A Contractual Approach To Social Media Governance

Introduction

Social media’s impact on individual rights and well-being, social stability, and democratic resilience has bred heated scholarly debate and a hodgepodge of regulatory regimes. Attempting to find ways to ameliorate the often harsh results of this impact, scholars and policymakers have been struggling to design governance regimes for social media that will, on the one hand, hold platforms accountable for the harms they generate and, on the other hand, avoid burdening them with crippling regulation that could stifle innovation and inhibit public debate. The vast majority of the scholarship in the field has adopted a public law approach, using public law language to analyze relevant questions, public law norms as the guiding compass toward reform, and public law tools for legal intervention.[1] Captured in this public law bias, the scholarship has devoted far too little attention to private law, generally, and contract law, specifically, as potential sources of legal norms for social media governance.[2] It has consequently yielded an incomplete analysis, that fails to address all relevant aspects of the issues at hand and overlooks potentially available legal tools for effecting improved governance norms. Furthermore, the lack of private law-based analyses in the literature has left courts substantially unequipped for adjudicating social media-related disputes and providing redress for ensuing harms.

Addressing this blind spot, this Article proposes a novel contractual approach to social media governance and adjudication. It seeks to show how better employment of the various resources of contract law in the context of social media could increase platforms’ accountability, without resorting to regulatory oversight over their decision-making; bolster the protection of individual and group rights, while accommodating platforms’ legitimate pursuit of profit; and, more broadly, bring greater alignment between the interests of platforms and users, as well as the interests of the public at large. For this purpose, the Article first sets out to dispel prevalent misunderstandings regarding the nature of the contractual relationship between platforms and their users, which is often far too narrowly construed. In contrast to commonly held views, this relationship cannot be understood based solely on Terms of Service agreements, and platforms’ services should not be construed as provided free-of-charge. Rather, social media contracting establishes a complex ongoing barter, that is profoundly distinct from typical consumer transactions. This barter, in which platforms’ services are traded for users’ non-fungible data and attention, is inevitably premised on platforms’ best-efforts commitments, whereby norms and expectations are created and continuously adjusted in a myriad of forms and can never be fully reduced into text.

Based on the suggested contractual analysis, the Article then turns to offer a novel consideration of the prevailing judicial approach to platform immunity with respect to content moderation, as it pertains to contractual causes of action, and argues that courts have erroneously granted platforms overly sweeping protection in this regard. This assertion applies to the judicial application of both contractual immunity, articulated in limits on liability clauses in platforms’ Terms of Service agreements, as well as statutory immunity under Section 230 of the Communications Decency Act of 1996.[3] Contractual causes of action, it is argued, should not be categorically barred. Rather, courts should extend immunity only to good-faith efforts undertaken by platforms to uphold their commitments with respect to content moderation, so as to avoid imposing strict contractual liability. Arbitrary, self-serving moderation choices that violate platforms’ contractual obligations should not be protected, and platforms should not have a right to make fraudulent misrepresentations about the services they sell. Not only does this discussion significantly contribute to the ongoing debate over platform immunity, it is also most timely given the apparently growing willingness of the Supreme Court to consider, for the first time, the proper scope of Section 230’s statutory immunity.[4]

The suggested contractual analysis of platform-user relationships also exposes the stated divergence of social media contracting from traditional consumer contracts. This divergence, the Article contends, mandates a parallel divergence in contract law analysis and application in order for courts to properly adjudicate contractual disputes between users and platforms. In other words, the Article shows that traditional contract law as customarily applied in the consumer-seller context is simply incompatible with the user-platform contractual relationship. Addressing this discrepancy, the Article comprehensively employs relational contract theory to identify emerging contractual norms between users and platforms and derive applicable legal duties. The suggested classification of user-platform relationships as relational contracts of a sort has already received some qualitative empirical support in the past.[5] This Article, however, is the first to offer a robust normative analysis of the applicability of relational contract theory to the social media context. Importantly, it is also the first to attempt bridging the gap between relational contract theory and contract law doctrine in this context, equipping courts and litigators with practical legal tools to address social media harms. By combining relational contract theory with contract law doctrine, the Article invites a thorough reimagining of the contractual duties social media platforms should be held to account for. It calls to subject platforms to duties of fairness and diligence in providing their services and offers the necessary tools to pursue this goal both cautiously and effectively through private law litigation.

The Article proceeds as follows: Part I explores the advantages of a contractual approach in the social media context, seeking to address the lopsidedness of the current prevailing analysis in the legal literature. Then, by examining the Terms of Service (ToS) agreements,[6] community guidelines,[7] and data policies[8] of three major platforms—Facebook (now operated by Meta), Twitter (now X), and YouTube—alongside other written arrangements and additional materials manifesting platforms’ conduct and perceptions, Part II sets out to clarify several ambiguities regarding the nature of the user-platform relationship that have substantially influenced case law, policymaking, and the legal scholarship in the field.

Part III addresses a crucial preliminary issue: should a contractual cause of action against social media platforms’ content-related decision making be barred by statutory immunity or, alternatively, defeated by platforms’ contractual exculpatory clauses? Through a combined doctrinal and normative analysis, this Part shows why reading into Section 230 robust immunity from contractual causes of action is untenable, as is unnuanced enforcement of exculpatory clauses. Indeed, Section 230 was designed specifically to bar causes of action in torts, not contracts, and it was certainly not legislated to allow platforms to make fraudulent misrepresentations about the services they provide. Furthermore, as argued in this Part, previous judicial interpretation of platforms’ exculpatory clauses has been heavily misguided, inter alia, in its adherence to the illusion that this is a contractual relationship in which platforms’ services are provided for free.

Part IV lays out the theoretical foundation for a contractual approach to social media governance and for identifying contractual norms in social media contracting. The discussion will show that unlike the classic model of arm’s-length consumer-seller transactions, social media contracting presents inherently relational structures and characteristics,[9] thereby warranting a different legal response. It is inevitably based on an incomplete and dynamically adaptable arrangement; it is aimed at facilitating long-term cooperation between users and platforms, which is premised on interdependency, mutuality, and reciprocal respect for the parties’ competing interests; it adopts an ongoing fair cooperation model, based on bilateral transactions, to create joint exchange surplus; and it inherently generates expectations for the protection of human rights and procedural fairness.

Part V provides a doctrinal analysis, offering practical tools for ensuring platforms’ compliance with contractual norms and obligations through adjudication. It shows that adopting the proposed contractual approach would enable courts to impose fairness and diligence duties on platforms with respect to both their content moderation and data-related practices. These duties would require that in providing their services, platforms engage in a constant balancing of interests—their own self-interests, users’ interests, and the public interest—and diligently employ adequate means and methods to achieve the appropriate balance. By applying these duties to content moderation and data-related practices, courts would be able to address some of the problems associated with platforms’ business model, and norms such as human rights and procedural fairness could be injected into the relational apparatus and accordingly legally enforced. However, to mitigate the risk of excessive judicial encroachment on platforms’ decision-making, this Part also proposes novel doctrinal tools that could conceivably guarantee platforms sufficient leeway in their decision-making while at the same time allow for judicial review of systemic design choices and for individual error correction in extreme cases. Such doctrines, it is suggested, can be developed through adoption of a deferential standard of review—a “platform judgment rule”—and by preferring equitable remedies to damages absent bad-faith decision-making.

I. The Complementary Roles of Public and Private Law in Social Media Governance

In recent years, a public law approach has largely dominated the legal debate over social media governance. Public law analogies and terminology are customarily deployed to describe social media and to justify the adoption of governance norms rooted in public law’s normative theory. As part of this discourse, social media platforms are often described as governors that perform judicial, administrative, or legislative functions, and, accordingly, bodies of law associated with public law (e.g., human rights law, administrative law, constitutional law), are invoked. Such analogies and strategies have provided analytical resources for evaluating the legitimacy of governance regimes and have led scholars and policymakers to propose regulatory frameworks for shaping governance norms that draw heavily on the public law arsenal of potential legal responses. However, as I explain below, despite the important contributions of this public law approach to understanding the intricacies of social media, its dominance has resulted in an incomplete picture, which omits relevant aspects of the controversy and important existing legal tools for effectuating desirable governance norms. Thus, as I argue below, a robust private law approach must be developed to supplement the public law analysis, and as suggested in this Article, contract law is a suitable body of law to start from.

Indeed, social media is, first and foremost, created by private entities that establish a complex network of relationships through contracts. Thus, as we analogize social media platforms to various state institutions, it is important to remember that platforms are, above all, private corporations that sell a service to private customers in accordance with a contract that allegedly binds both parties. This relational structure has both normative and legal implications that are all too often ignored. As a matter of both fact and policy, platforms’ contractual obligations toward users, from which governance norms can and should be derived, can never be fully described or properly addressed without a contract law analysis of the platform-user relationship.

A. Social Media Governance Through a Public Law Lens

This Section will present the public law bias that characterizes recent mainstream legal scholarship on social media governance. In a recurring process, this scholarship offers to replace one public law analogy with another, hoping to provide more accurate depictions of social media and more nuanced suggestions for public-law-based legal regimes to regulate its activity. This scholarly approach is certainly fundamental for understanding the field and, as explained below, has provided invaluable insights about social media’s modus operandi and on public perceptions regarding its legitimacy. However, such findings and insights should also inform a parallel private law analysis. A private law approach can better address horizontally developing norms governing interpersonal relationships[10] and provide a different toolset that would enable the mobilization of the judicial system for regulating these norms through private action. Such an approach would, therefore, be capable of addressing harms social media generates in novel ways.

Undoubtedly, one of the most influential public law analogies offered by legal scholars in this context is Kate Klonick’s depiction of social media platforms as “New Governors.”[11] From this perspective, platforms should be understood and imagined as engaged in governance,[12] and content moderators as acting “in a capacity very similar to that of a judge.”[13] This governance analogy, Klonick suggests, should supplant earlier public law analogies proposed for describing platforms’ activity, which, she contends, are inappropriate: namely, the First Amendment categories of state actors, editors, and broadcasters.[14] Perhaps to temper her analogy, Klonick also stresses that the New Governors’ motivation to “govern well” differs from that of traditional governors: the former is rooted primarily in economic incentives[15] that constantly push platforms to abide by users’ expectations and norms,[16] so as to foster their trust and increase their engagement.[17] Only by understanding platforms as governors, Klonick asserts, can we design an appropriate regulatory regime. As she argues in this respect, platforms’ underlying motivations render self-regulation and non-intervention the preferred policy approach, with intervention desirable in exceptional cases, such as where “equal access” or “democratic accountability” are lacking.[18]

Klonick’s governors metaphor greatly influenced the legal and political discourse on online speech regulation,[19] and understandably, notable scholarship, like the “digital constitutionalism” literature,[20] sought to extend it further. For instance, as Nicolas Suzor explains, if platforms are, indeed, governors, then the legitimacy of their decision-making should be evaluated through the lens of rule of law values like “[m]eaningful consent, equality and predictability, and due process.”[21] This rule of law perspective, originating in public law, supplies, according to Suzor, the necessary terminology for addressing concerns about “the relationship between platforms and their users”[22] and for informing the necessary “political discussion” about platforms’ responsibility to “protect the rights of users and how these responsibilities might be enforced.”[23]

Another public law analogy, offering additional layers to the comparison between platforms and states, is that of “platform law.” This term was originally coined by David Kaye, UN Special Rapporteur for the Promotion and Protection of the Right to Freedom of Opinion and Expression, who has dedicated significant amounts of his work to the introduction of human rights principles to social media governance. In Kaye’s framework, the term “platform law” was used to generally describe platforms’ opaque decision-making, policies, and practices with respect to content moderation, making them, according to Kaye, “enigmatic regulators.”[24] Later, Molly Land offered a more detailed articulation of platform law,[25] seeking to capture the multiplicity of norm-generating arrangements in platform-user relationships, through which social media governance is facilitated. These arrangements, she stresses, include not only Terms of Service agreements, but also “a range of substantive and procedural rules about user activity and remedies.” They are composed of “the formal and informal rules” generated by platforms to “govern the rights and liabilities of their users.”[26] In Land’s conception of “platform law,” the different norms which govern platform-user relationships are clustered into what can arguably be described as bodies of law, such as the platforms’ “contract law,” “procedural law,” etc.[27] Then, again adopting a human rights perspective, Land urges platforms to afford users “greater choice over the norms by which they want to be governed.”[28]

The depiction of social media platforms as all-powerful state-like “regulators,” who are engaged in “law-making” and “law-enforcement,” and the continuous search for legal and analytical tools for scrutinizing their decision-making, have led many—much like Kaye and Land—to enlist international human rights law (IHRL), as opposed to contract law for example,[29] to the struggle over social media governance. Under this approach, the IHRL version of the right to freedom of expression (as enshrined in Article 19 of the International Covenant on Civil and Political Rights[30] and in accordance with the United Nations Guiding Principles on Business and Human Rights[31]) was mobilized to provide a workable normative and doctrinal framework, under which “any restriction of an individual’s right to freedom of expression must satisfy the tripartite test of legality, legitimacy, and necessity.”[32]As Brenda Dvoskin thoroughly describes in this respect, “[a] large consensus among companies, scholars, and advocates has emerged” around making human rights law the touchstone of online speech governance.[33] This approach has thus not only gained significant traction in scholarship,[34] it was also explicitly adopted by dominant platforms, effectively influencing their moderation policies.[35]

Human rights law accordingly became both a practical standard for moderation-related decision-making, and a shibboleth for governance norms’ legitimacy, illustrating the powerful influence of public law reasoning on the field. But while the human rights approach is adept at handling individual cases, it seems to have its limitations when addressing the broader and more systemic issues relating to content moderation. Indeed in many respects, the human rights approach rests on what Evelyn Douek recently described as the “stylized picture of content moderation.”[36] This picture, according to Douek, depicts content moderation “as a process in which social media platforms write a set of legislative-style substantive rules and apply them in individual cases . . . similar to the offline justice system.”[37] However, she argues, this depiction is “misleading and incomplete”[38] since “many of the most important decisions” actually happen “upstream . . . before any individual content moderation case even arises.”[39] The diversity, pace, scale, and growing complexity of platforms’ governance mechanisms, argues Douek, make an individual rights-based public law analogy inappropriate, and it should therefore be rejected in favor of an improved one: content moderation as a system of “mass administration”[40] which is in many respects analogous to “administrative agencies.”[41] Douek’s suggested approach to content moderation thus seeks to focus on “ex ante institutional design choices,” and to hopefully better address complex governance questions for which an individual rights-based approach is unsuitable.[42] More practically, Douek proposes regulatory intervention that draws on “principles and practices of administrative law,” which are particularly suited “to bring oversight and accountability to massive unelected bureaucracies that determine rights and interests at scale within complex systems.”[43]

But whether public law-based proposals overly focus on individual rights, or rather, sufficiently include an institutional-aggregatory dimension, all share one similar blind spot: they assume away interpersonal norms of private relationships, alongside the law governing them (i.e., private law).

B. The Need for Private Law

Private law’s focus on interpersonal relational norms, its different institutional manifestation, and its distinct moral underpinnings, often make it more suitable—compared to public law—for addressing many of the questions concerning social media governance and harms. Leveraging the resources of private law to formulate effective strategies in this domain can yield significant consequences, offering both normative guidance and practical benefits. To be clear, this Section does not attempt to resolve the eternal dispute over the “true” distinction between private and public law.[44] Nor does it provide an exhaustive examination of private and public law’s comparative advantages within the context of social media. Instead, it draws attention to three particular aspects of the discussion around social media governance which emphasize the imperative need for a complementary private law analysis.

First, from a juridical perspective, the current debate around social media governance seems to have resulted in a certain irony: on the one hand, legal scholars deploy various analogies to devise strategies for protecting users’ expectations[45] and concur that platforms and users are private entities bound by a contract; yet on the other hand, contract law is hardly ever considered as a tool for achieving one of its most fundamental purposes—protecting the expectations of parties engaged in voluntary private interactions. This result is perhaps not surprising when looking at how legal concepts such as “contract,” and more generally, “contract law,” are often depicted in leading scholarship—either narrowly construed, reduced to platforms’ Terms of Service agreements,[46] or else ignored altogether. However, as this Article hopes to show, this perception of platform-user contractual relationships and of contract law’s doctrinal and theoretical reach is a far cry from their real characters.[47]

Second, from an institutional perspective, proposed tools for influencing social media governance tend to include either suggestions for imposing norms through regulatory or legislative action or, conversely, calls for the voluntary adoption of governance norms by platforms. Thus, in line with typical public law deliberation, the institutional inquiry predominantly assumes a bipolar structure, examining direct regulatory measures on the one hand, and self-regulation on the other, given constitutional limitations to the state’s police power. This approach accordingly often discounts, and occasionally completely ignores, the potential role courts may play in shaping governance norms through private law adjudication, without offering thorough analysis to support this position. The institutional opportunities private law litigation may present thus remain generally underexplored.

Third, from a moral perspective, the value of commitment or promise in interpersonal relationships is also often disregarded,[48] and the legitimacy of platforms’ conduct is repeatedly measured against public law perceptions of worthy rulemaking. This public law-rooted moral approach has real-life implications, as evidenced for example in Klonick’s empirical findings regarding the motivations behind platforms’ decision-making. “[T]he primary reason companies take down obscene and violent material,” she found, “is the threat that allowing such material poses to potential profits based in advertising revenue.”[49] Two additional motivations, Klonick identified, are platforms’ commitment to American free speech norms and perception of corporate responsibility.[50] Missing from Klonick’s description, however, is the influence of moral disdain for a different archetype of platforms’ failures, which has nothing to do with the “appropriateness” of content, free speech, or corporate responsibility: platforms’ failure to simply keep their word.

Focusing the moral reproach on platforms’ breach of a promise, rather than on the failure of for-profit companies to be “good” public agents or “serve” democracy and free speech, makes a difference. As Klonick showed, the moral framework through which platforms’ behavior is evaluated directly influences their actions, one reason being that moral reproach is a good cause for reputational backlash—a genuine concern for any for-profit company, including social media platforms. Moreover, the case for criticizing platforms for misleading users is arguably less controversial, and thus can be more effective: it is less paternalistic and more respectful of the platforms’ own liberties; and it is more attuned to the economic reality of private corporations that legitimately, we typically hold, prioritize profit maximization in doing their business over pursuing general conceptions of the public good. Contract law’s distinct moral underpinnings thus enjoy a practical advantage: moral criticism for a failure to keep a promise, as opposed to a failure to serve the public good, is almost always more justifiably translated into legal measures of redress.[51]

Several scholars have indeed proposed private law tools and approaches to tackle issues related to social media harms, and accordingly sought to realize some of their stated advantages.[52] Nevertheless, these contributions remain relatively modest in scope and influence when compared to the extensive body of public law scholarship in this field.[53] This general disregard for the private-horizontal dimension of social media governance and the potential influence of private law on its unfolding norms is both unjustified descriptively[54] a nd questionable normatively. In the face of ever-increasing public law abstractions, a comprehensive private law approach can offer more concrete grounds for conceptualizing governance norms for social media. Thus, instead of—or rather, in addition to—imagining social media platforms as state-like governors engaged in mass administration, adjudication, and legislation of platform “law,” that are subject to rule of law principles and international human rights, I propose conceiving of platforms as private corporations performing contractual obligations, that are subject to the edicts of contract law. This imaginative reconstruction, which might be considered radical in the current legal and scholarly environment, could be highly consequential, as further explored below.[55]

II. Social Media Contracting

The contractual relationship between social media platforms and their users cannot be understood based solely on Terms of Service (ToS) agreements. The classic model of consumer transactions is constructed on unilateral provision of goods or services in return for monetary remuneration, often based on boilerplate contracts that articulate most of the rules and norms governing the relationship. Social media contracting, however, diverges significantly from this model of transaction. The systematic analysis of the contractual arrangement governing platform-user relationships presented below will show that social media contracting establishes an ongoing barter, whereby norms and expectations are created and continuously adjusted in a myriad of forms and can never be fully reduced into text.

The analysis below begins with a brief outline of the sources of contractual norms and obligations in social media and their complex and dynamic interplay. It then proceeds to examine the structure of the contractual relationship and its three constitutive elements—the promised benefit, the standard of liability, and the contractual consideration—to respond to three respective questions: What service do platforms commit to provide? What would constitute a breach of that commitment? And what do users commit to pay in return for the service offered by the platforms?

A. The Sources of Contractual Norms in Social Media Relationships

To become a user on a social media platform, one must first consent to the platform’s Terms of Service.[56] However, ToS agreements, structured as boilerplate, or standard-form contracts, do not present the entire contractual arrangement and are supplemented by an array of norm-generating mechanisms. ToS agreements, as Tarleton Gillespie explains, “spell[] out the terms under which users and platform interact, the obligations users must accept as a condition of their participation, and the proper means of resolving a dispute should one arise.”[57] They address “not just appropriate content and behavior but also liability, intellectual property, arbitration, and other disclaimers.”[58] These agreements incorporate a variety of supplementary documents that elaborate on the parties’ rights and obligations. Such documents include the platform’s data policy[59] and community guidelines,[60] which are “written communications used to convey platforms’ moderation policies in an elaborative manner” and their “expectations of what is appropriate and what is not.”[61] They also typically include additional written arrangements that purport to govern other functionalities of the platform.[62] Together, all these written arrangements form a substantial body of rights and obligations pertaining to the interacting parties. They are all parts of the contractual relationship and operate in concert to generate and convey contractual commitments:[63] community guidelines, data policies, and other documents provided by the platform elaborate on the contents of contractual norms and obligations,[64] while the ToS agreement simply defines those contents as obligatory.[65] These documents, however, do not, and cannot, articulate the full array of contractual norms and obligations that govern platform-user relationships, particularly due to the innate indefiniteness of the service involved.[66] The parties’ correlative rights and duties are thus often dynamically conveyed in a variety of other means and mechanisms, during the course of performance, encompassing multiple forms of written and conduct-based communication:[67] they are embedded in the platform as part of its code;[68] mentioned in internal policy guidelines;[69] expressed through internal moderation procedures followed by platforms;[70] conveyed in public announcements, or official news releases with explicit declarations on updated policies;[71] and even posted on the platform by its CEO.[72]

Accordingly, while the ToS and associated documents provide a foundational contractual framework, the true essence of the platform-user relationship extends beyond these written agreements. In practice, platforms’ design and policy choices directly impact users’ rights and obligations and, consequently, inevitably affect their expectations about the functionality of the platform just as much as explicit rules do.[73] To illustrate, imagine if Twitter (now X) were to adopt a new internal moderation rule that instructs human moderators to remove any pro-choice content, with no right of appeal for users, and to update its algorithm to filter such content without notice. Such a rule would directly impede users’ rights of use on the platform, irrespective of whether the change was communicated explicitly in a written document stating, say, “Pro-choice content is not allowed.” Algorithmically empowered downranking,[74] or “shadowbanning,”[75] which covertly makes posts invisible to other users, produces a similar outcome.[76] These examples illustrate that contractual obligations and users’ expectations are created and adjusted regardless of the form in which moderation policies are conveyed. The contractual analysis of platform-user relationships must, therefore, take into account both the plurality of instruments used by platforms to create and adjust contractual expectations, as well as their ensuing substance. Thus, regardless of whether we relate to the multiplicity of norms governing platform-user relationships as “platform law” or simply “contract,” users base their expectations on a multitude of written arrangements and, to a large extent, on platforms’ conduct. To provide a comprehensive and substantive account of social media contracting, therefore, a contractual analysis of the parties’ relationship must give due consideration to such terms, implied or explicit, and to their dynamic nature.

B. The Elements of Social Media Contracting

1. The Service

Scholarly discussions of social media governance rarely consider the relationship between the adopted governance norms and the actual contractual arrangement between the parties. However, to understand the obligations imposed on platforms by the contractual arrangement and how they shape their decision-making, we must first clarify what services platforms contract to provide, for which users are willing to give consideration. The short answer, in many respects, is content moderation. It would thus be wrong to imagine that content moderation policies, practices, and design choices emerge and develop in isolation from the contractual setting: they impact the very essence of the service provided to users, generate users’ consent,[77] and accordingly shape[78] and are shaped by[79] contractual norms and expectations. In other words, platforms’ decisions with respect to moderation policies so inherently affect users’ rights because they are, in many respects, the very service bargained for.

The claim that content moderation is the very service social media platforms provide is not novel. Tarleton Gillespie, for example, explained in this respect that content moderation is the activity that truly distinguishes social media platforms from other types of online content carriers.[80] And accordingly, Gillespie concludes, “moderation is the essence of platforms, it is the commodity they offer.”[81] From a contractual perspective, however, Gillespie’s statement requires a certain qualification. Social media platforms arguably offer more than content moderation. At a more fundamental level, social media platforms offer “connectivity” or “access” to a network of people who wish to share content with one another. This function, however, as Gillespie rightfully stresses, is also shared by other online connectivity providers, including Internet Service Providers (ISPs), Instant Messaging services (IMs),[82] email services (e.g., Gmail), and even web hosting services (e.g., Amazon Web Services). Accordingly, it is insufficient to define social media services based on network access or connectivity alone. Simultaneously, the presence or absence of content moderation is also an insufficient criterion.[83] Instead, the nature of the service is determined by the specific interaction it seeks to facilitate between technology, people, and rules. That is, the service is defined 1) by the technologies and design choices it employs to enable the creation and dissemination of content, 2) by who participates in these activities, and 3) by how such participation is regulated. “Content moderation” at face value pertains only to the latter aspect of the definition. But at least in the social media context, given its intensity and scope, content moderation subsumes all other aspects of the service, and gives social media platforms their essential nature.

Content moderation as employed by social media platforms encompasses a long list of actions and methods: Platforms remove, filter, and suspend users and content.[84] They reduce or amplify distribution and virality,[85] shadowban[86] posts or users, or recommend them to others. They fact-check and flag problematic content. They curate content and create personalized feeds to increase engagement. More generally, social media platforms “actively and dynamically, tune the unexpected participation of users, produce the ‘right’ feed for each user, the ‘right’ social exchanges, the ‘right’ kind of community.”[87] As correctly defined by James Grimmelmann, content moderation is the “governance mechanisms that structure participation in a community to facilitate cooperation and prevent abuse.”[88] It follows, then, that the unique nature of social media is not defined by the mere existence of rules that regulate online interaction. Rather, it is defined by the degree to which and the techniques by which social media platforms moderate, as well as the goal such moderation is designed to achieve—forming an online speech “community.”

This analysis sheds light on the correct interpretation of the services social media platforms undertake to provide. Facebook’s ToS agreement, for instance, opens with the following declaration about its products and services: “Meta builds technologies and services that enable people to connect with each other, build communities, and grow businesses.”[89] Similar declarations are made by Twitter (now X)[90] and YouTube[91] in their ToS agreements. Yet while social media certainly “enables” users to “connect,” and despite the seemingly neutral formulation of these statements, this is by no means an impartial service, devoid of value judgments.[92] On the contrary, value judgments are inevitably made by platforms, at incredible scale and pace. These moderation judgments are embedded within the technology, shape all participation, and are crucial for the community that platforms wish to facilitate to be able to function. As Grimmelmann stressed, “[j]ust as town meetings and debates have moderators who keep the discussion civil and productive, healthy online communities have moderators who facilitate communication.”[93] Indeed, when moderators “do their job right, they create the conditions under which cooperation is possible.”[94]

Thus, what truly distinguishes social media platforms is not merely the existence of content moderation, it is the fact that all aspects of the interaction they facilitate—the content itself, the access to the content, as well as its active dissemination—are heavily moderated in a wholly non-neutral manner for the purpose of community building. Social media platforms do not simply provide content dissemination services. They provide moderated content dissemination services, which are designed to facilitate and incentivize online interaction in a virtual community of a prescribed nature. In this virtual environment, users cultivate their ‘virtual existence’ through profile building (anonymized or identifiable) and engage each other, either actively or passively, as members of the community. When a user logs on to social media, she accordingly seeks a service that allows her to immerse herself in an environment of content that, while not generated by the platform, has been moderated by it, and she expects this environment to have a certain nature dictated by the platform’s choices regarding content moderation. In a sense, therefore, social media platforms offer a doorway to Narnia of a sort; and content moderation defines “the essence” of this virtual kingdom, which lies beyond the screen.

What do these conclusions mean for real-world contractual analysis? From a contractual perspective, community guidelines should therefore be viewed as a primary mechanism by which platforms make representations about the material attributes of the services they sell. However, community guidelines are used only to convey generally applicable principles, which are illustrated only sporadically with specific examples. The guidelines do not, and cannot, prescribe solutions to every concrete issue that may arise.[95] As Klonick explained, “the internal rules of content moderation are iteratively revised on an ongoing basis, and much more frequently than the external public-facing policy.”[96] Thus, although adaptations of content moderation rules may occasionally be expressed through revisions of community guidelines, they are more commonly conveyed to users in other forms, thereby dynamically adjusting users’ expectations.[97]

Moreover, as Douek for instance shows, a comprehensive assessment of a platform’s content moderation service cannot be based solely on the substance of its explicit moderation policies. She explains that content moderation is “a complex and dynamic system”[98] that involves diverse methods, various actors, and many tradeoffs for platforms to achieve the particular balance in content they seek.[99] Each of these features of content moderation impacts users’ expectations regarding the service they receive from the platform: about whether the platform provides automated filtering; about whether it flags or downranks controversial content; about whether it outsources fact-checking to third parties and the identity of those third parties; about the strictness/leniency of the platform’s take-down policies on specific topics (e.g., nudity, anti-vax, political misinformation, and antisemitism); about a right to appeal, whether appeals are heard by external adjudicators, the professional credentials of those adjudicators, and whether their decisions are binding on the platform; or even about whether the platform outsources content moderation, wholly or partially, to the users themselves, allowing them greater control over their feed.

All of these institutional contexts and interrelationships are often more crucial for understanding and evaluating content moderation services than the actual substance of the moderation policy. This is even more cogent considering that “when it comes to details . . . the guidelines at the prominent, general-purpose platforms are strikingly similar,” at least with respect to the general categories of prohibited content.[100] Platforms commonly prohibit content of certain types, including sexual content, graphic violence and obscenity, harassment, hate speech, illegal activity, and promotion of self-harm.[101] True, each platform may define each category slightly differently[102] and take different measures to address it. But the similarities in categories of prohibited content are an indication not only of certain common “red lines” regarding permissible content on dominant platforms,[103] but also that a moderation system’s structure and efficacy may be a more significant factor than the actual policies in users’ choice from amongst alternative platforms.

From a contractual perspective, there are two important upshots to focusing on content moderation as the primary service provided by platforms. First, the inability to reduce all potential situations into specific rules, alongside the intrinsic elusiveness of moderation standards, renders social media contracting inherently incomplete with respect to the core attributes of the provided service. The materialization of unexpected contingencies will inevitably require dynamic adaptation and adjustment of existing moderation policies during the lifetime of the contractual relationship. Second, platforms provide a robust system of content moderation and not only community guidelines. Users’ expectations accordingly derive equally from platforms’ moderation policies, practices, and institutional design. Both of these attributes are stark divergences from the standard consumer contract.

2. The Standard of Liability

The expected standard of contractual liability in social media contracting is particularly significant for understanding the contractual norms governing platform-user relationships. Users understand that they are bound by moderation policies and expect that the content to which they expose themselves will be moderated in accordance with those policies. At the same time, however, users understand that error-free content moderation is implausible[104] and that flexibility and discretion in policy enforcement are an inherent and inevitable part of the social media contractual setting. What behavior, then, on the part of either party would constitute a breach?

Similarly to many standard consumer contracts, social media contractual agreements include various provisions that refer to dispute resolution and liability. These provisions are generally intended to exempt platforms from liability,[105] “arguably written with an eye toward avoiding future litigation, often indemnifying the company as broadly as possible against any liability for users’ actions.”[106] For example, Facebook’s Limits on Liability clause in its ToS agreement includes a disclaimer of all warranties to the extent permitted by law, alongside a justification for the exemption: the platform’s material inability to control the content posted by all users.[107] Similar provisions are included in YouTube’s[108] and Twitter’s[109] ToS agreements.

The inclusion of exculpatory clauses and warranty disclaimers in one-sidedly dictated standard-form agreements is not novel to the social media context, and their validity is examined below.[110] For now, it suffices to highlight the role these provisions play in creating expectations regarding the contractual standard of liability—i.e., the extent of the parties’ obligation to provide a contractually required outcome. Many, if not most, consumer contracts are based on a strict liability principle, requiring one party to provide a specific product or service to the other, in return for a sum of money. This is the typical standard of liability in consumer contracts negotiated at arm’s length. However, some contracts may set a different type of standard and replace the demand for strict performance with a “best-efforts” commitment.[111] Social media contracting inevitably provides for this latter type of performance expectations and standard of liability.

Given the scale, speed, fluidity, contestability, and contextual nature of online content moderation, platforms cannot possibly guarantee full control, ex ante, over each and every item of content uploaded to their network, nor can they provide “perfect” ex post correction. They can only commit to making their best effort at applying represented moderation policies and to setting up an operating system of content moderation aimed at striking an optimal balance between often contradicting considerations.[112] Content moderation is, therefore, inevitably offered on a “best efforts” basis and not as a “guaranteed result” service.[113] Exemplifying this is Facebook’s Limits on Liability clause, which begins with the following declaration: “We work hard to provide the best Products we can and to specify clear guidelines for everyone who uses them. Our Products, however, are provided ‘as is,’ and we make no guarantees that they always will be safe, secure, or error-free.”[114] This reflects the industrial norm in social media: guaranteeing efforts rather than results.[115]

Importantly, however, the best-efforts commitment does not render platforms’ implementation of certain moderation policies voluntary. Although moderation policies can be altered over time, the best-efforts commitment is immutably binding and continues to apply to content moderation, regardless of any substantial change the latter may undergo. In this context, it is important to distinguish between remedy and duty: Breaches by users are typically dealt with by platforms using self-aid remedies,[116] where the obligatory nature of moderation policies legitimizes the sanction.[117] This means that even where platforms enjoy contractual or statutory immunity from liability, which is the currently prevailing stance,[118] the binding force of the contractual norm expressed in moderation policies is not negated. Rather, immunity only proscribes extrinsic recourse in certain cases of default on the part of platforms, often leaving users with only the self-aid remedy of one-sidedly ending the contract. Immunity, therefore, does not extinguish the contractual norm itself.

The implication of the understanding that the social media contractual relationship is structured around a best-efforts commitment, as I explain below, is that platforms are in breach whenever they fail to uphold this commitment. This insight should inform any consideration of the soundness of extending platforms limitless immunity and, importantly, of the standard of conduct to which social media platforms should be held.

3. The Consideration

“The grand bargain of twenty-first-century media,” explains Jack Balkin, “looks like this: Privately-owned infrastructure companies”—for example, social media platforms—“provide you with many different valuable services,” and users “get all of these services . . . for free.”[119] Or rather, for free with one small caveat: “[I]n return [for these services], media owners get to collect [users’] data, analyze it, and use it to predict, control, and nudge what end-users do.”[120] Put differently, platforms’ services are by no means supplied without consideration.[121] Users pay with their personal data, which platforms monetize in various ways. Moreover, as explained below, users pay not only with their data but also with their time.

A variety of provisions, commonly included in platforms’ ToS and data policy agreements, govern the contractual consideration. Meta’s Terms of Service are quite explicit on this issue—stating, for example, that rather than monetary compensation, it will provide its services in exchange for users’ data and consent to be exposed to advertisements, thereby monetizing its users’ attention.[122] This is the customary business model in the social media industry,[123] which yields platforms enormous revenues.[124] In the current media ecosystem, the shortage of human attention has turned it into a valuable asset.[125] Thus, platforms are constantly engaged in attempts to effectively harness users’ attention and make robust use of their personal data to that end. Exploiting human attention and data, particularly in combination, has, accordingly, become an essential component in platforms’ business model.

Social media platforms’ compensation scheme is particularly unique since the personal data they collect is put to “a variety of uses and purposes including . . . the targeting of advertising, content moderation, and the improvement of platform functionality.”[126] Data, therefore, is used by platforms for two intertwined purposes: to improve users’ satisfaction with the service (i.e., enhancing moderation practices to increase engagement and attention spent) and to facilitate the platform’s monetizable business—targeted advertising.[127] Consequently, in social media contracting, there is a symbiotic and cyclical relationship between the contractual consideration and the provided services: better moderation generates more attention and data, which, in turn, further improves moderation, and so on. Under this business model, personal data serves as a raw material:[128] platforms purchase data from users in return for access to their service and then process the data into what they sell for monetary compensation—advertising—both by allowing user targeting and by increasing user attention. Data is thereby an exchanged commodity, and platforms’ ability to efficiently monetize attention is integrally contingent[129] on its continuous collection.

The notion that data and attention serve as remuneration for platforms’ services has already been affirmed by a substantial body of case law. In an action brought against Mariott for a data breach it experienced in 2018, for instance, Mariott argued that personal information lacks clear value, and thus the plaintiffs failed to allege cognizable harm. The court rejected Mariott’s argument, mentioning the “growing trend across courts that have considered this issue [] to recognize the lost property value of this information.”[130] The court further reasoned:

Neither should the Court ignore what common sense compels it to acknowledge—the value that personal identifying information has in our increasingly digital economy. Many companies . . . collect personal information. Consumers too recognize the value of their personal information and offer it in exchange for goods and services.[131]

Other courts have reached similar conclusions.[132] In another notable case, Klein v. Facebook, Inc.,[133] an action was brought against Facebook, citing antitrust violations. The plaintiffs argued that Facebook’s monopolization of the social media market “has harmed users because, without competition, Facebook can extract additional ‘personal information and attention’ from users.”[134] Like Marriot, Facebook disputed the tangible worth of data and attention, and its argument was similarly rejected. “[O]utside the antitrust context,” the court reasoned, “numerous courts have recognized that plaintiffs who lose personal information have suffered an economic injury.”[135] And as the court further stressed after reviewing the presented evidence, “there is no doubt that Consumers’ ‘information and attention’ has ‘material value’” and that the plaintiffs’ allegations in this respect have “ample support.”[136]

The unique qualities and structure of a contractual consideration that is premised on personal information and attention significantly distinguish it from typical consumer transactions. Rather than unilateral provisioning of a product or service in return for an impersonal fungible asset (i.e., money), social media contracting is structured as an ongoing barter: users exchange personal non-fungible assets—their personal data and attention—for a service that is uniquely tailored to facilitate a personalized experience for each individual user.[137] Understanding this exchange as a personalized ongoing barter has two critical ramifications. First, it dispels any lingering notion that platforms provide their services for free. Second, due to the unique qualities of the exchanged commodities in the social media transaction, the governing contractual norms diverge significantly from the norms of typical consumer-seller relationships: they inherently generate long-term reliance, they must mitigate increased risk of abuse, and they foster mechanized personalized treatment.

The unique concerns that arise in relation to the use of personal data as consideration have been addressed by both scholars and policymakers, who have articulated the legal constraints this transactional structure warrants. In Europe, the EU General Data Protection Regulation (GDPR)[138] regulates the market in data and imposes various limitations on data transfers.[139] While the situation is more complex in the United States due to the absence of uniform federal legislation to regulate the data market, the 2018 California Consumer Privacy Act is an example of state-level regulation that addresses these concerns: designed along the lines of the GDPR model, it was enacted to constrain the collection of personal information with respect to residents of California.[140] This notwithstanding, however, in the US, “privacy self-management” (that is, contract) generally remains “the primary check on companies” in their data collection and usage practices.[141]

Existing regulations in the EU and the US as well as contractual terms in platforms’ data policies ascribe transferred data with certain rights that “run with the asset.”[142] For example, a right often attached to data (by regulation or by contract) is non-transferability, which limits platforms’ ability to sell the data they collect to third parties or to share it with others.[143] This constraint does not mean that the data does not serve as contractual compensation. Rather, it merely signifies that rights pertaining to personal information persist with the transfer of the data and bind platforms and users to one another in the long-term. Multiple suggestions have been made from a private law perspective as to the “correct” way to treat privacy rights and data transfers in such transactions.[144] Notably, Jack Balkin and Jonathan Zittrain proposed defining platforms as “information fiduciaries” and thereby subjecting their data-handling to fiduciary duties.[145] The private law grounding of this proposed approach has been both criticized[146] and defended,[147] and currently, attempts are being made to implement this model by way of legislative reform.[148] Indeed, the information fiduciaries theory warrants a comprehensive analysis that is beyond the scope of this Article. But for the purposes of the argument here, it is important to note that this approach is grounded principally on users’ expectations with respect to platforms’ treatment of their data.[149] As Balkin explains, platforms “hold themselves out as trustworthy organizations who act consistent with our interests, even though they also hope to turn a profit,” and “present themselves to the public as responsible and upstanding organizations.”[150] In other words, Balkin stresses that platforms provide assurances that create legitimate expectations.[151] “By presenting themselves as trustworthy” and refusing transparency, platforms “induce relations of trust from us, so that we will continue to use their services.”[152]

The cogency of each of the proposed approaches for classifying and arranging privacy rights and data transfers in the context of social media contractual relations can be equally debated. Notably, however, three features emerge as common to all of the models. First, all recognize, at least implicitly, that data is transferred to benefit platforms, in lieu of monetary compensation, as part of a barter transaction. Second, transferring personal data to platforms is taken to make users vulnerable and to bind the parties together so long as the platforms continue to hold the data. Third, as stated explicitly by Balkin and implicitly in other proposals, consent to the “grand bargain” reflects an undiminishing expectation that personal data will not be weaponized to disproportionately harm its original owner.

III. Revisiting Platform Immunity from Contractual Claims

Before delving into the substantive contractual norms and obligations governing platform-user relationships and examining their necessary legal implications, this Part addresses a crucial preliminary issue: Should a contractual cause of action against social media platforms be barred by statutory immunity or, alternatively, defeated by platforms’ contractual exculpatory clauses?

As described above, platforms include limits on liability clauses in their ToS agreements, which are aimed at extending them broad immunity, including against contractual claims. These clauses are supplemented by Section 230 of the 1996 Communications Decency Act, which, under its currently accepted interpretation, provides statutory immunity from any claim pertaining to platforms’ content moderation decision-making. As this Part argues, however, these contractual and legislated immunities have been too broadly construed by courts, and platforms should not enjoy wholesale immunity from contractual claims. While their respective interpretations should certainly be consistent with one another and address issues of public policy, the interpretation of statutory and contractual immunities should also aim for a balanced result that reflects the expectations of both users and legislators with respect to platforms’ decision-making.

Reading into Section 230 robust immunity from contractual causes of action is untenable, I claim, as is unnuanced enforcement of exculpatory clauses. As this Part shows, Section 230 was designed specifically to bar causes of action in torts, not contracts; it was certainly not legislated to allow platforms to make fraudulent misrepresentations about the services they provide. Furthermore, judicial interpretation of platforms’ exculpatory clauses has been misguided, inter alia, in its adherence to the illusion that this is a contractual relationship in which platforms’ services are provided for free.[153] This Part, therefore, analyzes the interaction between Section 230, exculpatory clauses, and contractual obligations, to show that there is no conflict between the contractual liability standard and the goals of Section 230: both require that contractual causes of action be heard, on the one hand, while both seek restraint in judicial encroachment on platforms’ content-related decision-making, on the other. Such a balance, I assert, is in both cases achieved by immunizing good-faith decision-making and avoiding strict liability, and it can be further secured through proper judicial construction of remedy and standard of review.[154]

Not only is this discussion crucial for understanding platform immunity, it is also most timely given the apparently growing willingness of the Supreme Court to consider, for the first time, the proper scope of Section 230’s statutory immunity. A powerful indication of this willingness was recently given by the Supreme Court’s decision to grant certiorari in Gonzalez v. Google,[155] and Twitter v. Taamneh.[156] These cases originated in claims brought against Twitter, Facebook, and YouTube for allegedly allowing ISIS to make extensive use of their platforms to propagate terrorist agenda and solicit support. Based on the Anti-Terrorism Act (ATA),[157] the plaintiffs sought to hold the companies accountable for the third-party content made available on their platforms, and recommended by their algorithms—an activity arguably protected under Section 230’s immunities for third-party content.[158] In Gonzalez the plaintiffs explicitly sought to challenge the prevailing judicial interpretation of Section 230 in lower courts. Furthermore, the decision to hear these cases followed Justice Thomas’ previously expressed unease with Section 230’s emerging jurisprudence.[159] And thus, for a moment, it appeared that Gonzalez and Taamneh would provide the court an opportunity to scrutinize the precise extent of platforms’ statutory immunity. However, the specific causes of action alleged by the plaintiffs—”aiding and abetting” terrorist activity under the ATA—significantly narrowed the court’s opportunity to delve into Section 230, ultimately causing the court to sidestep the issue altogether.[160] The Supreme Court’s apparent readiness to reexamine Section 230 doctrine thus remains, suggesting that the current status quo concerning platform immunity is about to be dramatically challenged. Turning to contract, as this Part seeks to show, may provide the necessary tools for addressing the challenge of platform immunity.

A. Section 230

This Section will argue that Section 230 does not categorically preempt contractual claims against platforms and that the current application of its immunity with respect to contractual causes of action is flawed. This conclusion is derived from the history of the Section’s enactment, its text, and its purpose.[161] I assert that acknowledging a contractual cause of action that duly addresses the structure of social media contractual relationships and upholds the parties’ expectations, rather than imposing extrinsic norms on the relationship through tort liability, would be consistent with Section 230 and would actually advance its goals.

Section 230 accords two substantial privileges to online platforms: Section 230(c)(1) provides that no service provider “shall be treated as the publisher or speaker of any information” posted by a third party, rendering platforms immune to claims with regard to content that they decide to carry; Section 230(c)(2) provides that no platform shall be held liable for “any action voluntarily taken in good faith” to remove “objectionable” content.[162] These privileges have been interpreted very broadly by the courts,[163] granting platforms “sweeping” protection from any action or inaction related to content-moderation decision-making.[164] Moreover, some courts have noted that the tort/contract distinction should be rejected in this context and have, instead, adopted a substance-based test: where the platforms’ contested action was “editorial” in essence, it will be covered by Section 230 immunities, regardless of the classification of the cause of action.[165] However, Section 230’s legislative history casts doubt on the interpretation that the provision robustly bars contractual claims in this way.

As its legislative history reveals, the underlying purpose of Section 230 is to encourage online platforms to moderate content by providing them safe harbor, albeit under certain conditions.[166] The Section was intended to correct the outcomes of two defamation claims brought against online content intermediaries.[167] In Cubby v. CompuServe,[168] an online intermediary did not review any of the content uploaded to its platform and was therefore found not liable for defamatory statements that the platform carried. Conversely, in Stratton Oakmont v. Prodigy Services,[169] a content intermediary was found liable for defamation since it had taken some measures to remove objectionable content from the platform. Underlying these opposite outcomes was the established distinction in tort law between a “distributor” and a “publisher” of content in relation to liability for carrying another’s defamatory statements. Publishers are subject to a strict liability standard and are held to have knowledge about statements they publish. In contrast, distributors are held liable only upon proof of having actual knowledge about carrying defamatory content.[170]

From a public policy perspective, the Cubby and Stratton Oakmont outcomes met with unease. Legislators wished to strike a different balance: on the one hand, they sought to allow online intermediaries to more freely remove content they perceive as problematic; on the other hand, they wanted to disincentivize over-removal so as to protect users’ speech interests. Thus, rather than leaving the law to correct itself through the iterative process of common law, Section 230 was enacted to overturn Stratton[171] and release platforms from the speech-burdening doctrines of tort law.

This legislative purpose was also explicitly declared by the Fourth Circuit in Zeran v. America Online,[172] a seminal opinion in Section 230 jurisprudence:

The purpose of this statutory immunity is not difficult to discern. Congress recognized the threat that tort-based lawsuits pose to freedom of speech in the new and burgeoning Internet medium. The imposition of tort liability on service providers for the communications of others represented, for Congress, simply another form of intrusive government regulation of speech.[173]

Thus, not only is the purpose of limiting platforms’ tort liability easily discernable from Section 230’s legislative history and text, but the normative grounds for this limitation are also quite unequivocal. However, the concerns that underlie the limitation set on tort liability are not equally applicable to contractual liability, given the substantial differences between the two bodies of law. Two particular distinctions are worth noting in this respect.

First, while tort law generally engages in imposing extrinsic norms, contract law leaves room for private parties to set relational norms as they see fit. From this perspective, an uninhibited speech environment requires allowing private parties to reach agreements about which rules of discussion govern their privately held forum. In other words, freedom of speech includes the freedom to bind oneself to certain speech-related norms.[174] The second distinction relates to the important question of incentives. As the Fourth Circuit stated in Zeran, “[t]he specter of tort liability in an area of such prolific speech would have an obvious chilling effect,”[175] which works in one direction: toward content removal.[176] Allowing access to adjudication only to those who are harmed by content, and not to speakers, both creates a “consistent[] push toward removal” as well as “distorts courts’ opportunities to clarify the law,” since it generates an untenably imbalanced situation where “claims of people defending their expression rights” are not heard.[177] Allowing a contractual cause of action for harm caused by platforms’ decisions to carry or remove content would not create a similar problem, because it would operate in both directions: decisions to remove content could constitute a breach of platforms’ contractual obligations (toward speakers) no less than a decision to carry content could constitute a breach (toward listeners). This balance would mitigate the one-sided chilling effect generated by tort liability and would create a different incentive for platforms: to moderate as promised, in accordance with users’ expectations.

In sum, reading into Section 230 an intent to entirely invalidate contractual causes of action, thereby allowing platforms to arbitrarily breach reassurances and even directly mislead users regarding the very service they provide—seems like a far stretch.[178] Section 230 does not bar a contractual cause of action. Rather, for a contractual cause of action to be “consistent with” Section 230 immunities,[179] and absent an explicit commitment made by a platform to provide error-free moderation (which is unlikely to be made), courts must: (1) avoid applying a strict contractual liability standard when adjudicating content-moderation decision-making,[180] since such liability would both impose an extrinsic norm and make platforms de-facto “publishers,” and (2) courts should refrain from replacing platforms’ good faith judgment, which is protected under Section 230(c)(2), with their own.[181]

Adopting the contractual approach proposed here will not raise concerns about inconsistencies with Section 230 immunities.[182] No platform undertakes to provide services it can never provide (i.e., error-free content moderation), and if a platform decides to promise such services, it should be allowed to bear the costs and enjoy the benefits of its choice. The alternative legal rule—whereby platforms may freely reap the fruits of false undertakings, without paying the price for not upholding them—is a clear form of relational injustice or unfairness, which should be avoided. As discussed above, a reasonable platform will inevitably commit to best efforts in providing content moderation,[183] meaning that strict liability is ab initio not the contractual norm and good-faith attempts to uphold commitments are the very obligation the contractual relationship manifests.[184] A correct interpretation of the social media contractual arrangement, therefore, reveals that it is structured to facilitate, rather than circumvent, the balance Section 230 was in fact intended to achieve—a conclusion also supported in Klonick’s findings.[185] Moreover, given the advantages of contract law compared to tort law in this context, there is no good reason to prevent the enforcement of contractual obligations. This would not only be entirely consistent with Section 230, but would also advance its goals.

Interestingly, in Taamneh and Gonzalez, the Supreme Court eventually found that the plaintiffs had failed to state a claim under the ATA.[186] Explicitly for this reason, and despite its apparent initial willingness, the Supreme Court declined “to address the application of §230 to a complaint that appears to state little, if any, plausible claim for relief.”[187] Justice Jackson, in her very short concurrence in Taamneh, seems to have given further clues regarding her future willingness to put Section 230’s prevailing doctrine under scrutiny, potentially allowing alternative causes of action to be heard.[188] Eventually, it was the plaintiffs’ failure to explain why the platforms’ “passive nonfeasance,” which was alleged in the case—as opposed to the “affirmative misconduct” necessary to establish aiding and abetting under the ATA—should give rise to liability. As stated by the Court:

Because plaintiffs’ complaint rests so heavily on defendants’ failure to act, their claims might have more purchase if they could identify some independent duty in tort that would have required defendants to remove ISIS’ content . . . But plaintiffs identify no duty that would require defendants or other communication-providing services to terminate customers after discovering that the customers were using the service for illicit ends . . . To be sure, there may be situations where some such duty exists, and we need not resolve the issue today.[189]

A contractual approach to social media governance, as suggested in this Article, that acknowledges social media platforms’ affirmative contractual obligation to act in reasonable diligence and utmost good faith to remove terrorist content could overcome this obstacle.

B. Exculpatory Clauses

Another potential obstacle to bringing a moderation-related contractual claim against platforms is the exculpatory clauses commonly included in ToS agreements. I will show here, however, that such clauses should be invalidated: both doctrinal and public policy considerations dictate that courts disregard clauses that completely exempt platforms from providing the very service they committed to provide. Platforms must moderate content in good faith and in accordance with the representations they make in this regard, and courts should not enforce contractual terms that imply otherwise.

As noted earlier, all major platforms currently include limits on liability clauses in their ToS agreements.[190] Even though they are rarely read by users,[191] these online boilerplate agreements remain generally binding as long as the consumers had the opportunity to review them before manifesting consent.[192] Yet this notwithstanding, exculpatory clauses included in such agreements are subject to several special limitations. To begin with and as a general rule, exculpatory clauses are disfavored and “strictly construed against the benefited party.”[193] They are also subject to a duty of good faith,[194] to invalidation in the event of unconscionability,[195] and to public policy considerations. These doctrines can limit the scope of exculpatory clauses or even wholly invalidate them. The proposed understanding of the contractual relationship between platforms and users clarifies why each of these doctrines supports limiting these clauses. It also exposes the weaknesses in the case-law on platform liability-exemption clauses that are currently emerging in the California state courts, which are of particular relevance to any contractual claim against dominant social media platforms.[196]

Firstly, platforms’ limits on liability clauses should not be enforced because they are unconscionable. Courts can invalidate contractual terms based on “procedural unconscionability,” which “involves ‘oppression’ or ‘surprise’ due to unequal bargaining power,” and “substantive unconscionability,” which focuses on “‘overly harsh’ or ‘one-sided’ results.”[197] Though some evidence of both procedural and substantive unconscionability is necessary, the two types “need not be present in the same degree.”[198] Rather, courts use a sliding scale where “the more substantively oppressive the contract term, the less evidence of procedural unconscionability is required to come to the conclusion that the term is unenforceable, and vice versa.”[199]

From a procedural perspective, the California courts have generally accepted that standard-form adhesive contracts “establish at least some degree of procedural unconscionability,” which requires scrutiny of “the substantive terms of the contract.”[200] Compounding this presumption is an assumption that casts further doubt as to users’ ostensibly free choice in accepting platforms’ robust disclaimers and is based on the robust analysis of social media contracting provided above. As explained, content moderation, in many respects, is the commodity contracted for in the social media transaction.[201] More specifically, content moderation defines the essence of the platform generally, and shapes the community in which one agrees to participate. It subsumes all other aspects of the service: Moderation decision-making can result in complete denial of access to the platform through user banning; can completely or substantively defeat any sharing of content, through shadowbanning or automated down-ranking; can completely deny a user the content she legitimately expected to receive through unaccounted for filtering; and can entirely thwart any option of engagement in expected online relationships or debates by similar means. It necessarily follows that completely exempting platforms from any liability for failing to moderate content as expected means that platforms are essentially free not to provide the very service for which users contracted.[202] Such a result should be considered unconscionable. As the Restatement (Second) of Contracts provides, where an exculpatory clause in a boilerplate contract materially “eliminates the dominant purpose of the transaction,” there is no “reason to believe” that anyone would freely accept it, and accordingly, the clause should be disregarded.[203] Indeed, platforms should not have an indefeasible right to mislead.

This result is further supported by the complete one-sidedness and oppressive result of the clause,[204] allowing platforms to collect remuneration without bearing any risk, and even without providing the promised service. Yet California courts have thus far refrained from invalidating platforms’ limits on liability clauses on these grounds. Their jurisprudence reveals that the decisions to enforce such clauses are fundamentally based on substantive misperceptions about the nature of the social media contractual relationship and on an unsubstantiated doctrinal shift. In Lewis v. YouTube,[205] a user sued YouTube for breach of contract after YouTube suspended her channel and refused to reinstate it in its pre-takedown form. The California Court of Appeals refused to invalidate YouTube’s Limits on Liability clause, given the “general approval” of such provisions in California and that, “[a]s in the present case, these clauses are appropriate when one party is offering a service for free to the public.”[206] A California district court subsequently arrived at a similar decision in Darnaa v. Google,[207] addressing the exculpatory clause’s validity from two complementary perspectives: unconscionability and public policy. First, the court swiftly rejected the claim that Google’s Limits on Liability clause was substantively unconscionable, based on a similar rationale as articulated in Lewis, namely, that providing a service for free renders such clauses reasonable.[208] The court thereafter also rejected a public-policy-based argument for invalidating the clause and, consequently, the applicability of section 1668 of the California Civil Code, which permits invalidating exculpatory clauses that are counter to public policy.[209] This decision was grounded on the determination that section 1668 does not apply in breach of contract claims absent a showing of fraud or misrepresentation.[210]

Both Lewis and Darnaa seem to rest on shaky analytical ground. Denying the unconscionability claim based on the assumption that the platform’s services are provided for free is both factually erroneous and doctrinally uncompelling. From a factual perspective, the Darnaa and Lewis courts failed to acknowledge that social media contracting creates a barter, in which a non-monetary commodity is exchanged for a service, as elaborated above.[211] The uncanny implication of this approach is that platforms’ exculpatory clauses reallocate the risk of the interaction and place it entirely on users’ shoulders; this presumably allows platforms to refrain from fulfilling any of the contractual obligations they have undertaken, whether intentionally, recklessly, or negligently. This risk allocation seems grossly unfair and objectively unreasonable.[212] Perhaps were the services actually provided free-of-charge, such risk allocation could be justified, but as explained, users pay for this service with a powerful currency that generates enormous revenues for platforms.

From a doctrinal perspective, the foundations of the test applied by the Darnaa and Lewis courts to determine the validity of the exculpatory clause are somewhat tenuous. In Lewis, the court relied on case-law that it interpreted as endorsing exculpatory clauses where “the beneficiary of the clause is involved in a ‘high-risk, low-compensation service.’”[213] However, in using this test, the Lewis court erroneously applied the precedent: the original ruling[214] only noted this standard as a tool for examining the validity of liquidated damages provisions[215] and applied a different test to examine separately the conscionability of exculpatory clauses.[216] This erroneous application of the precedent should be revisited in future cases. But regardless, the determination that social media platforms—which are amongst the wealthiest companies currently in operation and enjoy a legislated safe-harbor—provide their services based on a “high-risk, low-compensation” model is significantly misguided, particularly in light of the aggregatory outcome of platforms’ business.

A second and even more easily applicable rationale for invalidating platforms’ limits on liability clauses is rooted in the duty to perform contracts in good faith. This duty is an implied warranty in every contract and is non-waivable.[217] Its enforcement is particularly crucial in consumer contracts that confer substantial discretion to the seller in determining the parties’ contractual rights and obligations. As suggested by section 5(b) of the Draft Restatement of Consumer Contracts, any term in such contracts that seeks to allow decision-making “unconstrained by the good-faith obligation or by other implied limitation” should be unenforceable.[218] This has direct implications for social media contracting, given that platforms are subject to an inherently limited standard of liability based on a best-efforts commitment, allowing them broad decision-making discretion.[219] Good faith in performing this commitment is a definitive element of the best-efforts commitment.[220] That is to say, where a contractual obligation is not subject to a standard of strict liability that would assure a specific result, all that remains is reliance on good faith performance. Remove that requirement, and no obligation is left.

Lastly, public policy considerations also support invalidating platforms’ limits on liability clauses. Both Lewis and Darnaa substantially lacked deliberation of the public policy implications for the analysis. As noted, the Darnaa court examined this issue through the lens of section 1668 of the California Civil Code, but in doing so it ignored the leading precedent on the matter: the California Supreme Court ruling in Tunkl v. Regents of the University of California.[221] In Tunkl, the plaintiff, upon admittance to the defendant’s hospital, signed a form releasing the latter from “any and all liability for the negligent or wrongful acts or omissions of its employees.”[222] However, although section 1668 explicitly invalidates exculpatory clauses in cases of “fraud,” “willful injury,” or “violation of law”[223] (a formalistic argument on which the Darnaa court relied to deny its applicability to the contractual claim brought against Google), the Tunkl court extended the section’s scope of protection to invalidate liability limitations in instances of negligently inflicted harm. The court moreover ruled that an “exculpatory provision may stand only if it does not involve ‘the public interest,’”[224] making public policy the focus of the scrutiny and not whether the cause of action is torts-based or contracts-based.

Given the patent bearing of social media platforms, particularly the major ones, on the public interest, it is unnecessary to comprehensively address each of the multiple factors examined in the Tunkl test.[225] In the social media contracting context, of particular pertinence is the test’s first prong, which examines whether the service (content moderation, in our case) is “suitable for public regulation.” On the one hand, platforms’ significant influence on public discourse and economic activity,[226] combined with their tendency towards monopolistic power, largely driven by network effects, necessitates some level of accountability.[227] On the other hand, First Amendment concerns push in the other direction, warranting caution with respect to governmental encroachment on platforms’ decision-making. In many respects, Section 230 itself inherently validates Tunkl’s first requirement, evincing that the field was subjected to public regulation practically since its inception. Ultimately, however, this question boils down to a policy analysis that is similar to what is discussed above in the context of Section 230 and the appropriate scope of its legislated immunities. As stated, while there are ample reasons for limiting social media platforms’ liability in torts—as Congress instructed—they should not be permitted to mislead their users about the content and quality of the services they provide or perform their undertakings arbitrarily or in bad faith. Instead, public policy requires that platforms would be held to their best-efforts commitment to uphold their obligations with respect to content moderation, and any attempt to exempt themselves even from this lenient standard of liability should be dismissed.

IV. Social Media Contracting as Relational Contracting

The significant divergence of social media contracting from traditional consumer contracts demonstrated above,[228] mandates a parallel divergence in applicable contract law theory and doctrine. And this latter divergence is crucial if we ever want courts to be able to properly adjudicate contractual disputes between users and platforms. That is to say, traditional contract law as customarily applied in the consumer-seller context is simply unsuitable for overseeing user-platform relationships. It may perhaps have been a presumed rigidity in the judicial application of contract law doctrine that led scholars and policymakers to prefer a public law strategy for shaping governance norms.[229] However, this rigidity is not inevitable, and contract law can and should be harnessed to align social media governance with the contractual norms and obligations freely adopted by the parties.

In this Part, I therefore provide the necessary theoretical framework for supporting this project, drawing on relational contract theory and applying it to the social media context. As I show below, social media contracting is better understood by analyzing it as a form of relational contracting. Indeed, the user-platform contractual relationship manifests central characteristics that are relational at their core. It is based on an inherently incomplete and dynamic arrangement with an array of rights and obligations that are constantly adapted and adjusted to changing circumstances. Social media contracting is aimed at facilitating long-term cooperation between users and platforms, premised on interdependency, mutuality, and reciprocal respect for the parties’ competing interests. It adopts an ongoing fair cooperation model, based on bilateral transactions, and it induces an expectation for the protection of human rights and for procedural fairness.

Given the inherent relational nature of these features, relational contract theory offers potent tools for identifying the contractual norms that govern the social media relationship and for crafting appropriate remedies to provide redress for harm. It justifies the inclusion of implicit terms in the contractual relationship, supports the necessary contractual adaptability, and calls for contextuality in adjudication, all of which are crucial for attending to the shortcomings in how social media harms are currently addressed. Indeed, notwithstanding its direct or indirect applicability to legal doctrine,[230] relational contract theory equips us with the necessary analytical tools to interpret parties’ words and conduct in ongoing, incomplete, cooperative relationships, and allows us to discern reasonable expectations deserving of legal protection. By shedding light on platforms’ behavior and users’ expectations from an intra-contractual perspective, the relational analysis presented below thus seeks to establish the normative groundwork for the subsequent implementation of legal doctrine outlined in Part V.

A. Relational Contract Theory

Relational contract theory was introduced by Stewart Macaulay[231] and Ian Macneil[232] during the second half of the twentieth century.[233] It is grounded on the basic premise that contracts—all contracts—must be read in the social context in which they were formed and intended to operate.[234] Norms stemming from the social relationship that produced the contract control the parties’ expectations, even where the parties were unable or unwilling to articulate these norms in explicit terms.[235] Articulating “definitive obligations,” as Charles Goetz and Robert Scott stressed, is often “impractical” either “because of inability to identify uncertain future conditions or because of inability to characterize complex adaptations adequately even when the contingencies themselves can be identified in advance.”[236] Relational contract theory therefore holds that express terms of relational contracts should serve their interpreters as “no more than an entry point to consideration of all the pertinent issues.”[237]

The relational aspects of contracts, Macneil explains, are not one-size-fits-all. Instead, contracts are situated along a continuum. At the one end of this continuum is the “discrete contract,”[238] which is the paradigmatic contractual relationship in classic contract law and is designed to govern an arm’s-length, isolated exchange of goods. It is unconcerned with the former or future relationship between the parties, and its contractual norms can typically be reduced into explicit terms. However, most contracts deviate from this paradigm, moving along a continuum that ends with the fully relational contract, which arranges a more complex or extensive long-term relationship and exists in a robust social context.[239] When this conceptualization was considered against emerging concepts of neoclassical contract law and fitted into its categorizations, it was recast into the now-prevailing view that “there exists a class of relational contracts that deserves treatment as a special subcategory of the general contract law.”[240]

The expectations of parties to relational contracts regarding contractual norms tend to differ from those of parties to discrete contracts. Unlike the discrete contractual arrangement, relational contracts emphasize “the common norms of a co-operative character,” aiming to ensure the “preservation of the relation” and premised on “contractual solidarity.”[241] Such solidarity, according to Macneil, facilitates the necessary “belief in being able to depend on another.”[242] Put differently, relational contracts are premised on “trust,”[243] and relational contractual norms are designed to reflect and facilitate that trust among the parties a priori.

Where contracts are sufficiently relational (i.e., are to a large degree formed to arrange long-term relationships requiring complex or extensive arrangements), the theory stipulates, they will inherently include a set of implied relational norms that govern the relationship.[244] These norms reflect legitimate competition between the parties’ competing interests, which is “bounded by integral acceptance of co-operation as operative within the contract.”[245]As Robert Gordon observed in this respect, “the object of [relational] contracting is not primarily to allocate risks” but, rather, to “signify a commitment to cooperate.”[246] Such a commitment requires that the parties respect one another’s mutual expectations and, to that end, go beyond the explicit terms of the contract. “In bad times,” Gordon stressed, “parties are expected to lend one another mutual support, rather than standing on their rights,” and “if unexpected contingencies occur resulting in severe losses, the parties are to search for equitable ways of dividing the losses.”[247] Thus, as part of the effort to facilitate mutual trust and cooperation, the parties to a relational contract are expected, as well as contractually obligated, to share the burdens and benefits of the transaction and refrain from opportunistic behavior.[248] This expectation and obligation are premised on notions of reciprocity and, importantly, fairness, which are essential for securing the long-term endurance of the relationship.[249] Opportunistic behavior and unfair usurpation of benefits are antithetical to trust and, therefore, prohibited. As Macneil describes, “wherever one side is seen as hogging the exchange-surplus or otherwise misusing power . . . the solidary belief is in danger since people may begin to question the benefits of continued interdependence and hence its very existence.”[250] Such unfair behavior, Macneil argues, may generate “disproportionate harm” to the contractual partner, in direct contradiction to the parties’ obligations.[251]

Interdependence is another significant attribute of relational contracts,[252] inherent to a relationship in which surplus can be produced only through cooperation and “division of labor.”[253] Interdependence does not mean, however, that power asymmetries do not exist in the relational contractual relationship. On the contrary, they are often inevitable[254] and can result from a preexisting imbalance or develop as circumstances change during the course of the relationship. Either way, abusing a power advantage with opportunistic behavior is strictly prohibited in relational contracts, and sharing the benefits fairly is required of the parties.[255]

Hence, inherent to relational contracts is a series of implied norms that govern the relationship, which are generally premised on ensuring fair cooperation. Clearly, the existence of such implied norms does not categorically preempt the existence of explicit norms conveyed in writing, nor their binding power. On the contrary, as Hanoch Dagan and Michael Heller stress, a typical feature of relational contracts is “the significance of what may be called ‘contract governance’” that puts “more emphasis . . . on the ‘operating relations’ of the parties and on ‘structures and processes.’”[256] Moreover, where parties “become highly interdependent . . . their ‘relations tend to include both sharp divisions of benefits and burdens and a sharing of them.’”[257] These contractual schemes, typically detailed in writing, assist to ensure the adaptability of the contract and, thereby, guarantee the perseverance of the relationship. In addition, such structures and processes often also govern disputes that may arise between the parties, allowing contingencies to be addressed so as to sustain the relationship, rather than by legalistic adherence to the terms of an incomplete express contract.[258] As Gordon succinctly explains, “each [party] will treat the other’s insistence on literal performance as willful obstructionism.”[259]

To sum up, relational contracts are characterized by high levels of solidarity and trust between inherently interdependent parties and are designed to facilitate cooperation. They are innately incomplete and entail ongoing adjustment[260] of obligations, require fair balancing of competing interests, and restrict opportunistic grabbing of the exchange-surplus. As I proceed to show, this is the inherent structure of social media contracting.

B. The Relational Nature of Social Media Contracting

The relationship between users and platforms in the framework of their ongoing barter transaction is sufficiently characteristic of a relational contract to be included in this category. It includes: indispensable reliance on trust with regard to both data utilization and content-moderation services; a commitment to fairness and diligence; inherent incompleteness, with ongoing proportional adaptations to address contingencies that may arise; and, importantly, interdependence and cooperation between the parties in generating joint surplus.

David Hoffman has presented qualitative empirical findings supporting this classification.[261] Based on interviews with social media personnel, he found that platforms’ subjective intention in designing the contractual scheme was to facilitate a long-term relationship of trust, where performance is “governed largely by users internalizing a set of rules created through brand alignment, informality, and interpersonal norms of reciprocity and fairness,”[262] similar to an attempt to create “a classic relational contract.”[263] Yet Hoffman raises some reservations about the validity of this classification of social media contracting, suggesting, instead, to frame it as “a new form of contracting,” which he terms “relational contracts of adhesion.”[264] This Section will challenge Hoffman’s proposed alternative conceptualization, arguing that it is unnecessary.

From a theoretical perspective, relational contract theory has never found it difficult to classify contracts of adhesion as relational. In fact, standard-form contracts were the first paradigm upon which Macaulay’s theory was originally tested;[265] for Macneil, relational contract theory provided the only available theoretical justification for enforcing boilerplate contracts in consumer-seller settings, despite the consumer’s clear lack of consent.[266] In other words, the “relationality” of a contract does not turn on its form but rather on its substance. This insight illuminates why Hoffman’s finding of platforms’ subjective intention to facilitate relational contracts with their users was to be expected. As this Section shows, social media contracting is inevitably relational, irrespective of the adhesive nature of ToS agreements. Social media’s inherent characteristics are what make its contracting relational, and the contractual scheme designed by platforms merely reflects this.

1. Incompleteness and Adaptability in an Ongoing Relationship

As described, a core characteristic of relational contracting is the inability to articulate definitive obligations in explicit terms, which results in an inevitably incomplete contract that is subject to future adjustment. Such adjustments are intended to support the endurance of the contractual relationship and should always balance between the competing interests. This characteristic is clearly evident in the social media contracting context.

As Klonick describes, platforms aim to create an environment “that reflects the expectations of their users,”[267] and the content-moderation system they have designed “to match the expectations of users and to self-regulate” is “impressively intricate and responsive.”[268] Under this system, internal rules and policies are “iteratively revised on an ongoing basis, and much more frequently than the external public-facing policy”;[269] they are “constantly being updated . . . to rapidly reflect the norms and expectations of [] users.”[270] Indeed, platforms aspire to create a service that attracts long-term user engagement through, amongst other things, constant adaptation to changing circumstances. Basic activities on social media, such as building profiles, developing communities, amassing followers, and establishing reputation, rely heavily on the long-term persistence of the platform-user relationship. In other words, social media contracting is, by its very nature, adaptable and ongoing.

The policies adopted by Facebook during the COVID-19 pandemic illustrate this dynamic process in which users’ expectations are formed and adjusted. In a March 2020 news release, Facebook announced that since January of that year, it had been removing “COVID-19 related misinformation that could contribute to imminent physical harm.”[271] This statement, alongside the adoption and publication of complementary moderation policies,[272] directly impacted users’ expectations with regard to the service provided by the platform: from the moment that Facebook began to remove such content, it was signaling to users that they were now prohibited from posting certain views about a prominent socially and politically controversial issue, which had previously generated substantial traffic. Significantly, users were also informed that they would generally not be exposed to such allegedly harmful opinions.

The adaptable nature of content moderation policies thus enables platforms to address new perceived dangers to users as they emerge. Indeed, little was known about the coronavirus when Facebook installed its COVID-19 policies. As it announced, “we regularly update the claims that we remove based on guidance from the WHO and other health authorities.”[273] This underscores the inherent need for flexibility in moderation policies and, at the same time, their inevitable incompleteness given the inability to foresee future contingencies and to translate terms like “harmful content” into comprehensive obligations. Facebook’s COVID-19 policies, as well as its explicit rationale for adopting them, also convey that moderation policies are designed to protect users in their exposure to content on the platform and that Facebook is committed to safeguarding their interests in its decision-making.

Moderation policies’ innate flexibility and adaptability to shifts in public interests have two crucial and complementary implications for understanding the essence and purpose of the contractual arrangement. First, the dynamic nature of moderation policies reinforces the understanding that the social media contractual arrangement is not intended to facilitate a one-time, short-term interaction (i.e., a discrete transaction) but, rather, to regulate an evolving, long-term relational interaction that requires constant adjustment. Second, in terms of the underlying rationale for policy changes, such adaptations are often intended to secure the safe space necessary to facilitate as much speech as possible in the long term. Indeed, moderation policies and practice communicate a message about the safety and character of speech on the platform and, in so doing, allow more speech and not less.[274] Keeping platforms safe for user participation is, therefore, in the interest of users (at least as a collective) and platforms alike.

This exposes another aspect of social media contracting that supports the relational contract classification: namely, in order to maintain a long-term relationship with users, platforms must at least occasionally demonstrate concern for users’ interests. To illustrate, when Twitter (now X) decided to suspend Donald Trump’s account, it asserted that this was in the interest of protecting its users’ social safety[275] and it did so despite the loss of benefits it reaped from the exceptional traffic his tweets generated. Facebook’s COVID-19 policies can be construed as similarly motivated. As Gillespie insightfully notes, “[p]latform moderators like to think that their guidelines already represent the values of users, and are responsive to shifting norms and practices, a ‘living document’ that moves with their user community.”[276]

Thus, moderation policies are designed to adaptively address users’ concerns and interests. Platforms uphold and adjust these policies, at times in a short-term, supposedly “altruistic,” fashion, and view these adaptations as crucial for facilitating a vibrant community in the long-term. Moreover, whether such adaptations are motivated by purely reputational gain is irrelevant. From a contractual perspective, a platform’s attempt to boost its reputation by providing certain assurances creates user expectations that are injected into the contractual relationship.

2. Interdependence and Cooperation in Creating the Exchange-Surplus

As explained, relational contracts are created to facilitate cooperation between interdependent parties. This interdependence is inherent to relationships with a sharp division of labor,[277] where each party fills a different role in generating the joint surplus. In such relationships, cooperation is vital. And indeed, one of the purposes of social media contracting is to enable such an interdependent and cooperative endeavor. Users are full participants in the production of social media[278] and are just as necessary as platforms for generating its value. The multiple roles users play in de facto producing platforms’ services make social media platforms definitionally dependent on users for having anything of value to sell. In other words, if “content moderation” is the commodity sold by platforms, then users produce its first half. As Elkin-Koren et al. stress in this regard, “while social media platforms generate their profits from advertising, it is users who provide the bricks from which the platforms build their business model.”[279] They further explain:

Users of social media platforms play multiple roles. Besides being consumers of services supplied by the platform . . . users themselves are also providers of content and supply added value to platforms, in ways that affect the myriad interests of other users of the platform and shape their . . . expectations. Since the value of usage to each user, and subsequently also to the platform, is generated by engagement among users, users should be conceived as “partners” with the platform in a contractual network that aims at a collaborative goal.[280]

         Users’ multiplicity of roles in social media reinforces the relationality of the transaction and is evidence of the scope and complexity of cooperation between the parties. Partnership—a typical example of relational contracting—indeed comes close to describing this relationship between platforms and users. This relationship is ultimately a joint effort aimed at creating joint surplus. The parties consensually enter into a relationship based on ongoing bilateral transactions (that is, each party both “sells” to and “buys” from the other) and assume the roles necessary for the success of the joint effort. Users thereby become both producers and consumers of content, as well as providers of data and targets of advertising. This allows platforms to focus on their distinct roles as disseminators and moderators of content, alongside coordinators of the advertising it carries, which ensures the endeavor’s economic viability. Indeed, barter exchanges, which are often present in relational contracts, tend to increase the parties’ interdependence and facilitate credible commitments for greater cooperation in the future.[281]

The duality of users’ participation in social media makes it tricky to analogize social media contracting to a single established category of relational contracts. However, when we combine two relational contract analogies, each pertaining to a different component of user participation, the interdependence embedded in the user-platform relationship becomes evident. Where users act as content producers, the contractual relationship can be easily analogized to a manufacturer-distributor or writer-publisher relationship. At the same time, where users act as consumers of content, paying platforms with their personal information, the interaction can be framed in supplier-fabricator terms: users are suppliers of data and attention, while platforms are fabricators that turn these raw materials into valuable advertisement spaces, which are subsequently monetized through third-party acquisition.[282] These are archetypical types of relational contracts that entail substantial interdependence based on division of labor and profound, ongoing cooperation.

3. Solidarity and Trust

The third fundamental aspect of the social media transaction that grounds its classification as relational contracting is the parties’ indispensable reliance on solidarity and trust, which are pervasive in every element of the contractual arrangement.

As described above, platforms use their moderation policies to generate users’ trust,[283] thereby “enhanc[ing] their brand.”[284] Community guidelines, as Gillespie asserts, “constitute a gesture” to users “that the platform will honor and protect online speech and at the same time shield them from offense and abuse.”[285] Platforms, in turn, must also trust users not to abuse their access to the service and rely heavily on user self-censorship in this regard.[286] Furthermore, as reflected in the contractual best efforts standard of liability, since moderation policies are impossible to reduce into a set of specific rules and since platforms avoid subjecting themselves to stringent transparency demands regarding the design of their algorithm and its impact,[287] users must rely on platforms’ good will in performing the contract. In other words, users must trust that platforms make their best effort to moderate in accordance with users’ expectations and have no substantive ability to monitor platforms’ activity.

The high degree of trust needed to sustain the user-platform relationship is also a constitutive element of the platforms’ business model, which is grounded on the monetization of personal data. As Balkin explains, platforms induce users to trust them so as to facilitate the extraction of their personal data, whether or not that inducement is also made explicit.[288] Thus, users allow platforms to use their data to curate their content and to target their attention and therefore should be able to trust that, in so doing, the platforms will not disproportionately harm them.

4. Fairness and Diligence

Platforms go to great lengths, and for good reason, to convince users that they conduct fair content moderation, both procedurally and substantively.[289] The social media contractual arrangement accords platforms with substantial discretionary power in delineating their own rights and obligations with respect to both content moderation and data monetization. As stated in the Draft Restatement of Consumer Contracts, “[a] contract or any term that grants the business discretion to determine its rights and obligations must be interpreted, when reasonably susceptible to such interpretation, to provide that such discretion will be exercised reasonably and in good faith.”[290] Thus, broad discretion is accompanied by the expectation that it will be exercised fairly and responsibly, requiring reasonable diligence in performance and good faith. As explained below, a best efforts commitment corresponds with this requirement, as it implies fairness and diligence in performing the contractual obligation.[291]

Moreover, given the unique privacy concerns attached to the possession of another’s personal data, its use is typically regulated more carefully, with concrete limits set on its utilization and a requirement of strict adherence to relevant legal and contractual norms.[292] Nonetheless, platforms retain substantial decision-making discretion regarding specific data monetization and utilization opportunities that fall within the scope of permitted uses. For example, though platforms are prohibited from transferring data to third-party opioid advertisers, they may choose to show their ads to users who demonstrate signs of opioid addiction. Balkin addresses similar concerns, highlighting users’ expectations that although platforms “should be able to monetize some uses of personal data,” they should refrain from using it “in unexpected ways” to users’ “disadvantage” or “in ways that violate some other important social norm.”[293] In other words, users expect platforms to balance their self-interests and users’ interests in a fair and responsible manner. Accordingly, users expect platforms to curb their profit-maximizing efforts when they could cause significant harm to users. This expectation is consistent with the relational contract model, which prohibits the infliction of disproportionate harm.[294] Behavior that causes disproportionate harm to the contractual partner, Macneil contends, is fundamentally inimical to the relationship. “Whenever one side is seen as . . . misusing power,” contractual solidarity is jeopardized, and the parties will question the benefits of continued interdependence.[295]

This is the context that should be interpolated into judicial interpretations of data monetization-related undertakings. For example, when Meta undertakes to use personal information so as to present users with advertisements that it believes are “relevant” to them, the term “relevant” should be interpreted as “potentially beneficial” and “not likely to cause disproportionate harm.”[296] Analogous to a best efforts commitment, this undertaking requires Meta to fairly and diligently balance the parties’ competing interests with respect to the personal data in the platform’s possession.

5. Human Rights and Procedural Requirements

Social media platforms often go a long way to show their commitment to human rights, democratic values, and public safety. One notable example can be found in the Facebook Oversight Board (FOB), as expressed in its Charter. The FOB is an institution charged with overseeing Facebook’s moderation decisions by reviewing selected appeals from impacted users.[297] It consists of a diverse and reputable board of members who enjoy substantial leeway.[298] The FOB’s decisions are binding on Meta when pertaining to specific content,[299] and it is designed to function similar to a “supreme court” for content moderation.[300] As for its substantive law, on top of commitment to apply Facebook’s Community Standards, the FOB’s Charter includes an explicit commitment to protect the freedom of expression[301] and an express undertaking to duly balance this human right against other considerations.[302] Similar commitments are also found in Facebook’s Community Standards themselves,[303] are conveyed in public announcements made by Facebook executives,[304] and are also reflected in other platforms’ publicly expressed positions.[305]

A recent tweet by Elon Musk, X’s owner and former CEO, provides another illustration of platforms’ express commitment to free speech and public safety. In response to a report accusing X of presenting advertisements next to neo-Nazi content and other hate speech,[306] Musk announced: “Above everything, including profit, X works to protect the public’s right to free speech.”[307] Such commitment to freedom of speech, Musk reasoned, is crucial for having a “thriving democracy” and a “flourishing society.”[308] This explicit pledge by X’s CEO to prioritize freedom of speech is naturally balanced against X’s commitment “to keep the platform safe from hateful conduct, abusive behavior, and any violation of Twitter’s rules”[309] or, for instance, to safeguard the integrity of democratic processes[310]; and striking a proper balance between these competing priorities inevitably requires X to establish suitable policies and processes, a commitment it readily accepts.[311]

The upshot of these examples is twofold. First, they show that leading platforms’ explicitly undertake to implement human rights and acknowledge their responsibility to protect public safety and democratic values. Second, these examples underscore that a commitment to this complex balancing of rights and interests inherently entails a commitment to designing fair and effective procedures.[312] Relational contract theory sheds light on these commitments and explains why they should be viewed as part and parcel of the contractual arrangement. Platforms’ undertakings in this regard attest to and reinforce the relational nature of social media contracting. As Robert Gordon contends, relational contract theory shows how “economic purposes and actions are deeply embedded in social fields, in densely woven webs of local customs, conventional morals, bonds of loyalty and entrenched power hierarchies.”[313] Furthermore, according to Macneil, relational contract theory offers “a vision of factors” that allows “humanitarian conclusions.”[314] Relational contracts, he explains, manifest “such broad norms as distributive justice, liberty, human dignity, social equality and inequality, and procedural justice” among others.[315] While these considerations do not necessarily appear in every relational contract, they are clearly compatible with the norms and expectations manifested by social media contracting. Platforms’ repeatedly expressed assurances regarding users’ free speech, public safety, and procedural fairness thus demonstrate and intensify the relational nature of social media contracting. Relational contract analysis supports recognizing human rights and procedural fairness requirements as inherent to the contractual arrangement and can justify their wider adoption as fundamental standards in social media governance.

C. Relational Contracting En Masse

A potential weakness in the approach proposed here is its seemingly individualistic perspective. As Douek argues with respect to content moderation, an individualized “post-by-post” focus is misguided and will likely result in an inappropriate legal response, for it is not a single user that counts but rather the “system” as a whole.[316] Indeed, an individualistic approach is inherently incomplete for understanding how social media operates. It may also lack the ability to fully account for all the interests involved. But the contractual approach proposed here can avoid this pitfall. It espouses a more comprehensive analysis, that looks beyond the bilateral dimension of user-platform relationships. As this Section explains, relational contract theory provides normative underpinnings for adding an aggregatory dimension to the contractual examination, thus demanding complex and widescale interest balancing that accounts for the multi-layered structure of the interests involved.

Let us first return to the insights implicit in the public law analogies for social media governance. When devising their content moderation policies, platforms necessarily take a network-oriented approach. They generate moderation rules and procedures necessary for supporting a public of users[317] and engage in “governance.”[318] Like states, platforms develop means and tools for managing the masses—a system of “mass speech administration”[319]—and this is how they perceive their business. An individualistic approach to content moderation is, therefore, insufficient for comprehensively understanding content moderation. Similarly, it is also impossible to make sense of data-related practices and policies using an individualistic lens alone. Rather, to understand how platforms harness and monetize data, we must grasp the “fundamentally collective nature of data,” i.e., “that the personalized content a user receives is strongly driven by rich data gathered about other users around the globe.”[320] Relational contract theory provides the framework for achieving such an understanding without stepping outside the borders of the parties’ agreement.

Relational contracts, as stated, are all about context; they are never detached from the social environment in which they exist. The parties’ expectations are necessarily formed within a social context, and contractual norms are designed to protect those context-contingent expectations. In the social media context, contractual norms can and do arise in all “dimensions” of the social interaction between users and platforms. They arise from: (1) the individualistic dimension (platform—specific user), (2) the communal dimension (platform—network/group of users), and (3) the societal dimension (platform—public as a whole). All such norms should be accorded equal attention, irrespective of the relational “dimension” from which they emerge. This means, inter alia, that ignoring the aggregatory (i.e. non-individualistic) dimensions of social media contracting results in important contractual norms being overlooked.

Going beyond users’ interests as “nodes” in the contractual network constellation enriches the analysis and makes it more accurate. A similar point was stressed by Elkin-Koren et al. in advocating for a “contractual network” approach to social media relationships.[321] This approach can arguably be framed as a subcategory of relational contracting[322] that allows focusing on network-oriented contractual norms.[323] This focus, however, which elevates one (important) dimension over others, lacks the commitment of relational contract theory to provide a holistic analysis that treats any norm-generating dimension of the interaction as rigorously as is contextually required.[324] Indeed, individual interests may diverge quite significantly from what is best for the network.[325] For example, a prohibition on generating disproportionate hardship to individual users, as mandated by a bilateral relational contract analysis, could produce a balance that occasionally favors the individual’s interests over those of the network.

Furthermore, as Balkin rightly notes, “large platforms like Facebook, Google, and Amazon have so many end users that a requirement that they must act in the interests of their end users effectively requires them to act in the interests of the public as a whole.”[326] In relational contract language, this means that once the public of users is large enough to converge in all important respects with the general public in its entirety, a comprehensive contractual analysis must address also aggregatory interests that have become public interests. For example, platforms’ ability to destabilize and engineer the social order, both through moderation policies and data exploitation, is nearly unparalleled, and users can legitimately expect that this ability will not be used in a way that is harmful to them.[327] This expectation requires platforms to weigh the public interest when addressing phenomena like misinformation, hate speech, organized silencing, targeted harassment, mass utilization of trolls and bots, and harvesting and exploitation of personal data.[328] Platforms are thus expected to consider the societal impact of their decision-making. Users’ interests in such cases inevitably extend beyond their individual or communal well-being and relate also to their need for social stability, public safety, and collective liberty. A similar point was recently stressed by Rory Van Loo, who thoroughly showed how an excessively narrow understanding of consumer law has led scholars and policymakers to overlook its ability to address various areas of concern, including social media and the inimical impact of disinformation on democratic resilience.[329] As he stated:

[F]ully weighing consumers’ interests with respect to a transaction means not only viewing the immediate economic effects that the transaction has on the consumer—such as the immediate financial harm—but also the ways that many such transactions, aggregated across society or over the course of a consumer’s life, affect that consumer’s less immediate personal interests, such as in democracy and long-term health.[330]

Van Loo’s analysis, however, primarily focused on public law aspects of consumer law,[331] calling for more comprehensive regulatory and legislative action to allow consumer law to fulfill its public purposes. The approach suggested here outlines a different path, offering tools and rationales for implementing public-oriented considerations through private law adjudication,[332] thereby circumventing the substantial constitutional and political barriers often associated with direct regulatory frameworks.[333]

The kind of multi-dimensional interest balancing suggested here is not a far cry from established practices. Indeed, platforms often engage in this balancing and openly undertake to act accordingly.[334] What follows, is that users are entitled to bring contractual claims to protect their legitimate expectations in all dimensions in which relational norms are generated and are operating, whether as individuals, as members of a network, or as members of the public as a whole. Importantly, recognizing the collective dimensions of the interaction, and enforcing the obligations they include, is distinct from a public policy analysis, which relies on extraneous norm-setting.[335] To more accurately reflect the parties’ expectations, it may be beneficial to conceptualize an aggregate party in the relational contract (such as the “network” or the “public”) and examine the platform’s responsibilities toward this collective entity. This theoretical tool may assist in analyzing relational contracting on a large scale, or “en masse.” It can be particularly effective in class action litigation, aiding to recognize group-bound commitments as they transition from aggregatory dimensions into bilateral interactions, thereby influencing contractual expectations within this framework.

D. Relational Contract Theory: Bridging the Gap Between Theory and Doctrine

The friction between relational contract theory and contract law doctrine has produced a somewhat ambiguous outcome. On the one hand, while many contract law scholars and economists have enthusiastically embraced this novel category of contractual arrangements, courts have generally been reluctant to develop and commit to a unified body of legal doctrine pertaining to “relational contracts.”[336] Consequently, the direct translation of relational contract theory into a well-defined body of doctrine remains scarce.[337] On the other hand, the development and influence of relational contract theory in academic discourse have been suspiciously correlated with significant shifts in various fields of contract law doctrine.[338] This correlation may suggest a substantial indirect influence of relational contact theory on contract law.[339] The intricate and largely empirical questions of whether and how relational contract theory affected contract law in practice go beyond the scope of this Article. However, for our purposes suffice it to note that courts possess ample doctrinal resources to enforce relational contractual norms, and they have arguably done so in other contexts in the past, without formally adopting the “relational contract” label.[340]

In particular, the tendency of modern contract law toward fragmentation[341] and contextualization[342] is highly consistent with relational contract theory. This tendency, Feinman argues, has effectively allowed for the “spinning off of subfields”—such as insurance law, landlord-tenant law, and product liability—that are premised on principles very much aligned with those stressed by Macneil.[343] As Feinman illustrated in this respect, by contextually employing traditional contract law doctrines and terminology (e.g. “reasonable expectations,” “good faith,” “[w]aiver and estoppel,” “misrepresentation,” “implied warranties,” etc.), courts have developed these fields to materially reflect many of the substantive and methodical insights of relational contract theory.[344] “Therefore,” Feinman argues, doctrinal implementations of relational norms can be pursued by “identify[ing] and develop[ing] additional subfields that can operate as independent relational contexts, as do insurance, landlord-tenant, and products liability law.”[345]

The theoretical and factual backgrounds provided above can offer justification for adopting this approach, as well as guidance for its implementation, in the context of social media contracting. As stated, the divergence of this contractual arrangement from typical consumer contracts warrants a similar divergence in contract law analysis and application for courts to properly adjudicate contractual disputes between users and platforms. Moreover, the prevalence of social media contracting today further underscores the need to design a specialized framework for addressing this wide-scale phenomenon. Such considerations arguably support the development of a new contract law subfield, aptly named “social media law,” which could produce a more robust and coherent legal framework. Nonetheless, distinguishing social media contracting from other fields of contract law is by no means a prerequisite for implementing the proposed approach. Nuanced implementation of existing contract doctrines can achieve substantial progress in this respect, and initial steps toward such implementation are accordingly proposed below.

V. Litigating Social Media: A Doctrinal Framework for Adjudicating Platform-User Disputes

Having crossed over into the world of private law, and having provided a theoretical framework to better understand relational norms and horizontal obligations in the social media context, we can now proceed to outline the potential practical implications of the proposed approach. In this Part, I accordingly provide a doctrinal analysis to equip courts and litigators with practical tools for redressing social media harms through contract-based adjudication.

Indeed, the application of contract law to social media contracting, properly construed, invites a reimagining of the legal duties of social media platforms. As will be explained below, applied correctly, contract law doctrine can certainly facilitate the imposition of fairness and diligence duties on platforms with regard to content moderation. Consequently, contract law doctrine can be applied to bar platforms from acting selfishly or opportunistically, from arbitrarily imposing one-sided policies, from inflicting disproportionate harm, and from making users bear the entire burden of failures to enforce moderation policies. The proposed contractual approach can also offer useful tools for addressing some of the problems associated with platforms’ data-related practices and even their business model itself, and it can importantly provide a long-sought legal hook for the judicial enforcement of norms such as human rights and procedural fairness—an outcome advocated by many.[346] However, to prevent excessive judicial encroachment on platforms’ discretion, the doctrinal solutions should be carefully selected, and a deferential standard of review, alongside a clear preference for equitable remedies, should be adopted. Should courts adopt the suggested doctrinal solutions, they will then have to contend with the significant challenge of gradually developing a common law of social media, a mission I believe to be long overdue.

A. Best Efforts Commitments: Good Faith and Diligence in Content Moderation

As stated above, one of the central differences between social media contracting and traditional consumer contracts is that the service platforms provide cannot be thoroughly defined in specific terms. Key aspects of content moderation are necessarily flexible and entail contextual, ad hoc decision-making. This inevitably affords platforms substantial discretion concerning the core features of the service they offer for sale. That is, platforms can define the substantive content of their obligations (e.g., what constitutes restricted hate speech) and have leeway to devise policies and systems for their performance (e.g., how the algorithm treats borderline cases). For this reason, in social media contracting, the strict liability standard is replaced by a best efforts commitment.[347] Inevitably, users must trust that platforms do, in fact, do their “best” to meet users’ expectations.

The term “best efforts” as a contractual commitment has generated much confusion over the decades. While some courts have equated the best efforts commitment with the duty to act in good faith,[348] Edward Farnsworth asserted that these duties should be distinguished from one another, with best efforts having “diligence as its essence.”[349] The best efforts obligation, he elaborates, imposes a “more exacting obligation” compared to the duty of good faith, “though it presumably falls short of the standard required of a fiduciary, who is required ‘to act primarily for the benefit of another in matters connected with his undertaking.’”[350] Thus, under Farnsworth’s construction, a best efforts commitment imposes a more rigorous duty than good faith but less rigorous than loyalty. For our purposes, we can call this a duty of fairness.[351]

In Bloor v. Falstaff Brewing,[352] the Second Circuit in fact adopted just such an “intermediate” standard, aspiring for “proportionality.”[353] In this case, the defendant undertook to make its best efforts to promote the beer manufactured by the plaintiff. It was found, however, that the defendant had engaged “in a number of misfeasances and nonfeasances, which could have accounted in substantial measure for the catastrophic drop” in the plaintiff’s beer sales.[354] The court concluded that despite the defendant’s “right to give reasonable consideration to its own interests,” it had “breached its duty to use best efforts” to promote the plaintiff’s beer.[355] Thus, the Bloor court’s finding aligned with Farnsworth’s assertion: namely, a best efforts obligation does not bar the promisor from considering her own interests. At the same time, however, it also imposes an “added obligation” to that originating from the implied warranty of good faith.[356] Indeed, “‘[b]est efforts’ requires more than ‘good faith’, which is an implied covenant in all contracts.”[357]

Like other case-law on this matter, Bloor applies a standard of reasonableness. That is, the promisor may “reasonably consider” her own interests when she takes measures to promote the promisee’s interests.[358] As another court stated in a subsequent case, “[c]ourts construing a best-efforts provision that does not specify the performance to be required commonly hold the promisor to the standard of the diligence a reasonable person would use under the circumstances.”[359] Notably, a similar standard to this was adopted by the drafters of the Restatement of Consumer Contracts for cases where sellers have substantial discretion in defining reciprocal contractual rights and obligations and translating them into discretionary performance.[360]

Best efforts clauses often appear in relational contracts,[361] such as distributorship and publishing agreements. Indeed, “relational contract theory has excelled in its treatment of . . . specific types of express or implied terms, like best-efforts provisions.”[362] As Goetz and Scott explain, “the best efforts cases hinge on two factors, strategic adaptation to the conflict of interest between the parties and the problem of managerial incompetence.”[363] Best efforts clauses, according to them, are premised on the genuine attempt to maximize the joint exchange-surplus, using an interest-balancing approach.[364] And these clauses reflect a contractual model of cooperation, trust, and inability to foresee all future contingencies in a long-term relationship.

Holding social media platforms to their best efforts commitment is not only compatible with the relational nature of social media contracting, but it would also effectively support the call to subject platforms to a duty to fairly balance their self-interest against users’ competing interests.[365] The latter duty exceeds the implied warranty of good faith but is less stringent than the fiduciary obligation. Platforms would accordingly be restricted from acting selfishly or opportunistically, grabbing the exchange-surplus and burdening users with the full cost of failures to enforce moderation policies. They would instead be required to affirmatively consider users’ interests and equitably balance those interests against their own and to provide their service diligently, as a “reasonable platform” would do under the circumstances.[366] Additionally, as described at the outset, many scholars favor subjecting platforms to various types of public law norms, such as due process and transparency requirements, international human rights, freedom of speech, and democratic participation.[367] The proposed contractual approach offers interesting potential in all these respects, laying out normative and doctrinal foundations for recognizing such implied, and certainly explicit, contractual guarantees of human rights protection and social responsibility.[368]

The first significant upshot of this framework would thus be the potential subjugation of social media platforms to a contractual duty to institute procedures and design systems that ensure fair and diligent content moderation.[369] Platforms’ promises to provide the “best” content moderation or to “strive” to enforce moderation policies “with uniform consistency” as accurately and “as quickly as possible” should not be treated as mere puffery.[370] While platforms convey their intent to preserve discretion and acknowledge the inevitability of mistakes, they nonetheless remain committed to exercising this discretion with good faith and diligence and attempting to minimize the rate of error.[371] To uphold these duties, platforms must set up appropriate moderation procedures and design systems apt to enforce their policies at the necessary scale and pace.[372] It follows that platforms’ failure to properly structure such moderation systems would constitute a breach of contract actionable in court.[373]

Furthermore, as stated above, though human rights have become the touchstone for content moderation, all have failed to recognize an enforceable legal obligation to uphold such norms—absent robust legislative and regulatory intervention that is often practically unattainable. The suggested contractual approach offers interesting potential in this respect as well. As demonstrated above, platforms’ explicit commitments to human rights, freedom of speech, and public safety are part and parcel of the contractual apparatus.[374] To be sure, and as further explained below,[375] platforms must enjoy substantial deference in this context, and courts should generally avoid replacing platforms’ judgment with their own with respect to substantive content-related decision-making. Nonetheless, platforms’ obligation to enforce their policies with best efforts—i.e. fairly and diligently—stands. Free speech must be weighed, as promised, while fairly balanced against other considerations like public safety and democratic integrity,[376] as well as platforms’ self-interest in profit. This act of balancing is social media platforms’ voluntarily undertaken legal duty, from which specific obligations may be derived. In this context, “fairness and diligence” may denote non-discriminatory algorithms, consistently enforced moderation policies, and genuine efforts to foster an unbiased public forum. Conversely, a willful attempt to influence elections or public opinion, arbitrary and inconsistent interpretations and takedowns of hate speech, or failure to diligently deplatform terrorists could each constitute an actionable breach of contract.

B. Fair Data Utilization

Imposing relational contract norms on platforms could also impact the way the data economy is currently administered. As stated, platforms offer their service in exchange for users’ personal data and users’ consent to be exposed to third-party advertising.[377] And being profit-driven corporations, a decision by platforms not to offer users an alternative payment structure, is presumably based on an estimate that extending such an offer will likely result in lower revenues.

Indeed, a business model constructed on data in lieu of monetary remuneration appears to offer at least two significant advantages for platforms. First, in addition to the benefit of not having to nudge users to rethink payment, this model means that users cannot evaluate the actual price they are paying for the service because no monetary price has been set. This is both because they cannot fully understand what data is collected and how it is monetized and because users are unaware of network effects, which make the value of personal data hinge on the aggregate effect of data collection.[378] Moreover, some (like the Lewis and Darnaa courts[379]) may even fall for the illusion that the service is provided for free. Second, this model allows platforms to conduct price discrimination, charging a different value from each user based on the monetizability of their personal attributes,[380] further increasing their share in the exchange surplus.

These advantages suggest that the data-in-lieu-of-monetary-remuneration model is utilized by platforms to grab uninformed users’ surplus. That is, users may be paying more than they are willing to pay, and platforms are reaping a greater share of the joint exchange surplus than what users’ would have agreed to.[381] This may potentially lead to entirely inefficient bargains, where users engage in transactions in which the user loses more value than the platform gains. For example, a user may value the privacy interests attached to her data, together with the utility derived from devoting her time to alternative activities, in the amount of $10 while the platform’s revenue from her activity, including her contribution to network effects, amounts to only $5. Moreover, personal data collected from individual users can be utilized to conduct algorithmic price discrimination in markets for other goods. Using personal information in this way may create additional harm to users, of which they are mostly unaware, further increasing the price they are in fact paying for the services social media platforms provide.[382]

Under the relational contract structure, a pricing strategy aimed at grabbing uninformed users’ surplus, let alone generating negative utility for them, could be prohibited. As stated, relational contracts require parties to fairly share the benefits and burdens arising from the relationship. They are required to reasonably balance between competing interests and should never abuse power imbalances to extract unwarranted surplus at the expense of the other party. Furthermore, adopting a contracting en masse approach for particularly big platforms would expose their data monetization practices to an aggregate welfare analysis, thereby requiring platforms to avoid economically inefficient activities or activities generating public harm. Concrete implications may vary. Platforms could be restricted from utilizing personal data in a manner that unreasonably harms users, and may be required to provide users with alternative payment models—data-based or fee-based. Such an option could endow users with more autonomy to decide what price they are willing to pay and whether they want to surrender personal information. This could also possibly mitigate some of the price discrimination exercised by platforms, a questionable policy to start with under a “fairness” requirement. By preventing platforms from grabbing the exchange-surplus, the relational contract model might contribute to rebalancing the scales and facilitate more equitable distribution of revenues, a great share of which platforms currently hog, inter alia, by enjoying the barrier to competition in the field.

In many respects, imposing the suggested fairness duty on platforms’ data-related practices would produce a similar outcome to that Balkin and Zittrain aim to achieve in their information fiduciaries proposal, and for a similar rationale. As Balkin also acknowledges, due to the barter structure of social media contracting, “people do not expect the same degree of concern from online service providers” as they do from doctors or lawyers, for example.[383] Therefore, Balkin argues, platforms should be held to a “more limited” version of fiduciary duty of loyalty.[384] Under this duty, they would be allowed “to monetize some uses of personal data” but not to “use the data in unexpected ways to the disadvantage of people who use their services or in ways that violate some other important social norm.”[385] Platforms, according to Balkin, are therefore required to engage in interest balancing. However, whereas an appeal to a standard of “loyalty”—which is premised on a commitment to selflessness—generates significant difficulties,[386] a fairness duty is more suitable for regulating this requirement.

Of course, this is only an initial analysis. Requiring platforms to change their business model based on the imposition of a “fairness” requirement will involve many tradeoffs which, in turn, will require careful deliberation. Flexible equitable remedies, examined in the context of collective litigation procedures, can support this type of cautious intervention.

C. The Contributions of Equity: Judicial Discretion and Remedial Approach

An important caveat must be attached to any attempt at mobilizing private law for reforming social media governance: the judiciary’s inherent institutional limitations, alongside the potential impact on speech interests of its meddling in the free operation of the social media market, are serious grounds for concern. This concern is further exacerbated in light of the need to employ open-textured fairness and diligence-based standards in reviewing contract-based causes of action, which would require that courts conduct a flexible and contextual analysis.[387] As I explain below, however, private law has the necessary resources to address these concerns. Indeed, the fear of judicial over-encroachment, and the judiciary’s institutional limitations can be substantially mitigated by thoughtful application of various principles and doctrines associated with what can be called “equity-mode” adjudication—or, as put by Henry Smith, by “going meta.”[388]

Equity is arguably a necessary mode of justice where the regulated activity is so complex and unpredictable[389] that its governing arrangements cannot be fully reduced into predetermined universal rules of conduct.[390] And as human lives become reliant on ever more complex and interconnected technologies, relationships, and arrangements, which develop at an exponential rate, resorting to equity-style adjudication becomes unavoidable.[391] In this sense, Smith explains, “[e]quity is part of law’s response to the world’s inevitable complexity.”[392] According to Smith, equity operates as second-order law: a system that takes “regular” first-order law as its input and is capable of correcting errors “from without,” thereby allowing first-order law to remain more general and certain.[393] Thus, Smith stresses, equity “solve[s] complex and uncertain problems”[394]—particularly those involving “polycentricity, conflicting rights, and opportunism”[395]—“by going to a new level of law. Equity is law about law, or meta-law.”[396]

In line with this conceptualization of equity, equity-based adjudication seems naturally suited to deciding disputes related, generally, to relational contracts,[397] and, more specifically, it appears especially crucial in the context of social media contracting, given the inherent nature of the problems it generates. Social media disputes are likely to “involve many items (people, objects, activities) and many interdependencies, leading to complexity,”[398]—i.e., they are polycentric in nature.[399] Moreover, they will tend to involve context-dependent conflicting presumptive rights, making it hard to determine, let alone foresee, which right will prevail in specific circumstances or to ex ante design sufficiently detailed universal rules.[400] And importantly, social media contracting inevitably generates substantial risks for opportunism on the part of platforms, which enjoy nearly uninhibited control over users’ rights that is practically impossible to oversee.[401]

Despite its bad reputation,[402] equity does not bestow on courts a free pass to exercise unbounded ex-post discretion.[403] Rather, as Smith argues, equity, correctly construed, operates through the use of cabined, or tailored, ex-post standards, often accompanied by ex-ante prophylactic rules, and seeks to avoid “broad blunderbuss invocations of ex post fairness and morality.”[404] By using various “safety valves,” including triggers (such as “deception, bad faith, vulnerability, and hardship”[405]), maxims, presumptions, and self-imposed restraint, by granting various defenses, and by limiting itself to certain remedies, equity avoids a “broad ex post approach,” which would be “not just chilling but destabilizing and inimical to the rule of law.”[406]

To be effective, courts’ adjudicatory approach to social media contracting disputes must accommodate cabined contextuality and flexibility, alongside a possibility for deliberation of the systemic impact of concrete choices made by platforms. An instrument capable of achieving these goals is the class action,[407] itself an outgrowth of equitable jurisdiction.[408] But aggregated litigation is not always suitable, nor does it offer sufficient safeguards against judicial over-encroachment on platforms’ decision-making. To prevent arbitrary or paternalistic judicial intervention and to allow social media to develop organically in ways that honor conflicting interests and expectations, additional cabining of judicial discretion is necessary.[409] Among the various equitable tools suitable for this purpose, two promising possibilities are briefly discussed below.

1. The Platform Judgment Rule: Insulating Good Faith Decision-Making from Judicial Scrutiny

Concerns about judicial over-encroachment on managerial discretion in activities requiring complex decision-making have long arisen in equity adjudication. This was particularly acute in corporate law, where general reluctance to allow courts to replace directors’ good faith judgment with their own, and thereby hinder the free operation of the business sector, gave rise to the business judgment rule. I will contend below that establishing an analogous “platform judgment rule” could incentivize platforms to adopt due process norms in their content moderation and improve their moderation systems. Platforms will thus be able to insulate themselves from judicial scrutiny and retain leeway in conducting their affairs while overall accountability will dramatically improve.[410]

The business judgment rule is a “presumption that in making a business decision the directors of a corporation acted on an informed basis, in good faith and in the honest belief that the action taken was in the best interests of the company.”[411] Under this presumption, liability is rarely imposed “simply for bad judgment.”[412] As long as the presumption’s elements are not rebutted by the plaintiff, only a grossly negligent decision[413] lacking “any rationally conceivable basis” could give rise to liability.[414] However, where the plaintiff demonstrates, for example, that a contested decision was tainted by conflict of interest, resulted from substantial failure to exercise oversight,[415] or made on an uninformed basis,[416] courts will review the merits of the decision itself and pass muster on its “fairness.”[417]

The rationales underlying the business judgment rule can strongly inform an analogous platform judgment rule. In the corporate context, substantial insulation from liability is strongly justified by the voluntariness of risk assumption by shareholders, the need to allow such risks for the business environment to flourish, and the imperfection of “after-the-fact litigation” for evaluating corporate business decisions, often made quickly and under conditions of uncertainty.[418] Similar rationales can ground a platform judgment rule. First, users, like shareholders, voluntarily agree to assume the risk of inaccurate and imperfect moderation, namely, encountering undesirable content,[419] and this risk is imperative for the maintenance of a vibrant speech environment. Accordingly, and similar to the business judgment rule, insulation from scrutiny is necessary for the users themselves to be able to benefit from the interaction. Second, conducting a post-hoc evaluation of the reasonableness of specific content-related decisions is often very difficult because of the rate, scale, and uncertainty in which such decisions are made.[420]

From a more general perspective, the business judgment rule can be said to be “designed to effect a compromise—on a case-by-case basis—between two competing values: authority and accountability.”[421] These values “refer, respectively, to the need to preserve the board of directors’ decision-making discretion and the need to hold the board accountable for its decisions.”[422] The way in which the business judgment rule is currently applied to resolve the tension between authority and accountability is replete with doctrinal and theoretical nuances but this is beyond the scope of the discussion in this Article. What is noteworthy, however, is that an analogous platform judgment rule appears highly appropriate for inclusion in any future scheme of social media governance where adjudication takes a central role.

Clearly, such a rule would gradually evolve and be fine-tuned by courts to develop its own unique attributes. But from the outset, ensuring platforms’ insulation from judicial scrutiny—conditional on making informed decisions that are untainted by conflicting interests and rest on professional rationales—would powerfully incentivize them to guarantee users’ due process rights and incorporate safeguards in their systemic structure to counter error-making and mechanism biases. For instance, a platform judgment rule could incentivize the inclusion of independent bodies in platforms’ complaint review processes,[423] similar to the Facebook Oversight Board, for example.[424] Charged with overseeing Facebook’s moderation decisions, the FOB consists of a diverse and reputable board of members who enjoy substantial leeway in reviewing Facebook’s decision-making.[425] A platform judgment rule would subject the FOB’s decisions only to standards of good faith and independence in using its discretion for Meta to be entirely insulated from substantive judicial review. Such a rule would strongly incentivize Meta—and, in its wake, other platforms—to further guarantee the FOB’s independence, as well as bolster the esteem of its members, and expand the scope of its authority.[426]

2. Equitable Remedies versus Damages

Though damages are allegedly the “most prevalent remedy at law,”[427] they should be the exception and not the rule in social media cases. Equitable remedies, in contrast, would allow courts to provide a far more nuanced and dynamic remedial response to breaches of duties by platforms, facilitating just results in concrete cases while also addressing systemic concerns. Moreover, as opposed to damages, equitable remedies would not generate negative incentives that could lead to over-removal of content or make the platform’s continued operation unviable.

Equity has developed an assortment of unique remedies, geared mainly toward reversing unjust gains and preventing wrongs from materializing or persisting, rather than toward providing compensatory damages or punishing so-called immoral behavior. The various trajectories of the different equitable remedies (such as the constructive trust, injunctive relief, disgorgement, specific performance, and equitable rescission)[428] make them not only generally more suitable for relational contracts,[429] but also particularly crucial for redressing social media harms. One fundamentally relevant attribute of equitable remedies is their ability to address ongoing systemic concerns. For example, courts can grant injunctions that can later be altered to respond “to events that were unseen when the remedy was first granted,”[430] while the power to hold parties in contempt gives courts the ability to ensure long-term compliance.[431] Perhaps most interestingly, courts may also appoint “what could be called ‘equitable helpers,’” that is, “officers of the court who take discovery, dispose of property, or investigate compliance.”[432] This tool assists in mitigating courts’ deficiencies in expertise and expediency, and it helps judges design better instructions for platforms, which have systemic implications. At the same time, equitable defenses such as undue hardship, unclean hands, or laches, alongside equity’s self-imposed constraints, can guarantee that equity’s powerful remedies will be exercised with restraint.[433]

Compounding the averseness to award damages in the social media context is their inherently inimical effect on free speech. Social media contracting facilitates and promotes powerful speech interests, and this can be directly impaired should monetary damages become routine and, particularly, if courts award presumptive or punitive damages.[434] The availability of such damages as a remedy for users would expose platforms to massive liability and materially deprive them of the necessary breathing space for conducting good faith contextual balancing in content moderation. Not only would this undermine the purpose of Section 230,[435] but it would also unreasonably expand platforms’ contractual obligation. As explained, social media contracting is premised on a best efforts commitment on the part of the platform, implying users’ expectation that mistakes may occur[436] and their assumption of a risk of resulting damages. Nonetheless, given the ongoing nature of the relationship, errors should be corrected (at least where correction does not create disproportionate hardship) to allow the contract’s continuation.

One exception to a “no-damages” rule should be made, however, to accommodate cases where gross negligence or bad-faith behavior is demonstrated. Platforms must be deterred from turning a blind eye to calamities in the making and, more generally, from blatantly disregarding their contractual commitments. In such extreme cases, compensatory and even punitive damages would be morally justified[437] and should remain available to plaintiffs.[438] Importantly, moreover, since the difficulty to detect a breach would necessarily result in underenforcement, platforms would be under-incentivized to uphold their duties in the absence of damages in such cases. Punitive damages can be used to mitigate this deterrence gap.[439]

Conclusion

This Article has aimed to fill a substantial and substantive gap in the academic, political, and legal discourse by suggesting a contractual approach to social media governance. As described, most of the current scholarship in the field looks exclusively to public law for answers and seeks to develop regulatory tools to address current challenges. The ensuing result, however, is far from satisfying, as privately owned social media platforms continue to determine “matters of such public significance without any form of accountability, transparency, or meaningful public input.”[440] By providing both doctrinal and normative frameworks for subjecting platforms to judicial scrutiny, this Article challenges the legal status quo and offers a promising way forward.

Both relational contract theory and contract law doctrine offer practical and theoretical resources to contend with many of the harms caused by social media, without running the risk of throwing the baby out with the bathwater. The contractual approach proposed here would not be vulnerable to platforms’ immunity claims or constitutional constraints and would facilitate the contextual and nuanced deliberation necessary for examining the difficult tradeoffs social media governance requires. In addition, the proposed approach can duly account for horizontally developing expectations between the private parties who jointly create social media. And thus, rather than attempting to impose extrinsic norms through regulatory encroachment on speech-enhancing, privately owned services—with all the constitutional and political obstacles such an approach entails—a contractual approach offers innovative legal solutions. It would be attentive to platforms’ need for leeway in designing speech norms which they deem fit and would be able to dynamically address changing expectations and adjusting norms. And importantly, the suggested approach would bring courts back into the social media governance arena and harness their institutional advantages, including their apolitical nature and the iterative and contextual process of common law adjudication.