s framework, one cannot have a right unlesssomeone else has a duty, and vice versa), and so must be decided together in analysisthat takes into account all of the interests that are relevant to both determinations.Here, I will engage in an analysis not only of the interests that the public has inconstraining private power over free speech but also the interests that these powerfulprivate agents have. If the overall analysis concludes that these private powers havea duty to uphold the free speech rights of others, then they do not themselves haverights to do otherwise; any conflicting rights that they previously might have beenthought to have do not hold. In the end, a select few of those agents whom I havecalled quasi-state agents might not have duties at all because of the strength of theircountervailing interests—but this is not the typical case. Moreover, even for thoseon whom duties are properly imposed, the contours of those duties may be affectedby these countervailing interests. Where they lack speech duties, quasi-state agentsmay retain some speech rights.176176 On some of the models for operationalizing quasi-state agent duties discussed in PartIV, speech duties might be adjudicated on a case-by-case basis. In these cases, it is best tothink of the actual duties and rights as remaining unsettled and the analysis in each case2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 655A. Personal LibertyThe most common defense of the state-only-duties principle says that holdingprivate agents to a duty of impartiality like the state’s would impinge on personalliberty.177 One might even think that a private agent’s ability to be partial is pro-tected by other rights. The most obvious candidates, as I explain below, are expres-sive and associational rights—themselves protected by the First Amendment—andproperty rights. But these rights claims are most persuasive, I argue, in the case ofprivate individuals and small organizations, not the massively powerful corporationsthat quasi-state agents are likely to be.1. Expressive and Associative LibertySay that an individual, just because she dislikes another person’s speech, refusesto invite him onto her property, excludes him from a parade she is organizing,removes him from her will, or refuses to include his article in her edited volume. Allof these choices are expressive or associative: that is, they are motivated by her desireto express her dislike of the speaker’s speech or what that speech reveals about himand not to be associated with him on that basis. At the same time, her actions canhave an impact on his speech: he is directly denied a speech opportunity (in the caseof her parade, or edited volume) and he may be deterred from speaking similarly inthe future (so as to be invited again to her house, or to be re-added to her will). Inother words, the individual’s freedom of expression, perhaps counter-intuitively,includes a kind of freedom of suppression—freedom to deter or block others’ speechin certain ways. One might therefore be concerned that enforcing a duty of impar-tiality—a requirement of respecting others’ free speech rights—to private partieswould negate the duty holder’s own free speech rights.178 Expression and associationare such an important part of everyday human life that constricting those activitiesacross the board—requiring, for instance, close association with those whose beliefsone finds abhorrent—might take a heavy toll on any individual.But the objection seems substantially less compelling when the obligation isborn only by the quasi-state agents identified in the last Part. Three reasons whyinvolving that of free speech (and other) interests of different parties. See discussion infraSections IV.C, IV.D.177 Cf. Lillian BeVier & John Harrison, The State Action Principle and Its Critics, 96 VA.L. REV. 1767, 1785–87 (2013). For more support, see also Manhattan Cmty. Access Corp.v. Halleck, 139 S. Ct. 1921, 1928 (2019) (declining to find state action in order, in part, toprotect a “robust sphere of individual liberty”); Lakier, supra note 127, at 2336 (recountinghow employers resisted restrictions for firing workers based on speech as a violation ofproperty and liberty); Barak, supra note 111, at 230–31 (describing the individual autonomyloss from horizontally applying human rights).178 For a version of this objection, see Schauer, supra note 21, at 449–50.656 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615these agents’ expressive or associative claims are diminished stand out. First, insofaras nearly all quasi-state agents will be corporations, the free speech rights of corpo-rations are less weighty than those of individuals.179 Corporations lack the autonomyinterests that individuals possess.180 Indeed, the Supreme Court itself has expresslyrecognized only an instrumental rationale for corporate speech rights.181Corporate speech can also serve an instrumental purpose: multi-membered cor-porations can speak with voices that are not identical to that of any of their members,thereby improving the diversity and ostensibly overall quality of the marketplace ofideas.182 But for this purpose, corporations’ ability to suppress others’ speech is sub-stantially less critical than their ability to issue statements on their own behalf183—anability which would not be affected by a state-like duty of impartiality. I delve moredeeply into these instrumental concerns in the next subsection.Second, even if corporations have associational rights for noninstrumental rea-sons, the associations of the largest commercial corporations are typically lessintimate and less valuable than personal associations. Corporate relationships tendto be transactional, whether they are employment relationships, seller-buyer rela-tionships, or provider-user relationships. The interactions within these relationshipsare contracted and focused on the corporation’s narrow goals—in the case of acommercial corporation, usually profit. The larger the corporation, the more thisholds true. Walmart almost certainly has less-valuable relationships with its custom-ers and employees than the owner of almost any local deli has with its own. Beingforced to be roommates with a person with abhorrent political views is almostincomparable to being forced to engage in a commercial transaction with that personamong many thousands of others. Moreover, very large corporate owners are un-likely to be perceived as sympathetic to the views of those whom they are forced toassociate with, given their ability to disavow any such sympathy and to explicitly179 Cf. Adam Winkler, Corporate Personhood and the Rights of Corporate Speech, 30SEATTLE U. L. REV. 863, 863 (2007) (arguing that corporate speech rights are lesser thanthose held by individuals).180 See, e.g., C. Edwin Baker, Realizing Self-Realization: Corporate Political Expendi-tures and Redish’s The Value of Free Speech, 130 U. PA. L. REV. 646, 648–58 (1982); seealso Avital Mentovich et al., The Psychology of Corporate Rights: Perception of CorporateVersus Individual Rights to Religious Liberty, Privacy, and Free Speech, 40 LAW & HUM.BEHAV. 195, 195 (2016); Daniel J.H. Greenwood, Essential Speech: Why Corporate SpeechIs Not Free, 83 IOWA L. REV. 995, 996 (1998).181 The Supreme Court itself has expressly recognized only an instrumental rationale forcorporate speech rights and so the Court could follow this reasoning consistently with itsown doctrine. Baker, supra note 180, at 655. Of course, there are reasons to think that theCourt would not actually treat corporate speech rights as less weighty than individual speechrights. See, e.g., Sorrell v. IMS Health Inc., 564 U.S. 552, 580 (2011).182 See Citizens United v. FEC, 558 U.S. 310, 354–55 (2010).183 Cf. id. at 354 (“By suppressing the speech of manifold corporations, both for-profit andnonprofit, the Government prevents their voices and viewpoints from reaching the public andadvising voters on which persons or entities are hostile to their interests.”).2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 657explain their legal obligations (e.g., “While we do business with these individualsas required by law, we abhor their messages”).Third, a parallel argument applies to expressive rights. Even if corporations ingeneral have expressive rights for noninstrumental reasons, the largest corporations’decisions about allocating benefits and burdens (including the awarding or with-drawing of speech opportunities) may have only minimal expressive significance.The expressive significance of an action necessarily depends on its context, includingthe number of recipients of the benefits or burdens: the more recipients, generallythe less expressive significance. For instance, a painter’s choice to accept a commis-sion to paint a portrait of a private party sends a more meaningful message than amassive corporation’s choice to accept the same private party’s online order. Eachdecision “to deal,” so to speak, will be approached and perceived in a different man-ner. In the former case, a message is more likely to be sent and perceived. Part of thedifference is that the painter can, assuming she has many potential customers, bechoosy; given that she will only take one customer at a time anyway, it conveys amessage—even if just mere non-repulsion—that she takes this particular customer.By contrast, the largest corporations often have a default of dealing with everyoneand so any affirmative choice to deal will typically not be felt or perceived asmeaningful.184 And again, corporations’ disavowals should significantly alleviateany such feeling or perception.For illustration, consider Meta. Its platforms Facebook and Instagram acceptnearly every user onto their platforms and so an acceptance does not send any nu-anced message. Indeed, some scholars and courts have argued that a platform likeMeta is not speaking when it accepts or excludes users’ content, and therefore hasno expressive claim.185 At the least, it seems clear enough that Meta’s expressiveclaim is diminished. One could argue that Meta desires to send the blunt messagethat certain fringe, especially abhorrent ideas are unacceptable, and thus to eject thosewho espouse those ideas from the platform. A duty of impartiality would prevent itfrom doing even this; the state cannot reject even speech it deems abhorrent. But thebluntness of the message makes its expression less imperative.186Notice that the arguments above apply most strongly to large corporations thathave “open[ed] up” their services “for use by the public in general”187—those who184 This argument would not apply to even large media companies, such as televisionnetworks, that publish only a small selection of speech. Yet while television networks likeFox News and CNN may have a large viewership, the case for quasi-state agency is for themmuch weaker than for online-only companies that reach a majority of Americans, such asFacebook and Google.185 See, e.g., Trouillard, supra note 5, at 265–66.186 But see Hurley v. Irish-Am. Gay, Lesbian & Bisexual Grp. of Bos., Inc., 515 U.S. 557,580–81 (1995) (concluding that the fact that city-authorized parade organizers had excludedalmost no one from their parade did not prevent them from exercising expressive rights whenthey excluded an LGBTQ group).187 Marsh v. Alabama, 326 U.S. 501, 506 (1946).658 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615are arguably common carriers. Yet the arguments can also apply to corporations thatexclude a small number of speakers for public relations reasons and not because ofa strong mission. This maps onto the approach of the original Marsh decision.I should offer one qualification to the entire discussion above. Because a firm’semployment relationships are in virtually every case less numerous than its customerrelationships, a firm’s decision to employ someone will usually be taken and per-ceived as at least more expressive/associational than a decision to deal with some-one. In the case of a very large firm, the expressive/associational significance of thedecision will still be nil. But in the case of a smaller firm, her relationship with othermembers of the firm may be somewhat more intimate and their hiring of her may bedeemed more expressive. That is, the associational claims of even a quasi-state firmwill grow as the firm shrinks. I return to this qualification in Part IV.2. Property RightsMany of the same decisions that private agents make about allocating benefitsand burdens based on speech implicate property rights.188 The removal of someonefrom your house or your will, for instance, is an associative act but, at the same time,a choice about how to manage your property. Similarly, a shopping mall owner’srefusal to allow a protestor on mall grounds is a decision about property.189 Thereare three reasons, however, why quasi-state agents’ claims to be partial are not sub-stantially stronger just because property is involved.First, quasi-state agents choose to grow their power (or, in the rare case wherethey do not, choose to keep it). As our power grows, so can our responsibilities.190One makes a choice to simultaneously grow one’s power and to limit how one canlegitimately exercise it. If a corporation truly wanted to avoid First Amendment ob-ligations, and thereby to retain full freedom to choose how to dispose of its property,then it could do so simply by not seeking to expand the corporation’s power. Forinstance, a broad-power quasi-state agent might sell off part of their media corpora-tion to avoid triggering First Amendment obligations.Second, quasi-state agents’ property claims are likely to concern commercial,corporate property and one’s interest in disposing of this sort of property is signifi-cantly more limited than one’s interest in disposing of, say, residential or personal188 Nearly all jurisdictions I have encountered that apply free speech rights against privateentities temper that application based on potential losses to those entities’ commercial andproperty interests. See, e.g., Bundesgerichtshof [BGH] [Federal Court of Justice], July 29,2021, Entscheidungen des Bundesgerichtshofes in Zivilsachen [BGHZ] III ZR 179/20, ¶ 88(Ger.); State v. Schmid, 423 A.2d 615, 623–33 (N.J. 1980); Batchelder v. Allied Stores Int’l,Inc., 445 N.E.2d 590, 592–93 (Mass. 1983).189 See, e.g., Lloyd Corp. v. Tanner, 407 U.S. 551, 567–70 (1972); Pruneyard ShoppingCtr. v. Robins, 447 U.S. 74, 88 (1980).190 See, e.g., Erin L. Miller, With Great Power Comes Great (Individual) Responsibility,20 POL. PHIL. & ECON. 22, 39 (2020).2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 659property that one uses in one’s private life. These latter types of property tend to bemore closely linked to the self and to be used in one’s private and intimate plans andprojects. Intuitively, one has a stronger interest in deciding whom to allow to entertheir home or to whom to gift their favorite sweater than in deciding whom to allowto enter their place of business or to sell to/buy from. This intuition is reflected inhow aggressively most societies regulate business activity in particular. But again,the intuition can falter when it comes to small businesses, where a mom-and-pop’sstorefront may be as treasured as the family home.Third, our interest in our property arguably diminishes marginally as the propertyaccumulates. A person’s interest in their first ten thousand dollars of income in ayear is much weightier than their interest in the tenth ten thousand dollars they makein the same year, assuming they have no savings. The first ten thousand dollars areneeded for the necessities of human life, such as shelter and food, whereas the tenthten thousand dollars are free to be spent on non-essential goods or even luxuries.Insofar as a duty of impartiality would merely reduce the profits that an alreadyexceptionally lucrative firm can make, this interest cannot stand up to the potentialharms that quasi-state agents can inflict on others’ First Amendment rights.191Antitrust is an illustrative case. Antitrust law prohibits certain business firmsfrom engaging in free market transactions that would be permissible but for themarket power of the firm.192 We deem the integrity of an entire competitive market-place to be more important than any one, extremely powerful firm’s ability to dobusiness exactly as it would like (i.e., dispose of its property exactly as it wouldlike). For instance, in Associated Press v. United States, discussed in the next sec-tion, the Court states:While it is true in a very general sense that one can dispose ofhis property as he pleases, he cannot “go beyond the exercise ofthis right, and by contracts or combinations, express or implied,unduly hinder or obstruct the free and natural flow of commercein the channels of interstate trade.”193If we make this sacrifice of liberty in order to prevent the monopolization of theeconomic marketplace, it is unclear why we would not make the same sacrifice inorder to prevent the monopolization of the marketplace of ideas or the control of thefree speech choices of individuals. Indeed, Associated Press assumes that propertyrights end prior to this point.191 Some support for this argument is actually found in judicial precedent. See, e.g.,Buckley v. Valeo, 424 U.S. 1, 19–21 (1976) (recognizing a lesser free speech interest in do-nating more money to a candidate).192 See Associated Press v. United States, 326 U.S. 1, 15 (1945).193 Id.660 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615The case for property rights might be stronger, however, when it comes to thesurvival of one’s business. Owning a business is central to many people’s livelihoodand even identity. If First Amendment obligations would somehow harm a com-pany’s actual existence, and not merely its size or profits, then there might bearguments cutting against the enforcement of those obligations—or at least their fullforce. I expect that this concern will seldom arise for quasi-state agents, but I willreturn to it in Part IV.The argument from personal liberty against imposing a duty of impartiality onquasi-state corporations is generally quite weak, perhaps too weak to override thefull-value First Amendment interests on the other side of the ledger: the First Amend-ment interests that quasi-state agents’ partiality would infringe. However, as theremarks above acknowledge, personal liberty may be a more compelling defenseagainst a duty of impartiality when it comes to (a) smaller corporations and (b)corporations that stand to lose their business altogether.B. Societal HarmsAnother potential defense of the state-only-duties principle is that enforcing aduty of impartiality on private agents would intolerably burden society as a whole—including in ways that undermine free speech interests. Specifically, private agentsneed to be able to exclude others based on their views (as revealed, among otherplaces, in their speech) in order to create certain types of valuable social groups thatdrive diversity, energy, and experimentation in civil society.The first group is the like-minded community. We want people to be able toconvene events, clubs, and online forums, to which only those of certain interests,values, or political persuasion are invited. Such communities are obviously invalu-able insofar as they allow others to enjoy the company of those of similar views andvalues and sometimes even to debate nuances and applications of their shared views.In addition, a like-minded group sometimes becomes more of a collective agentwhen it comes together not just to hang out or debate but also to achieve a collectivevision. Some degree of coherence among participants’ beliefs and values is neces-sary for effective collective action. A group that aims to combat racism wouldobviously struggle to achieve its goals if it was forced to include white supremacists.In the First Amendment context, a vibrant marketplace of ideas itself dependson the existence of such viewpoint-discriminatory collective agents: the media.Media organizations are needed to filter the morass of speech created each day andelevate above the rest only the speech they deem worthy of reaching larger audi-ences. That is, we need decidedly partial media, albeit a diversity of them. If everymedia organization was required to publish speech on a first-come or some otherimpartial basis, the quantity of speech available would be unmanageable for anyindividual and we would lack anything resembling a common public sphere.2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 661Yet notice that, insofar as these are instrumental goals, they should be achiev-able so long as the vast majority of corporations lack impartiality duties. If onlyquasi-state agents are subject to impartiality obligations, then the remaining corpora-tions’ pursuit of decentralized private aims may well be enough to sustain a diversecivil society and functional marketplace of ideas. For instance, if Reddit was aquasi-state agent, a subreddit could still pursue its own ideas and kick out anymembers who failed to live up to them, even if the overarching platform could noteject members based on viewpoint.Let me return to our two types of quasi-state agents: broad-power and dense-power. Imposing First Amendment obligations on broad-power quasi-state agentsis especially unlikely to create a substantial dent in the vibrance of civil society andthe marketplace of ideas. The media corporations most likely to rival the state in thebreadth of their power over public discourse are also the least likely to be engagingin the serious exclusion and filtering of speech that is vital to the working of themarketplace of ideas. This is a function of market principles; in order to attract thelargest possible numbers of viewers/listeners/users, a company must appeal to a maxi-mally diverse segment of the population. A communications giant like Meta tendsto include on its platforms vastly more speech created by users than it excludes.194Its platforms are thus not primarily relied on, like newspapers and news televisionshows, to amplify a small number of ideas and subjects over others based on quality.At the same time, when it comes to social media platforms in particular, wemight worry that their very possibility depends on the existence of a certain kind ofspeech discrimination: content moderation, or the ability to remove and demotecertain speech. A platform that could not exclude or demote any speech or speakersmight soon find itself overwhelmed by bots and spam.195 If so, it might cease to bethe sort of platform that attracted large numbers of users in the first place.Now consider dense-power quasi-state agents. The example I have offered is anemployer in a loose labor market with easily replaceable employees. Some suchemployers might still be pursuing specific private, even non-profit, aims. If theseemployers were not permitted to fire or demote employees who contradicted thecompany’s vision in private or in public, then the company might not be able toeffectively pursue those aims. This remains a fairly strong argument against theapplication of the First Amendment against these private employers, at least in casesinvolving speech core to the mission of the employer.Overall, both types of quasi-state agents, if bound by a duty of impartiality,would be less likely to effectively achieve their private aims—a loss of uncertain194 Facebook reports removing only minute percentages of content, such as .02% catego-rized as hate speech or .08% categorized as nudity and sexual activity. Community StandardsEnforcement Report, META (2023), https://transparency.fb.com/data/community-standards-enforcement/ [https://perma.cc/P9DS-CY2M].195 See, e.g., Jack M. Balkin, How to Regulate (and Not Regulate) Social Media, 1 J. FREESPEECH L. 71 (2021); Ashutosh Bhagwat, The Law of Facebook, 54 U.C. DAVIS L. REV.2353, 2353–54 (2020).662 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615value to society. This would therefore probably be the main point of disagreementin the implementation of quasi-state agent duties. Again, the more broadly an agentopens up its services for use by the public, offering a service to all without a message,the more likely it is to have private aims that are frustrated by free speech duties. InPart IV, I examine how these losses could be somewhat limited by imposing onlya restricted duty of impartiality.Just one subset of quasi-state agents has an especially strong set of countervail-ing reasons against being subject to full First Amendment obligations: small privateemployers. Due to their small size, their associative and expressive liberties aresomewhat stronger than those of other quasi-state agents. They may also be more atrisk, without viewpoint-discriminatory employment practices, of not being able tokeep their businesses afloat or of losing the chance to realize their ideological aims.Imposing full First Amendment duties on other quasi-state agents, on the otherhand, raises few compelling liberty concerns. It might however, make it harder forthose whose business models or missions are built around viewpoint discriminationto thrive. The extent of the social cost resulting from this difficulty is up for debate.I am personally inclined to think that the cost would be slighter than we might fear,given the possibility of new business models emerging and the remaining ability ofsmaller, less powerful corporations to engage in viewpoint discrimination. But giventhe substantial unknowns, I explain in the next Part how partial First Amendmentduties on these agents might offer a fair compromise.IV. DOCTRINAL MODELS FOR BINDING STATE-LIKE PRIVATE AGENTSThe preceding sections have made the case that extraordinary power over speech,whether public or private, should not be free of all First Amendment constraints.Private entities with state-like power over others’ speech, which I call quasi-stateagents, should be treated differently by First Amendment doctrine than private indi-viduals with the power of only their own voice. In this Part, I explore a range ofoptions for what that differential treatment might look like.One relevant consideration is that quasi-state agents are, while state-like, notexactly like the state. And some quasi-state agents will be more state-like than others.The main sticking point, as explained in the last Part, is this; private agents—evenmassively powerful ones—often pursue valuable private goals that the state is notcapable of pursuing, given the latter’s charge to fulfill the public interest. Respectfor these goals—not for any particular goal, but for the general practice of pursuingprivate aims—could call for a softening of quasi-state agents’ constitutional bur-dens. I am not entirely persuaded, except for the exceptional class of small privateemployers, but I have acknowledged compelling arguments for such a softening.In what follows, I present three distinct models for how quasi-state agents mightbe bound by the First Amendment, with varying degrees of flexibility to accommodate2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 663these agents’ private aims.196 The first and most obvious option is that quasi-stateagents are treated just like the state under the First Amendment; the agents haveduties directly from the text—the same duties that courts have recognized the stateas bound by. I call these direct and identical duties. I explain, however, why thismay still allow for the substantial pursuit of private aims, given how the doctrineholds even the government to a relaxed impartiality standard when it is acting in amanagerial capacity.The other two models add even more flexibility. The second option is that quasi-state agents have direct and lesser duties. This option is the same as the last, exceptthe duties of the quasi-state agents are less demanding than those imposed on thestate—for instance, they might result from weighing the private interests of a com-pany against the damage they can do to free speech values. The government hashigher duties because it has no competing private interests. The third option is thatquasi-state agents have indirect duties. Here, quasi-state agents’ duties would be“indirect” in the sense that they are not self-enforcing (do not come directly from thetext, like direct duties) but instead would require that legislatures or courts takeaction to activate and specify those duties in further detail—and potentially in moreflexible, less demanding ways than First Amendment law has.These models are not mere analytical possibilities: they have been used by realcourts in many jurisdictions to interpret the constitutional duties—including freespeech duties—of private agents.197 Most obviously, the U.S. Supreme Court itselfhas recognized constitutional constraints on private action in isolated pockets of itsjurisprudence using all three of these models.198 The discussion in Part I touched onsome, but not all, of these examples. But other jurisdictions have more openly andliberally applied free speech rights against private agents. Several U.S. states, includ-ing most notably New Jersey, Massachusetts, and California, have applied state-constitutional free speech guarantees against limited numbers of private actors usingeither direct or indirect duties.199 In addition, many countries apply their constitu-tional rights, sometimes including free speech rights, against private actors.200 Thisis known in comparative constitutional law as the “horizontal application” of196 An awkwardness of Supreme Court substantive due process jurisprudence is thatfederal agents are bound by speech rights via the First Amendment and state agents arebound by arguably identical speech rights via the Fourteenth Amendment—though both setsof duties (those of federal agents and those of state agents) are expounded through the samebody of case law (consisting of both state and federal cases). A private entity judged to havegovernment-like power over speech could thus be found to be bound by either the FirstAmendment or the Fourteenth Amendment.197 See, e.g., infra notes 208, 216 and accompanying text.198 See, e.g., infra notes 208, 247 and accompanying text.199 See, e.g., supra notes 48–51 and accompanying text; see also infra note 216 andaccompanying text.200 See infra note 204 and accompanying text.664 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615constitutional rights and is selectively practiced by countries including Canada, India,Ireland, Germany, South Africa, and Switzerland.201 I draw on the practices of manyof these jurisdictions in articulating the models of quasi-state agency below.A. Expanding State ActionBefore launching into my three options, I want first to set aside the one proposalfor constraining private parties by constitutional rights that has already appeared inAmerican constitutional legal scholarship, and explain why it is not well-suited tothe free speech context.Some critics of the state action doctrine, writing primarily in the context ofracial discrimination, have attempted to expand the concept of state action so far asto effectively abolish the distinction between private and state action.202 On theirview, states authorities can be seen as engaging in state action whenever theyenforce private choices. To the extent that those private choices infringe on constitu-tional rights, the enforcement of them is invalid.203 So, for instance, if a person wasto exclude someone from their property on the basis of their race, and called on thepolice to enforce that decision, then they would implicate the state in their racialdiscrimination. Because of the state’s property rights, the state always sides with theprivate owner anyway. Extended into the First Amendment context, this would sug-gest at least that private owners could not enforce partial decisions toward others’speech, in cases involving property or other material under the owner’s control.You can see how this would work. If a shopping mall attempts to exclude my pro-test from its grounds, and I attempt to continue to protest anyway, then the policemight remove me from the property. If Facebook excludes me from its platform andI somehow (very unrealistically, given my lack of computer skills) managed to hackmy way back onto the platform, I could be fined for breaking laws against hacking.In all of these cases, the police’s action would itself be considered a speech-discrim-inatory act of the state.This is the only model that I would recommend against adopting, for two rea-sons. First, as discussed in Part III, the countervailing reasons against imposingspeech-respecting duties on private parties apply most strongly when those dutiesare applied against all private parties. Under the argument considered here, anydistinction among the private agents would not matter; all that matters is the role ofthe state in enforcing those agents’ choices. I therefore suspect that a more nuanced201 See infra note 205 and accompanying text.202 See, e.g., Gary Peller & Mark Tushnet, State Action and a New Birth of Freedom, 92GEO. L.J. 779, 789 (2004); Cass R. Sunstein, State Action Is Always Present, 3 CHI. J. INT’LL. 465, 467–68 (2002).203 On the most obvious (but not prevailing) interpretation of the case Shelley v. Kraemer,334 U.S. 1, 20–21 (1948), the Court followed exactly this line of reasoning in holding thata court enforcement of a racially restrictive covenant on real estate sales was invalid.2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 665approach like the three canvassed below would be more appropriate to the FirstAmendment in particular (however well this argument might work with respect toother rights).Second, an expansion of state action would likely fail to constrain a great dealof quasi-state agent conduct that threatens vital First Amendment interests. Oftenspeech-discriminatory decisions by these agents are not sought to be enforced by law.For instance, when Facebook seeks to exclude a user from the platform, it simplydoes it—without any police or judicial assistance. Any hackers who did break throughcould be easily removed fairly quickly, without resort to the law.B. Direct and Identical DutiesTo reiterate, the direct-and-identical-duties model conceives of the First Amend-ment as directly binding quasi-state agents in the same way that it binds full stateagents. Quasi-state agents could look to existing First Amendment jurisprudence forthe clearest guidance for their conduct. The directness of the binding means thatquasi-state agents’ duties would be immediately enforceable by courts, without anyneed for action by other government decision-makers.204 Quasi-state agents couldbe directly challenged by other private parties, just as the state can be, for violatingthe Speech Clause or the Press Clause.This model for the horizontal application of constitutional rights is relativelyrare in the world but has been selectively implemented in Ireland and SouthAfrica.205 Ireland’s highest court, for instance, has recognized that the nation’sjudges have a duty to create causes of action for violation of constitutional rights byprivate actors.206But the direct-and-identical-duties model is not entirely alien to the Americancontext, either. Marsh v. Alabama can be read as one example under the FirstAmendment. In addition to repeatedly implying that Gulf had violated the constitu-tional rights of the public to receive information, the opinion explicitly states that:“The more an owner, for his advantage, opens up his property for use by the publicin general, the more do his rights become circumscribed by the . . . constitutionalrights of those who use it.”207 Logan Valley, though overruled, arguably followedthis model as well.208 In addition, the Thirteenth Amendment, as mentioned above,204 Courts might need to do one thing: create new causes of action, as they have done withrespect to certain constitutional rights in other countries. See Stephen Gardbaum, The “Hori-zontal Effect” of Constitutional Rights, 102 MICH. L. REV. 387, 396–98 (2003).205 Id.206 See Hosford v. John Murphy & Sons, [1987] IR 621, 626 (Ir.) (“Uniquely, the IrishConstitution confers a right of action for breach of constitutionally protected rights againstpersons other than the State and its officials.”).207 Marsh v. Alabama, 326 U.S. 501, 506 (1946).208 391 U.S. 308, 325 (1968).666 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615is widely recognized as establishing direct-and-identical duties for all private actorsnot to subject others to slavery or involuntary servitude. This Amendment, as op-posed to the Fourteenth Amendment, grants Congress power to regulate privateconduct directly.209 The Court has also recognized that causes of action that areordinarily available only to sue state agents can be used to sue private parties forThirteenth Amendment violations.210Notice one implication of this model: if a quasi-state agent has any First Amend-ment right that conflicts with its duties under the First Amendment then the dutypresumably takes precedence, insofar as the same occurs for state agents. So, forinstance, Gulf Shipbuilding could not claim that it had a free speech right to sup-press speech in the streets that it owned because such a right would conflict with theduties the Court declared it to have.Direct-and-identical duties offer the least accommodation for the private aimsof quasi-state agents because they treat those agents just like the state, which has noprivate aims. But, perhaps surprisingly, this model may still offer some amount ofaccommodation given contemporary free speech doctrine. While I emphasized theburdens of a duty of impartiality in Part III, the reality is that even the state’s dutyhas many exceptions that relieve its total burden. The first and most obvious, men-tioned earlier, is that the state may speak on its own behalf. State actors frequentlyissue statements condemning or praising not only the actions of private individualsbut also their speech. The state can even choose to fund some organizations but notothers based on their speech.211 The second is that the state is able to directly sup-press speech as part of its role as an employer, the administrator of public schoolsincluding universities, and the owner of other spaces such as government buildingsand military bases. Robert Post has referred to these special spaces as “managerialdomains.”212 In each of these domains, courts recognize that these state institu-tions—whether the public office or the public school—have certain purposes that,to be effectively accomplished, may require suppressing speech based on its content.Thus, for instance, the state is able to regulate speech content that is substantiallydisruptive of workplace or learning environments and may even fire employees orsuspending students on this basis.213Imagine that quasi-state agents were treated under First Amendment law like statesrunning their own special managerial domains. If so, then any duty of impartiality209 See The Civil Rights Cases, 109 U.S. 3, 20 (1883).210 See Jones v. Alfred H. Mayer Co., 392 U.S. 409, 438, 441–44 (1968) (holding that 42U.S.C. § 1982 applies to private suits under the Thirteenth Amendment).211 See Rust v. Sullivan, 500 U.S. 173, 178, 203 (1991); Rumsfeld v. FAIR, 547 U.S. 47,59–60 (2006).212 See ROBERT C. POST, CONSTITUTIONAL DOMAINS: DEMOCRACY, COMMUNITY, MAN-AGEMENT 254 (1995).213 See Pickering v. Bd. of Educ., 391 U.S. 563, 574–75 (1968); Connick v. Myers, 461U.S. 138, 150 (1983); Tinker v. Des Moines Indep. Cmty. Sch. Dist., 393 U.S. 503, 512–13(1969).2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 667would likely contain exceptions for (1) speech that is clearly in the name of thequasi-state agent and, most significantly, (2) suppressions and demotions of speechthat undermine the basic purpose of the domain. In the case of social media plat-forms, this seems likely to allow for the continuation of platform “recommendation”lists and, at the least, basic housekeeping acts like the removal of bots and overtspam. In the case of employers, it would likely allow for disciplining employeesbased on rude or disruptive speech—certainly when that speech is uttered as part oftheir job responsibilities.214But direct-and-identical duties would likely not allow discrimination based onpolitical or ideological viewpoints. Therefore, it would continue to still impair com-panies’ pursuit of certain private goals, as discussed in the last Part. Some additionalroom for maneuvering in this space might be allowed under the next approachdiscussed.C. Direct and Lesser DutiesA second possibility is that quasi-state agents are directly bound by First Amend-ment duties, but their duties are not identical to those the state bears. If quasi-stateagents were only subject to lesser duties, this might permit them to pursue privateaims beyond the basic housekeeping and corporate statements allowed by direct-and-identical duties. Justice Marshall may have hinted at this approach in LoganValley when he noted that private businesses subject to customers’ First Amendmentrights would still have the “power to make reasonable regulations governing theexercise of First Amendment rights on their property,” and that these rights wouldbe “at the very least co-extensive with the power possessed by States and municipal-ities . . . .”215 Indeed, several of the states that have applied their state constitutionalfree speech guarantees against private agents in the wake of Logan Valley havearguably adopted the direct-and-lesser-duties model.216Courts would have to identify these lesser duties, on a separate doctrinal trackfrom that devoted to full state agents. Perhaps the most straightforward way of doingso would involve balancing the rights of quasi-state agents (including any claimedexpressive and associative rights) against the free speech rights they threaten tohinder. Many of the same factors examined in Part III, including the expressive andproperty interests of quasi-state agents, could be used in the balance. Open balancing214 Cf. Garcetti v. Ceballos, 547 U.S. 410, 421 (2006) (limiting the freedom of speech ofeven public employees to speech not uttered pursuant to the employee’s “official duties”).215 Amalgamated Food Emps. Union Loc. 590 v. Logan Valley Plaza, Inc., 391 U.S. 308,320 (1968) (emphasis added).216 See Batchelder v. Allied Stores Int’l, Inc., 445 N.E.2d 590, 595–96 (Mass. 1983) (rec-ognizing a right to collect signatures for ballot access on private property, though only whenbalanced against the property interests of the owner); see also Glovsky v. Roche Bros. Super-markets, Inc., 17 N.E.3d 1026, 1032 (Mass. 2014) (referring to this analysis as balancing).668 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615is often perceived as alien to the First Amendment, but the doctrine does employ it onoccasion. For instance, under the Pickering test, courts balance the free speech rightsof employees against the state’s interest in maintaining a functioning workplace.217We can see how a balancing model might work from judicial opinions in NewJersey, which has adopted a direct-and-lesser-duties model for interpreting its stateconstitution’s own free speech clause. In State v. Schmid in 1980, the SupremeCourt of New Jersey held that the (state) freedom of speech applies not just againstthe government but against certain private entities that “assume[] a constitutionalobligation not to abridge the individual exercise of such freedoms because of thepublic use of their property.”218 In each case, the court said it would balance theinterests of the property owner against the speech and assembly interests of the pub-lic.219 It would need to consider, among other things, how the expression mightaffect the normal uses of the property, the scope of any (even implied220) invitationto the public to use the property, and the extent to which the property was needed toconvey the speakers’ message.221 The court clarified, however, and has reiterated inlater cases, that the burden on the private property owner should not be too great.222Using this analysis, the court overturned Schmid’s conviction for trespassingwhile distributing political literature on Princeton University’s campus.223 Universitypolicy required anyone not affiliated with the university to have an invitation orpermission to be on campus, and Schmid lacked both.224 The court concluded thatSchmid could invoke the First Amendment against Princeton because of the univer-sity’s educational mission and its general invitation to the public to congregate onits grounds for purposes relating to free speech.225 But this did not mean that Prince-ton was required to let everyone onto its campus just as a city must let everyone intoits parks; the university has due deference in administering its educational institu-tion.226 At the same time, the court insisted that the university must have a “reason-able regulatory scheme” governing public free speech on its campus, rather than astandardless and discretionary permission requirement that did not even attempt tograpple with the free speech interests at stake.227Germany also offers an example of how such balancing might work in morecontemporary cases. Under the German system, private agents’ duties under the Basic217 See Pickering, 391 U.S. at 568.218 423 A.2d 615, 628 (N.J. 1980).219 Id.220 Cf. N.J. Coal. Against War in the Middle E. v. J.M.B. Realty Corp., 650 A.2d 757,780–81 (N.J. 1994) (clarifying this fact).221 See Schmid, 423 A.2d at 623, 630.222 Id. at 630.223 Id. at 633.224 Id. at 617.225 Id. at 631–33.226 Id. at 632.227 Id. at 633.2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 669Law (the constitution) are indirect on the model explained below.228 But in decidingthe exact contours of those duties, its courts often consider the competing constitu-tional rights of private parties and attempt to give each effect to the greatest extentpossible.229 Many of these cases balance the free speech rights of individuals versuscorporations, including social media platform owners.230 Factors that can strengthenan individual’s rights claim vis-à-vis a platform’s can include the platform’s domi-nant position in an economic market; the corporation’s “structural advantage,” orgreater bargaining power (from, e.g., superior information or resources); or the ne-cessity of the platform to individuals’ participation in social life.231 Conversely,factors that can strengthen the corporation’s rights claim include its property interestin the successful operation of its business model and its interest in not being heldliable for others’ actions.232Consider a recent pair of German cases against Facebook. Two individuals whohad posted right-wing, anti-immigrant content had their posts removed and theiraccounts blocked for violating Facebook’s Community Standards against hatespeech.233 To determine what Facebook’s constitutional obligations, if any, were tothe users, the Federal Court of Justice balanced the individual users’ free speechrights against Facebook’s “commercial interest in creating an attractive communica-tion and advertising environment,” its interest in not being liable for third-partycontent, its free speech rights, and the free speech rights of society as a whole.234 Itsconclusion was strikingly similar to Schmid’s: Facebook had a right to develop itsown Community Standards and enforce them—but it has a legal obligation, in doingso, to take into account its users’ speech rights.235 As a result, it must offer rules thatare clear and objective, so that users will understand them, and it must provide certaindue process to users whose content or accounts are removed, including notice, a228 See generally BVerfG, 1 BvR 3080/09, Apr. 11, 2018, https://www.bundesverfassungsgericht.de/SharedDocs/Entscheidungen/EN/2018/04/rs20180411_1bvr308009en.html[https://perma.cc/47FE-233J].229 Id. ¶ 32.230 Bundesgerichtshof [BGH] [Federal Court of Justice], July 29, 2021, Entscheidungendes Bundesgerichtshofes in Zivilsachen [BGHZ] III ZR 192/20, ¶ 8 (Ger.).231 See, e.g., BVerfG, 1 BvR 3080/09, ¶ 41; see also BVerfG, 1 BvQ 42/19, § 2, May 22,2019, https://www.bundesverfassungsgericht.de/SharedDocs/Entscheidungen/EN/2019/05/qk20190522_1bvq004219en.html [https://perma.cc/8V6G-Y6J5] (citing the “considerablemarket power in Germany” of Facebook and the fact that the company is “used by more than30 million people in Germany every month”).232 See BVerfG, 1 BvR 3080/09, ¶¶ 33, 35–36.233 Bundesgerichtshof [BGH] [Federal Court of Justice], July 29, 2021, Entscheidungendes Bundesgerichtshofes in Zivilsachen [BGHZ], ¶ 7; id. ¶ 5.234 Id.; Matthias C. Kettemann & Torben Klausa, Regulating Online Speech: Ze GermanWay, LAWFARE (Sept. 20, 2021, 8:01 AM), https://www.lawfaremedia.org/article/regulating-online-speech-ze-german-way [https://perma.cc/CQ8V-E94Y].235 Kettemann & Klausa, supra note 234.670 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615statement of reasons, an opportunity to respond, and a second review.236 Just asSchmid did, the court emphasized that these requirements would not significantlyjeopardize the economic interests of Facebook.237Balancing could at least partially accommodate quasi-state agents’ pursuit ofprivate aims, in part because, unlike direct-and-identical duties, direct-and-lesserduties might not fully negate the corporations’ own expressive rights claims. Instead,the corporations’ rights would simply be diminished in the sense that they could bepartially or fully outweighed by the speech rights of the public. A court might, forinstance, conclude that a non-profit corporation could fire or demote employeesbased on speech counter to the corporation’s goals but not simply based on back-ground ideological disagreements. Or they might decide that corporations could fireemployees in leadership positions based on viewpoints but not other employees.When it comes to social media platforms, courts interpreting the First Amendmentmight follow the lead of Germany and New Jersey and impose procedural dutiesfalling short of impartiality, such as duties to constrain their own discretion in thesuppression of speech, duties to respect due process, and/or duties of transparencyand disclosure.D. Indirect DutiesA quite different option is that quasi-state agents could have indirect dutiesunder the First Amendment. These duties would be “indirect” in the sense that theywould not be self-enforcing but would need to be both activated and defined throughlegislation, regulation, and judicial interpretation. Like the direct-and-lesser-dutiesmodel, this model leaves open the possibility of less demanding duties for quasi-state agents than for full state agents, depending on the choices of the relevantlegislators, regulators, or judges. Peculiar as indirect duties for private might sound,it has been occasionally adopted by foreign courts, U.S. state courts, and the SupremeCourt itself. Below, I consider first how judges could play a role in the articulationof indirect duties, and then how political actors could do the same.1. Judicial DefinitionIn most jurisdictions where judges define indirect duties, they do so by interpret-ing existing non-constitutional law—private law, criminal law, etc.—so that it isconsistent with or promotes constitutional values.238 In Germany, for instance, the236 Bundesgerichtshof [BGH] [Federal Court of Justice], July 29, 2021, Entscheidungendes Bundesgerichtshofes in Zivilsachen [BGHZ] ¶ 85; id. ¶ 97.237 See id. ¶¶ 76, 89.238 See BVerfGE, 1 BvR 400/51, Jan. 15, 1958, https://www.bundesverfassungsgericht.de/SharedDocs/Entscheidungen/EN/1958/01/rs19580115_1bvr040051en.html [https://perma.cc/JGN7-BJM3]; Retail, Wholesale & Dep’t Store Union, Loc. 580 v. Dolphin Delivery Ltd.,[1986] 2 S.C.R. 573, 603–04 (Can.); see also Gardbaum, supra note 204, at 404, 406.2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 671main vehicles for this “indirect” effect of the Basic Law are the undefined and broadlegal terms of private law statutes.239 So, for instance, German courts have interpreteda statutory requirement that each party to a contract in “good faith” not try to un-reasonably disadvantage the other party to demand that the parties respect variousvalues guaranteed by the Basic Law.240 These include dignity and equality,241 as wellas the freedom of speech of all parties and of other persons affected by the con-tract.242 While American private law is mostly spelled out in common law, ratherthan in statutes as in Germany, it, too, could be adapted to constitutional values. Inthe context of quasi-state power over speech, these values might include ensuringthat people have adequate alternatives for speaking freely or that the primary chan-nels of communication provide access to a diversity of viewpoints on public issues.243Indeed, the U.S. Supreme Court has already arguably interpreted private law inlight of constitutional free speech values in one context. In one of its most famouscases, New York Times v. Sullivan, the Court considered a defamation action, a dis-pute of private law: the public safety commissioner of Montgomery, Alabama, suedthe New York Times, in his personal capacity, for publishing an advertisement bysupporters of Martin Luther King, Jr. that made various slightly exaggerated state-ments about actions taken by the Montgomery police against civil rights protestors.244The Court held that the defamation action violated the First Amendment, even thoughthe suit was brought by a private individual, not a state actor.245 (Generally speaking,a court decision in a private law case is not considered “state action.”246) The Courtthen drew on the First Amendment’s purpose of encouraging the robust criticism ofgovernment and public officials to modify common law defamation doctrine: hence-forth, public figures bringing defamation suits would have to prove not only that thechallenged statement was false but also that the defendant made it with “actualmalice,” i.e., reckless disregard for its truth.247 This diverged from the earlier libel rulethat the defendant was liable unless she could establish the truth of her statement.248239 Gardbaum, supra note 204, at 404.240 See BVerfG, 1 BvR 3080/09, Apr. 11, 2018, ¶ 38, https://www.bundesverfassungsgericht.de/SharedDocs/Entscheidungen/EN/2018/04/rs20180411_1bvr308009en.html [https://perma.cc/7SSP-G7DE].241 Id. ¶¶ 40–41.242 Bundesgerichtshof [BGH] [Federal Court of Justice], July 29, 2021, Entscheidungendes Bundesgerichtshofes in Zivilsachen 230, 347–89, paras. 74–75 (interpreting nationalcontract law).243 See, e.g., Red Lion Broad. Co. v. FCC, 395 U.S. 367, 369 (1969) (describing thepurpose of the fairness doctrine).244 376 U.S. 254, 256–58 (1964).245 Id. at 264. One could argue that the state action in the case was simply the substantiveshape of defamation law. Still, the case interprets a duty that one private party owes toanother (not to defame) in light of constitutional principles.246 See Jackson v. Metro. Edison Co., 419 U.S. 345, 349 (1974).247 Sullivan, 376 U.S. at 279–80.248 Id. at 267.672 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615Ultimately, a plaintiff was disabled from using tort law to remedy her injury in theordinary way, because her use of that remedy would stifle free speech interests.States have gone much further in interpreting their common law considering(state) free speech values, often specifically in the case of powerful private agents.California is perhaps the most notable, but not the only, example.249 In Robins v.Pruneyard in 1979, the California Supreme Court interpreted private property rightsto be limited by the California Constitution’s free speech guarantee.250 Specifically,a privately owned shopping mall, which provided an “essential and invaluableforum for exercising those rights,” was blocked from invoking the state’s trespasslaw in the ordinary way against members of the public collecting signatures for apolitical petition on the mall’s property.251 The court observed that property rightsin California had been interpreted to be “responsible and responsive to the needs ofthe social whole.”252 Given “the strength of ‘liberty of speech’ in this state,” thecourt concluded that property rights must give way to the free speech claims of thesignature gatherers.2532. Legislative and Regulatory DefinitionThe responsibility of defining quasi-state agents’ indirect First Amendmentduties could lie not only with courts but also with the political branches. Legisla-tures or agency regulators could give those duties more specificity than judicialinterpretation could and would have more flexibility to craft narrowly drawn dutiesto address quasi-state agents’ actual harms to free speech interests in light ofchanging societal conditions and needs. These statutory or regulatory duties couldalso, unlike indirect duties through judicial interpretation, be enforced even absenta civil lawsuit or criminal trial.This is effectively the approach in Associated Press and Red Lion. The Courtdid not suggest in either of those cases that a member of the public could sue the APor NBC under the First Amendment; that is, it did not suggest that these large mediacompanies had direct constitutional duties. It did, however, give the nod to the Depart-ment of Justice’s antitrust action against the AP and the FCC’s fairness doctrine—even though these policies may well have encroached on these companies’ traditionaleditorial rights.254 It also, in both cases, indicated that these encroachments were249 See, e.g., State v. Shack, 277 A.2d 369 (N.J. 1971) (requiring that a property owner,on principles of freedom of speech, give federal employees access to private property toconvey to workers information about available federal assistance programs).250 Robins v. Pruneyard Shopping Ctr., 592 P.2d 341, 342 (Cal. 1979).251 Id. at 347.252 Id. at 345.253 Id. at 346, 348.254 See Associated Press v. United States, 326 U.S. 1, 19–20 (1945); Red Lion Broad. Co.v. FCC, 395 U.S. 367, 375 (1969).2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 673justified by the First Amendment interest in ensuring the dissemination of a diversityof viewpoints over the essential channels of communication upon which individualsand democratic discourse more broadly depend.255 In other words, the First Amend-ment itself justified the imposition of duties that overrode free speech rights.Today’s mass media and telecommunications companies face many legal con-straints on their power that arguably could be justified under the First Amendment.For instance, common carrier laws impose duties of non-discrimination upon masstelecommunications companies that provide telephone and internet services.256 Whilethese laws are ordinarily upheld by courts because the companies are thought to bemere “conduits” for communication and not actual publishers with editorial rights,257the distinction is somewhat difficult to uphold and has engendered substantial debatewhen it comes to its application to social media platforms.258 A growing number ofvoices are also calling for antitrust actions against Google and Meta.259 Indeed, theDepartment of Justice just filed a new, broader lawsuit against the former allegingthat it is monopolizing certain digital advertising technologies.260 Still others are cal-ling for legislation or regulation that requires the biggest tech companies to followcertain due process and transparency rules in their handling of speech.261 One wayof seeing all of these projects—along with that of the obsolete fairness doctrine—isas enacting a scheme of indirect First Amendment duties against quasi-state agentsin particular.E. Distrust of the StateYet, recall that our justification in Part II for imposing free speech duties onquasi-state agents was the terrible threat that the state itself poses to free speechinterests. It might seem odd, having rooted our analysis in this point, to turn around255 Associated Press, 326 U.S. at 20; Red Lion Broad. Co., 395 U.S. at 389, 392.256 For details on these laws, see generally Yoo, supra note 32.257 See, e.g., U.S. Telecom Ass’n v. FCC, 825 F.3d 674, 741 (D.C. Cir. 2016); see alsoStuart Minor Benjamin, Transmitting, Editing, and Communicating: Determining What “TheFreedom of Speech” Encompasses, 60 DUKE L.J. 1673, 1686–87 (2011).258 See Bhagwat, supra note 195, at 2382 (summarizing Sabeel Rahman’s argument). Seegenerally Volokh, supra note 5 (arguing that social media platforms could be classified ascommon carriers specifically with respect to their hosting of speech); Yoo, supra note 32.259 See, e.g., Lina M. Khan, Sources of Tech Platform Power, 2 GEO. L. TECH. REV. 325(2018).260 Press Release, U.S. Dep’t of Just., Justice Department Sues Google for MonopolizingDigital Advertising Technologies (Jan. 24, 2023), https://www.justice.gov/opa/pr/justice-department-sues-google-monopolizing-digital-advertising-technologies [https://perma.cc/4XAE-AL8Z].261 Lakier, supra note 127, at 2371 (describing regulations that might enact free speechvalues).674 WILLIAM & MARY BILL OF RIGHTS JOURNAL [Vol. 32:615and hand more power over free speech interests to the original agent of concern.Indeed, entrusting state actors to regulate speech in any way—even how privateagents treat that speech—cuts against core First Amendment tradition.However, note that our situation involves a clash of powers: free speech inter-ests will be severely threatened, one way or another, either by a private power or apublic power. In this stalemate, there are reasons to release the constraints on thepowerful democratic agent. First, the release would be for the narrow purpose ofconstraining the powerful, undemocratic agent. Entrusting the state to define theindirect First Amendment duties of quasi-state agents would not give the state ablank check to regulate however it pleases. Those duties could not, for instance,require promoting or demoting any particular viewpoint. For instance, Red Lion wasclear that the First Amendment restrained the FCC from requiring that “the officialgovernment view dominat[e] public broadcasting.”262Second, a democratic state, as described in Part II, is subject to certain pressuresto conform to these limitations on its power. Consider a worst-case scenario inwhich a broad-power quasi-state agent, like the most widely used social mediaplatform, intervenes just prior to an election to significantly tilt public discourse infavor of a particular candidate or party. If the state is not allowed to regulate, thereis nothing more to be done and no way to prevent further such abuses of power. Bycontrast, if the public version of this worst-case scenario happened and the FCCabused its power to regulate the quasi-state agent in order to drastically tilt publicdiscourse in another direction, then there are “higher” authorities that could step into stop the abuse or prevent further such abuses—from the other branches of govern-ment to the voters themselves.263CONCLUSIONThis Article has argued that the inflexible line in First Amendment doctrinebetween state and non-state agents is an empty formalism. If the freedoms of speechand the press matter, it is because they protect the ability of individuals to expresstheir opinions freely and for public opinion to form freely. This means that whoeverhas the power to disrupt these outcomes flouts those freedoms. In nearly every case,the state has this power. This explains why courts have applied the First Amendmentto all agents backed by the power of the state and not just to Congress, as theconstitutional text demands. But at least in our time, the state is not the only agentwith this power.262 395 U.S. 367, 396 (1969).263 Of course, voters are limited in their time and ability to gather information aboutgovernment conduct, understand the political and policy issues at stake, engage in soundcausal reasoning, and free themselves of biases. See CHRISTOPHER H. ACHEN & LARRY M.BARTELS, DEMOCRACY FOR REALISTS: WHY ELECTIONS DO NOT PRODUCE RESPONSIVEGOVERNMENT 1–4 (Tali Mendelberg ed., 2016). But they provide a partial check on at leastthe most severe abuses.2024] THE PRIVATE ABRIDGMENT OF FREE SPEECH 675I have made the case that two types of private agents also hold the power tothwart First Amendment speech rights and thus should be held—in some manner—tobe bound by those rights. Placing a functional inquiry about power, rather than aformalist inquiry about state status, at the heart of free speech doctrine offers severaladvantages. First, it renders the doctrine coherent with the purposes that ostensiblymotivate it, to protect the free expression of individual opinion and the free forma-tion of public opinion. Second, it explains multiple threats to free speech in themodern world, from social media companies to especially powerful employers.Third, unlike much contemporary thinking about powerful media companies, itleaves open the possibility that the activities in which these powerful agents areengaged may be, on some level, speech. This matters when less powerful agentsengage in them.Once First Amendment doctrine has embraced a more functional line betweenprivate and public agents, it confronts the next task of deciding how exactly therelevant private agents are to be bound. This Article has offered several models fordoing so, drawing on the experience of both domestic and foreign jurisdictions.Each provides a different level of flexibility in accounting for the reality that evenstate-like agents are still not exactly the state. |
|---|