Abstract
Distributed punitive events—commonly described as pile-ons, cancellations, or online mobbing—are frequently analyzed through the lens of individual moral failure or cultural pathology. This paper proposes an alternative analytical frame: the political economy of collective punitive participation. Drawing on economics, social psychology, organizational behavior, and media studies, the paper argues that pile-ons are not primarily expressions of spontaneous moral outrage but are structurally produced outcomes of identifiable incentive systems. Three structural conditions are examined in detail: the engagement incentives built into digital platform architectures, which make punitive participation individually rational regardless of aggregate consequences; the moralization of participation, which converts engagement incentives into the appearance of moral obligation; and the mechanisms by which harm is crowd-authored in ways that systematically distribute responsibility so thinly that no individual actor perceives themselves as causally accountable for the aggregate outcome. The paper concludes that pile-ons are a predictable product of current institutional design, and that preventing them requires redesign at the level of the incentive structures that make them rational, rather than merely at the level of the individual moral choices that those structures shape.
Introduction
The pile-on is, on its face, a puzzling phenomenon. A large number of individuals, most of whom have no personal connection to the target and no direct stake in the outcome, choose to invest time and attention in the public condemnation of a person they do not know, for conduct they have typically encountered only through a highly mediated and abbreviated account. The investment is individually small—a post, a comment, a share—but the aggregate is substantial, and its effects on the target can be severe and lasting. Why do so many people participate? And why does their participation so consistently exceed what proportionate response to the original offense would require?
The dominant popular explanation appeals to the moral emotions: people are outraged, and they express their outrage. This explanation is not wrong, but it is radically incomplete. Moral outrage is a common human experience that does not, under most circumstances, produce pile-ons. People experience outrage at conduct they observe daily—in their workplaces, their communities, their personal relationships—without producing the kind of distributed, amplified, rapidly escalating collective condemnation that characterizes a pile-on. The difference between those ordinary experiences of outrage and the pile-on is not a difference in the intensity of moral feeling; it is a difference in the institutional environment in which the feeling occurs and the structural incentives that environment provides for expressing it in particular ways.
This paper applies the tools of political economy—the analysis of how incentive structures shape behavior in the absence of centralized coordination—to the phenomenon of distributed punitive participation. The term “political economy” is used deliberately to signal that the analysis is concerned with the structural conditions that make certain behaviors rational and others irrational, rather than with the moral evaluation of the behaviors themselves. The argument is not that pile-on participants are bad actors; it is that they are actors in an environment that makes punitive participation individually rational, that converts that rationality into the appearance of moral necessity, and that then distributes the resulting harm so diffusely that no actor bears visible accountability for it. Understanding this structure is prerequisite to designing environments in which the incentives operate differently.
Section I: Engagement Incentives
1.1 The Architecture of Engagement
Digital platforms are not neutral channels for human expression. They are institutional architectures designed to maximize a specific behavioral outcome: engagement, measured in the time users spend on the platform, the frequency with which they return, and the rate at which they generate and interact with content. This design objective is not incidental or secondary; it is the commercial foundation on which the dominant platforms are built, because engagement drives the advertising revenue and data collection that are the primary sources of platform income (Zuboff, 2019). Every feature of platform design—notification systems, algorithmic content curation, reaction mechanisms, follower counts, visibility metrics—is, at minimum, informed by and, in many cases, specifically optimized for this objective.
The relationship between engagement and punitive content is not accidental. It is a consistent empirical finding, replicated across platform types and cultural contexts, that content expressing moral outrage generates more engagement than content expressing most other emotional registers. Brady et al. (2017) demonstrated, across a dataset of millions of social media messages, that the inclusion of moral-emotional language—language expressing outrage, disgust, contempt, or moral condemnation—was associated with significantly higher rates of sharing and amplification. Each moral-emotional word in a message increased the rate of diffusion by approximately twenty percent. This finding has been replicated and extended by subsequent research (Brady et al., 2021; Rathje et al., 2021), consistently showing that outrage-expression is among the highest-engagement content categories across social media platforms.
The implication for pile-on dynamics is direct: platforms whose architecture is optimized for engagement will systematically amplify outrage-expressing content above content expressing other registers, including deliberative, qualifying, or contextualizing content. The pile-on does not merely occur on platforms; it is accelerated by the structural features of those platforms in ways that are predictable from first principles of engagement optimization.
1.2 Individual Rationality and Collective Irrationality
The concept of a collective action problem, developed in economics and political theory (Olson, 1965; Hardin, 1968), describes situations in which individually rational choices aggregate into collectively suboptimal outcomes. The pile-on exhibits the structure of a collective action problem in reverse: rather than the under-provision of a collective good due to free-riding, it produces the over-provision of a collective harm due to individually rational participation.
For each individual participant, the decision to join a pile-on involves a calculation, typically implicit, of costs and benefits. The costs are low: the time required to post a comment or share a condemnatory message is minimal, and the social risk of participation—in an environment where the participation is visible and aligned with apparent majority sentiment—is similarly low. The benefits are tangible: social visibility, expression of in-group alignment, the affective satisfaction associated with moral condemnation, and, in many cases, measurable platform metrics in the form of likes, replies, and follower growth. The individually rational calculation favors participation.
At the aggregate level, however, the outcome is frequently disproportionate to any legitimate evaluative or corrective function, as established in White Papers 1 and 2. The target of a pile-on does not receive proportionate feedback that enables learning and improvement; they receive a volume of condemnatory attention that produces outcomes—reputational destruction, professional loss, psychological harm—that no individual participant intended or calculated. Each participant’s individually rational act contributes to a collectively harmful outcome that is, at the aggregate level, irrational relative to any stated corrective purpose.
This is the core political economy of the pile-on: it is a stable equilibrium produced by individually rational choices in an environment that does not price the aggregate harm into the individual decision. Because the harm is diffuse and the benefit is concentrated, participation remains rational for each individual actor even as the aggregate becomes increasingly disproportionate.
1.3 Asymmetric Visibility and Participation Incentives
The incentive structure of pile-on participation is further shaped by asymmetric visibility—the condition in which the target of commentary is more visible than any individual participant. This asymmetry has two effects on participation incentives. First, it reduces the perceived power of the individual participant, which reduces their sense of causal responsibility for the aggregate outcome: each participant can truthfully observe that their single contribution, relative to the scale of the pile-on, is negligible. Second, it increases the perceived safety of participation, because the social risk of expressing a minority view—the risk that comes from being visible and identifiably wrong—is substantially reduced when the participant is one of thousands expressing the same view against a single, highly visible target.
Both effects reinforce participation incentives. The reduction of perceived causal responsibility removes a friction that might otherwise inhibit participation. The reduction of social risk removes another. The result is that asymmetric visibility systematically lowers the threshold for participation, ensuring that the pile-on is accessible to actors who would not participate in a more symmetric evaluative environment where their individual contribution would be more visible and their individual accountability more legible.
1.4 Temporal Dynamics and Momentum Effects
The incentive structure of pile-on participation changes over time in ways that are themselves structurally significant. In the early stages of a pile-on, participation involves some social risk: the verdict on the target has not yet been established as majority sentiment, and early participants are taking a position that might not be vindicated by subsequent social consensus. As the pile-on gains scale, this risk diminishes and then inverts: the majority position having been established, the social risk now attaches to non-participation or to dissent. Participants who might have been ambivalent in the early stages now face a different calculation: the cost of expressing skepticism about the pile-on is higher than the cost of joining it, because skepticism is now a minority position in a moralized social environment.
This temporal dynamic produces momentum effects that are independent of any new information about the original offense. The pile-on grows not because each successive wave of participants has independently assessed the offense and found it worthy of condemnation, but because the social momentum of the pile-on has altered the incentive structure for participation in ways that make joining the rational choice regardless of independent assessment. Sunstein (2019) identifies this as a form of social cascading in which the expressed preferences of earlier actors provide informational and normative pressure on later actors that is independent of, and frequently overwhelms, their independent judgment. The result is that the scale of a pile-on is a poor signal of the severity of the underlying offense; it is, instead, a signal of the pile-on’s own momentum.
Section II: Moralized Participation
2.1 The Conversion of Incentives into Obligations
The analysis of engagement incentives in Section I might suggest that pile-on participation is purely self-interested behavior dressed in moral language. This reading, while partially accurate, misses a crucial structural element: the moralization of participation is not merely a rhetorical disguise for self-interest. It is a genuine psychological process through which engagement incentives are converted into felt moral obligations, and through which the act of participation is experienced by the participant not as a choice but as a duty. Understanding this conversion process is essential to understanding why pile-ons are so difficult to interrupt from within.
Moral psychology has extensively documented the phenomenon of moral licensing—the process by which prior moral behavior increases the subjective permissibility of subsequent self-serving behavior—and its inverse, moral credentialing (Monin & Miller, 2001). Less attention has been paid to the inverse process at work in pile-on dynamics: the conversion of self-serving behavior into moral behavior through the mechanism of social moralization. When an act that is individually beneficial—participation in pile-on engagement—is framed by the social environment as morally required, the actor experiences the act not as a self-interested choice but as a moral performance. The engagement incentive and the moral obligation become, in the actor’s experience, indistinguishable.
2.2 The Structure of Moral Obligation in Pile-On Contexts
Several mechanisms produce the experience of moral obligation in pile-on participation.
The framing of non-participation as complicity. Perhaps the most powerful mechanism for converting engagement incentives into felt moral obligations is the framing of non-participation as a form of complicity with the offense being condemned. In this framing, silence is not a neutral position but an endorsement; failing to add one’s voice to the condemnatory chorus is treated as equivalent to defending the target. This framing is structurally powerful because it removes the option of principled non-participation: the actor who does not join the pile-on is not merely abstaining from a social activity but is, within the moral logic of the framing, actively choosing the wrong side. The choice set is effectively reduced from three options—participate, abstain, or dissent—to two: participate or be complicit.
Waytz et al. (2013) identify the fairness-loyalty tension as a central dynamic in whistleblowing decisions, but an analogous tension operates in pile-on participation: actors face pressure to express loyalty to the in-group’s moral consensus, and loyalty expression through participation is functionally difficult to distinguish from genuine moral conviction. The social environment does not provide the actor with tools to disentangle these motivations, and the actor typically does not attempt to disentangle them.
Moral emotion as participation trigger. Haidt (2001) and subsequent researchers in moral psychology have established that moral judgments are frequently produced by rapid affective processes and subsequently rationalized rather than primarily produced by deliberative reasoning. In pile-on contexts, this means that participation is typically triggered by a moral-emotional response—outrage, disgust, contempt—that precedes and shapes rather than follows from assessment of the relevant facts. The moral emotion authenticates the participation: because the actor genuinely feels outrage, the participation feels genuinely moral rather than self-interested, regardless of whether the outrage is proportionate, well-informed, or categorically appropriate in the senses established in White Papers 1 and 2.
The authenticity of the moral emotion is, in this sense, not in dispute. What is in dispute is whether authentic moral emotion in response to a mediated, incomplete, and algorithmically selected account of a situation is a reliable basis for participation in a punitive event. The political economy of pile-ons exploits the authentic character of moral emotion by providing conditions—algorithmically amplified outrage content, social consensus signaling, framing of non-participation as complicity—that reliably produce that emotion and then channel it into participation before deliberative processes can introduce moderating assessment.
Competitive moral signaling and the ratchet effect. White Paper 1 introduced competitive moral signaling as a mechanism of disproportionality. In the context of moralized participation, this mechanism operates as what might be described as a moral ratchet: once participation in condemnation has been established as the morally correct position, the intensity of condemnation becomes a signal of moral seriousness. Actors who express mild condemnation are, within this logic, less morally committed than actors who express severe condemnation. The result is systematic upward pressure on the intensity of expressed condemnation that is independent of any new assessment of the underlying offense.
Cikara and Van Bavel (2014) identify this dynamic in intergroup moral contexts, where the intensity of outrage expression functions as an in-group loyalty signal. In pile-on contexts, the same dynamic operates without requiring established group membership: the shared event of the pile-on creates a temporary moral community in which intensity of condemnation serves as the primary credential. Actors who moderate their condemnation risk being perceived as insufficiently committed to the moral consensus—a social risk that, as established in Section I, the incentive structure of pile-on participation makes costly.
2.3 The Immunization of Participation Against Correction
The moralization of participation has a particularly consequential structural feature: it immunizes participation against the kinds of correction that would, in a non-moralized context, be available. When a participant is presented with evidence that the account of the offense on which they based their participation was incomplete, inaccurate, or contextually distorted, the moral framing of their participation creates resistance to updating.
Several mechanisms produce this resistance. Moral consistency motivation—the desire to maintain consistency between one’s expressed moral positions and one’s subsequent behavior—creates pressure against acknowledging that the participation was based on inadequate information, because such acknowledgment implies that the moral condemnation was not fully warranted (Festinger, 1957). Social commitment effects make it costly to withdraw from positions that have been publicly expressed and that have generated social reinforcement from the in-group. And motivated cognition produces selective evaluation of new information: evidence that supports the original condemnatory verdict is weighted more heavily than evidence that complicates or contradicts it (Kunda, 1990).
The result is that moralized participation is significantly more resistant to correction than non-moralized participation would be. A participant who joins a pile-on on self-interested grounds and subsequently learns that the account was distorted faces a calculation that is, in principle, straightforward: the self-interested basis for participation no longer applies, so participation should be withdrawn. A participant who has experienced their participation as moral duty faces a much more complex set of pressures that resist the same withdrawal. The moralization of participation thus not only enables the pile-on to grow more rapidly than self-interest alone would produce; it prevents it from being corrected once it has begun.
Section III: Crowd-Authored Harm
3.1 The Problem of Distributed Causation
The most challenging analytical feature of the pile-on as a harm-producing event is the radical distribution of its authorship. The harm caused to the target of a pile-on is real, often severe, and frequently well-documented: loss of employment, destruction of professional reputation, sustained psychological harm, and in some cases more severe consequences (Ronson, 2015; Citron, 2014). But identifying the author of this harm—the actor whose conduct caused it—is extraordinarily difficult under conventional frameworks of individual causal responsibility, because the harm is the aggregate product of many individually minor acts, no one of which, considered in isolation, could plausibly be identified as the cause.
This is not merely a philosophical puzzle. It has direct practical implications for institutional design and accountability. In legal contexts, the difficulty of establishing individual causation in distributed harm events is a significant barrier to civil remedy (Citron, 2014). In moral contexts, the distribution of authorship enables each participant to truthfully maintain that their individual contribution was not the cause of the harm, because relative to the aggregate, it was not. In social contexts, the diffusion of authorship prevents the development of shared accountability norms: because no individual participant experiences themselves as having caused the harm, the development of norms against participation does not benefit from the feedback loop that would exist if causation were visible and concentrated.
3.2 The Mechanics of Harm Distribution
The distribution of harm authorship in pile-on events is not random; it follows identifiable structural patterns.
Threshold models and critical mass dynamics. Granovetter’s (1978) threshold model of collective behavior describes situations in which individuals decide whether to join a collective action based on the number of others already participating. In pile-on contexts, this model predicts that participation will accelerate as the pile-on gains scale, because each new participant lowers the threshold for subsequent participation by demonstrating that participation is both socially legitimate and socially safe. This dynamic produces rapid growth in participant numbers that is mechanically independent of any assessment of the underlying offense, making the scale of the pile-on a function of its own internal dynamics rather than of the severity of the conduct being condemned.
The threshold dynamic also distributes moral responsibility in a way that is structurally protective of individual participants: each participant can truthfully observe that they joined a large ongoing event rather than initiating harm, that many others made the same choice, and that their individual contribution to the aggregate was negligible. This observation is accurate at the individual level and misleading at the aggregate level, but the individual level is where moral and legal accountability frameworks are designed to operate.
The layering of different contribution types. Pile-on harm is rarely a single-type aggregate. It is composed of qualitatively different contributions that interact to produce the overall harm: initial condemnatory posts, amplification through sharing, secondary commentary that elaborates and extends the original condemnation, contact with the target’s employers or professional associations, creation and circulation of compilations of the target’s prior statements or conduct, and in some cases coordinated harassment extending across multiple platforms. Each of these contribution types involves different actors, different levels of intentionality, and different degrees of causal proximity to the ultimate harm. The layering of these types makes causal attribution not merely diffuse but analytically fragmented: different elements of the harm have different proximate causes, none of which is the complete cause of the aggregate harm.
Citron (2014) identifies this layering as a primary obstacle to legal remedy for online harassment, noting that existing legal frameworks for individual tortious conduct do not map cleanly onto distributed harm events in which the proximate cause of each element of harm is different from the proximate cause of the aggregate. The same analytical problem applies to moral and institutional accountability frameworks, which were similarly designed for environments in which causation was more concentrated and attributable.
Temporal distribution and the persistence of searchable harm. Pile-on harm is not bounded by the duration of the event. The content produced during a pile-on remains searchable, shareable, and re-circulatable indefinitely after the event has concluded. The target’s name, associated with condemnatory content, may remain the top search result for years. New actors who encounter the target in professional or social contexts may search their name and encounter the archived pile-on content without knowing that the event has concluded or that the account on which it was based may have been incomplete. This temporal distribution of harm extends the causal chain of the pile-on far beyond its active participants and far beyond the event’s temporal boundaries, while simultaneously making it impossible to identify a discrete set of actors responsible for the ongoing harm. Platform design that preserves and surfaces archived condemnatory content without contextualizing its origin is a structural participant in this extended harm causation, though it is rarely analyzed as such.
3.3 The Absence of Feedback Between Participation and Consequence
A central feature of crowd-authored harm is the systematic absence of feedback between individual participation and aggregate consequence. In environments where causation is concentrated, actors receive feedback: the lawyer who loses a case, the doctor whose patient is harmed, the engineer whose bridge fails—each receives feedback that connects their conduct to its consequences in ways that inform future behavior and sustain accountability norms. This feedback is the mechanism through which individual actors learn the real-world weight of their decisions and through which institutional norms about the acceptable limits of conduct are maintained.
In pile-on environments, this feedback mechanism is absent. Individual participants do not observe the aggregate harm; they observe only their individual contribution, which is small. They do not receive information about the target’s actual experience of the aggregate; they receive, at most, additional content about the event’s further development, which is typically framed in the evaluative terms of the pile-on itself. And because no formal accountability structure connects their participation to its consequences, they receive no structured feedback of the kind that would, in other contexts, inform their future behavior.
The absence of this feedback loop has two consequences. First, it prevents the development of calibrated personal norms about pile-on participation: actors do not learn, from experience, what the actual weight of their participation is, because the weight is distributed and invisible. Second, it prevents the development of shared social norms against participation: because no individual participant experiences themselves as having caused harm, the social learning process that would otherwise generate prohibitive norms against harm-causing behavior does not engage. The result is that each pile-on occurs in a social environment that has not been reformed by the consequences of prior pile-ons, and that the same structural conditions that produced prior events will produce future events with equal efficiency.
3.4 Platform Complicity and Structural Accountability
The analysis of crowd-authored harm would be incomplete without addressing the role of platforms as structural participants in the harm-production process. Platforms are not passive channels through which pile-on participation flows; they are active architectural participants whose design choices—algorithmic amplification of outrage content, notification systems that draw additional participants into ongoing events, search and discovery features that make pile-on targets and content mutually findable, and content preservation policies that extend harm indefinitely—contribute materially to the aggregate harm. Zuboff (2019) frames this participation in terms of the surveillance capitalist model that makes engagement-optimized architecture commercially rational for platforms regardless of its social consequences. Gillespie (2018) analyzes the governance dimensions, examining how platform policy choices about content moderation, algorithmic curation, and terms of service enforcement constitute de facto institutional positions on what kinds of harm are acceptable, even when those positions are never explicitly articulated.
The difficulty of holding platforms accountable for their structural participation in crowd-authored harm mirrors, at a larger scale, the difficulty of holding individual participants accountable: causation is distributed, the connection between architectural choice and aggregate harm is mediated by many individual decisions, and existing legal and regulatory frameworks do not readily accommodate distributed institutional causation. Section 230 of the Communications Decency Act in the United States, for example, provides broad immunity for platforms from liability for user-generated content, effectively insulating platforms from legal accountability for the harm-amplifying consequences of their architectural choices (Citron & Wittes, 2017). The political economy of pile-on harm is, in this sense, partly a political economy of platform immunity: the commercial rationality of engagement-optimized architecture is sustained by a legal framework that does not require platforms to internalize the costs of the harms that architecture produces.
Conclusion
The political economy of pile-ons reveals that distributed punitive events are not spontaneous expressions of collective moral judgment. They are structurally produced outcomes of incentive systems that make participation individually rational, that convert engagement incentives into felt moral obligations, and that distribute the resulting harm so diffusely that no actor—individual participant or institutional platform—bears visible accountability for the aggregate. This structural analysis does not exculpate individual participants; it locates individual behavior within the institutional conditions that shape and enable it, and it directs attention toward the level at which effective intervention must occur.
Interventions designed at the level of individual moral education—campaigns asking people to think before they post, appeals to empathy for pile-on targets, or calls for greater online civility—are structurally insufficient responses to the problem identified here. They address the individual decision without modifying the incentive structure that makes participation rational, the moralization process that converts participation into felt duty, or the harm distribution mechanisms that prevent feedback from operating. Effective intervention requires redesign at the level of these structural conditions: platform architectures that do not amplify outrage content above other registers, accountability mechanisms that make the aggregate consequences of distributed participation visible to individual actors, and institutional frameworks that impose on platforms an obligation to internalize at least some portion of the costs their architectural choices impose on pile-on targets.
White Paper 4 addresses the cultural infrastructure dimension of these failure modes, examining the long-term costs of environments in which sincerity is not protected and early-stage formation is not safeguarded. The argument developed there complements the structural analysis of this paper: if the political economy of pile-ons explains why they occur, the degradation of cultural infrastructure explains what is cumulatively lost when they continue to occur without structural constraint.
Notes
Note 1 — The term “political economy” is used here in its classical sense, referring to the analysis of how incentive structures, resource distributions, and institutional arrangements shape collective outcomes in the absence of central coordination. It does not imply a specifically Marxist or any other particular ideological framework; the analytical tools employed here are drawn from mainstream economics, public choice theory, and institutional analysis and are intended to be applicable across ideological perspectives.
Note 2 — The Brady et al. (2017, 2021) findings on moral-emotional language and diffusion rates are among the most methodologically robust empirical findings in this literature, having been replicated across multiple platforms, time periods, and cultural contexts. They are cited here as evidence of a structural relationship between engagement architecture and outrage amplification, not as a claim about the intentions of platform designers. The relationship between engagement optimization and outrage amplification is, in significant measure, an emergent property of optimization processes rather than an intentionally designed feature.
Note 3 — The application of Granovetter’s (1978) threshold model to pile-on participation is an extension of its original application to collective violence and social movements. The model’s predictive power in these contexts has been well established; its application to online collective condemnation events is a more recent development. For readers seeking empirical work on threshold dynamics in online collective behavior specifically, Watts (2002) and Centola (2010) provide useful entry points.
Note 4 — Section II’s analysis of the immunization of moralized participation against correction should not be read as a claim that moral outrage is never warranted or that pile-on participants are always wrong in their assessment of the original offense. The argument is structural: that regardless of whether the initial condemnation is warranted, the moralization of participation creates psychological and social conditions that resist the incorporation of corrective information. This resistance is harmful even when the initial assessment was accurate, because it prevents the proportionality recalibration that legitimate evaluation requires.
Note 5 — The legal analysis in Section III draws primarily on United States law because the relevant empirical and doctrinal literature is most developed in that context, and because the Section 230 framework has been particularly influential in shaping the global digital platform environment. The analysis is not intended as a comprehensive comparative legal treatment; readers interested in the comparative regulatory dimensions are directed to Keller (2019) and Suzor (2019) for broader international perspectives.
Note 6 — The distinction between individual moral accountability and structural institutional accountability that runs throughout this paper reflects a methodological commitment rather than a moral position. The paper does not argue that individual participants bear no moral responsibility for their participation; it argues that focusing exclusively on individual moral responsibility, without analyzing the structural conditions that make such participation rational and feel morally obligatory, is analytically insufficient and practically ineffective as a basis for intervention.
Note 7 — Readers may note that the analysis of crowd-authored harm in Section III has significant structural parallels to economic analyses of negative externalities—costs imposed on third parties that are not reflected in the prices paid by those who produce them. This parallel is deliberate. The political economy of pile-ons is, among other things, an externality problem: the costs of pile-on participation are borne by targets rather than by participants, and no pricing mechanism internalizes those costs into the individual participation decision. Pigouvian solutions—taxes or regulations designed to make external costs visible to those who impose them—are one family of potential institutional design interventions, though their application in this context raises significant practical and constitutional questions.
References
Brady, W. J., Wills, J. A., Jost, J. T., Tucker, J. A., & Van Bavel, J. J. (2017). Emotion shapes the diffusion of moralized content in social networks. Proceedings of the National Academy of Sciences, 114(28), 7313–7318. https://doi.org/10.1073/pnas.1618923114
Brady, W. J., McLoughlin, K., Doan, T. N., & Cikara, M. (2021). How social learning amplifies moral outrage expression in online social networks. Science Advances, 7(33), eabe5641. https://doi.org/10.1126/sciadv.abe5641
Centola, D. (2010). The spread of behavior in an online social network experiment. Science, 329(5996), 1194–1197. https://doi.org/10.1126/science.1185231
Cikara, M., & Van Bavel, J. J. (2014). The neuroscience of intergroup relations: An integrative review. Perspectives on Psychological Science, 9(3), 245–274. https://doi.org/10.1177/1745691614527464
Citron, D. K. (2014). Hate crimes in cyberspace. Harvard University Press.
Citron, D. K., & Wittes, B. (2017). The internet will not break: Denying bad samaritans Section 230 immunity. Fordham Law Review, 86(2), 401–423.
Festinger, L. (1957). A theory of cognitive dissonance. Stanford University Press.
Gillespie, T. (2018). Custodians of the internet: Platforms, content moderation, and the hidden decisions that shape social media. Yale University Press.
Granovetter, M. (1978). Threshold models of collective behavior. American Journal of Sociology, 83(6), 1420–1443. https://doi.org/10.1086/226707
Haidt, J. (2001). The emotional dog and its rational tail: A social intuitionist approach to moral judgment. Psychological Review, 108(4), 814–834. https://doi.org/10.1037/0033-295X.108.4.814
Hardin, G. (1968). The tragedy of the commons. Science, 162(3859), 1243–1248. https://doi.org/10.1126/science.162.3859.1243
Keller, D. (2019). Who do you sue? State and platform hybrid power over online speech. Hoover Institution Press.
Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulletin, 108(3), 480–498. https://doi.org/10.1037/0033-2909.108.3.480
Monin, B., & Miller, D. T. (2001). Moral credentials and the expression of prejudice. Journal of Personality and Social Psychology, 81(1), 33–43. https://doi.org/10.1037/0022-3514.81.1.33
Olson, M. (1965). The logic of collective action: Public goods and the theory of groups. Harvard University Press.
Rathje, S., Van Bavel, J. J., & Van der Linden, S. (2021). Out-group animosity drives engagement on social media. Proceedings of the National Academy of Sciences, 118(26), e2024292118. https://doi.org/10.1073/pnas.2024292118
Ronson, J. (2015). So you’ve been publicly shamed. Riverhead Books.
Sunstein, C. R. (2019). Conformity: The power of social influences. New York University Press.
Suzor, N. P. (2019). Lawless: The secret rules that govern our digital lives. Cambridge University Press.
Tufekci, Z. (2017). Twitter and tear gas: The power and fragility of networked protest. Yale University Press.
Watts, D. J. (2002). A simple model of global cascades on random networks. Proceedings of the National Academy of Sciences, 99(9), 5766–5771. https://doi.org/10.1073/pnas.082090499
Waytz, A., Dungan, J., & Young, L. (2013). The whistleblower’s dilemma and the fairness–loyalty tradeoff. Journal of Experimental Social Psychology, 49(6), 1027–1033. https://doi.org/10.1016/j.jesp.2013.07.002
Zuboff, S. (2019). The age of surveillance capitalism: The fight for a human future at the new frontier of power. PublicAffairs.
