Accelerationism and Algorithms of Violence: Mind that Beef
Abstract: This article focuses on the role of social media algorithms in online-to-offline violence among vulnerable, high-risk child and youth populations. Specifically, we present the understudied phenomenon of internet banging in communities with gang-involved young people. We highlight how seemingly trivial affective exchanges on performative social media spaces have escalated quickly to visceral violence in the real world. Long-term exposure to these visible affective-to-visceral violence norms may degrade already-troubled communities, lowering their threshold to accept both violent action and inaction in response to aggression. Such violence-mediated, ideology-independent nihilism forms the basis of accelerationist thought, typically linked to extremist communities, but increasingly being linked to public violence perpetrated – performatively – by children. We therefore recommend that affective-to-visceral behavioral pathways to violence on social media be taken as serious threats to societal safety and resilience, including the risk of adversarial state manipulation for a visceral form of cognitive warfare.
Introduction
Accelerationism is a nihilistic belief that the contemporary social and political structures in Western society must be destroyed, and that violence should instrumentalize this collapse by aggravating social tensions. Rooted in Marxist and cultural underground thought and currently associated with militant extremism in neo-Nazi communities, accelerationism has also been linked to recent cases of children who have committed mass shootings and to nihilistic violent extremism (NVE) communities with several child-aged violent perpetrators known to have forced their victims to degrade themselves for community entertainment.
The current third-generation wave of online radicalization increasingly involves accelerationist appeal at the expense of clear-cut ideological signaling, targeting vulnerable children to commit violent attacks with accelerated “flash-to-bang” timelines, using humor, memes, and inflammatory “shit-posting” to desensitize and manipulate behaviors. Studies suggest that ideology has taken a back seat in contemporary online nihilistic radicalization among child and youth populations, with violence being the only clearly identifiable goal.
We argue that contemporary social media algorithms can quickly escalate online affective engagements, typically involving performative insults, to visceral violence within their communities in the real world.
In this article, we specifically explore the role of social media in online-to-offline violence perpetrated by juveniles and young people, with three assumptions: (1) contemporary social media algorithms of exposure and engagement are highly effective in shaping the behavior of attention-captive audiences; (2) hostile, aggressive, and violence-prone messaging between two “beefing” parties on social media mirrors real-time physical altercations, particularly if the parties are easily accessible to each other in the real world; and (3) in the long term, these social media beefs can shift a given community’s thresholds to accept both violent action and inaction in response to visible aggression.
We demonstrate this with the growing phenomenon of internet banging among juveniles and young people affiliated with gangs. We argue that contemporary social media algorithms can quickly escalate online affective engagements, typically involving performative insults, to visceral violence within their communities in the real world. The communities in which these gang-involved bangers live or operate typically have a history of sustained exposure to violence, including high levels of homicide. These communities are further characterized by chronic degradation of social capabilities including collective efficacy, limited trust in formal security providers (i.e. local police), and heightened perceptions of dying before age 20 among young residents.
We further argue that some hostile nations already understand these pockets of vulnerability in American societal contexts, posing a risk that these reflexive affective-to-visceral behaviors might be influenced by malign actors.
We are suggesting that internet banging cases, deemed “out-of-control” by local law enforcement in several of these troubled communities, demonstrate a specific digital pathway to violence involving affective-to-visceral exchanges between participating youngsters in the community. This moves the lever of escalation from grievance typically associated with ideological violence in fringe groups, to a naturalized tendency in more mainstream society to react viscerally and fast to known, measurable triggers especially by known rivals (e.g., insults or threats of violence). We further argue that some hostile nations already understand these pockets of vulnerability in American societal contexts, posing a risk that these reflexive affective-to-visceral behaviors might be influenced by malign actors.
Empirical and Theoretical Considerations
To explore this violence-mediated pathway to conditions of insecurity and collapse in high-risk, degraded communities, we tapped into the emerging phenomenon of internet banging. This refers to the use of social media platforms (e.g., Facebook, Twitter/X, or YouTube) by juveniles and youth affiliated with gangs or similar less organized groups to exchange insults or threats online which then escalate to real-world violence in their communities. Typical incidents started as seemingly trivial performative “beefs” between rivals and ended in fatal or non-fatal violence including retaliatory homicide and spiraling tensions communitywide.
Before proceeding to a more detailed characterization of internet banging, we would like to underscore two very important caveats. First, it would be inaccurate, overly alarmist, and “street-illiterate” to assume violent social media posts exchanged between young bangers instantly or always translate to real-world violence. Second, the demonstrated pathway to collapse is purely hypothetical, with no real cases of such communitywide collapse known to exist in the U.S.A., meaning that the associated threat is not imminent; however, if left ignored for long, this extreme scenario can be both possible and potent.
Grief, Beefs, and the Attention Economy
Studies based on New York, Chicago, and UK digital street culture consistently show that online platforms play a role in intensifying visibility and exposure to drill, the rap-based interpersonal playout between gang-involved youth, which has long been associated with hostile posturing, confrontation, specific references to violence including murder, and escalations to legal battles, violent escalation, and even death. These studies further show that while online violent personas may not necessarily translate to violent intent, the online violence itself can be driven by the attention economy that lead to increased youth behaviors of self-disclosure and surveillance of others on social network sites. Drill-related “diss tracks” on social media can additionally spill over to real-world rivalries in different postcodes, culminating in homicide.
Another study on escalation mechanics in disinvested neighborhoods showed that, while adolescents “do not go online with the intention to fight,” they identified specific social media features (e.g., public comments, video-streamed content) to be particularly hard to ignore, which could intensify trivial interpersonal insults and conflict (e.g., over threats and romantic relationships) into beefs leading to violent offline confrontations. Patton and colleagues, whose work originally presented the concept of internet banging, found in a later study that social media served as a vector for reciprocal violence, facilitating the exchange of online violence in real time. Studies also find that young gang-involved people use social media to express both grief and aggression, and that expressions of loss frequently preceded aggressive posts.
Notably, the role of social media in internet banging is deemed one of “adaptive structuration” that unintentionally primes and directs the behaviors of vulnerable, conflict-prone young people to violence. Together, these studies suggest that among young populations exposed to high levels of violence and the degradation of capabilities within their communities, social media has the capacity to unintentionally instrumentalize conflict, further leading to conditions of collapse within these troubled communities.
Specifically, social media algorithms optimized for high performance around exposure and engagement metrics may offer the following opportunities: (1) providing visibility to real-time insults and hostility especially in vulnerable moments, leveraging the captive attention of beefing parties who are always on social media; (2) presenting beefing parties with compelling and creative methods to express and retaliate, extending existing practices (e.g., drill); and (3) compressing the timeline to escalation by allowing real-time co-writing of an online script whose authors may have easy physical access to each other.
Algorithmic Exposure, Priming, and Violence Thresholds
Advertising and social learning theories appear to form the basis on which social media engagement systems are operated and optimized. In particular, the effectiveness of algorithmic content delivery is measured in terms of “impressions” that quantify exposure to the message, and operationally, these involve three key dimensions – reach, frequency, recency. Reach refers to the number of unique users who are exposed to a message. Frequency is the number of times a message is served to the target, and this was introduced in advertising research in 1972 based on Krugman’s findings that “three exposures may be enough” to repeat a message so it achieves the three psychological goals of exposure (i.e., curiosity, recognition, reminder/decision). Recency was later introduced as a challenger dimension to frequency, and it emphasized “when” rather than “how many times,” meaning that messages are more relevant when served closer to the anticipated purchase window.
Social media changed many of the traditional media planning approaches (in print, TV), given the ubiquitous reach of mobile phones and their ability to leverage always-reachable, attention-captive audiences for maximum or highly customized exposure. These platforms have the ability to facilitate engagement between an influencer and their “community” of followers, as well as direct engagement between two or more individuals who may not even be known to each other. Contemporary social media operations have progressed even further, with automated algorithms of engagement-optimized recommendation systems.
For instance, a study by Brady and colleagues shows that these algorithms amplify PRIME (prestigious, in-group, moral, emotional) content, and certain human–algorithm interactions may lead to conflict, misinformation, and misperceptions. People may also form skewed, hyperpersonal impressions of others online, and choose to overlook some qualities of someone online due to the limited and curated nature of digital interaction. In the internet banging context, this means that the algorithm-driven, curated engagement on social media may be shaping the extreme ways in which young beefing parties respond; this response may be reflective of their relationships within their in-groups, their attitudes toward rivals, their normative beliefs toward aggression, and the long-term accumulation of their direct and indirect affective experiences within their community.
This idea of behavior optimization is furthermore consistent with social learning and reinforcement theory. In the early 1960s, Bandura demonstrated that aggression can be socially learned by an individual from observation alone, mediated by four cognitive processes (i.e., attention, retention, motivation, reproduction); furthermore, individuals were more likely to imitate aggressive behavior that they could observe being rewarded. Another thread of social learning theory, postulated by the General Aggression Model, suggests that exposure to violent media increases aggression in individuals through three “internal states” i.e., aggressive thoughts, angry feelings, and increased physiological arousal.
Studies by Carnagey et al. (2007) and Bushman & Anderson (2009) further show that repeated exposure to violence can reduce both neurocognitive response and helping behaviors against violence, with implications of individual desensitization to violence as well as the degradation of cohesive capability in a community to counter aggression and conflict. This is consistent with the Threshold Model of Collective Behavior (Granovetter, 1978), according to which an individual’s threshold of joining violence within a community will depend on a cost-benefit tradeoff shaped by several considerations (e.g., situational, continuity of behavioral dispositions, bandwagon-type response to observing other community members participate in violence). Loosely based on threshold models, instigators of violence have a threshold of zero, violence-prone individuals will have low threshold values, and the spread (or conversely, the stopping) of violence within the community will depend on the population distribution of thresholds. This framing is also consistent with the well-stablished epidemiological “Cure Violence” model of community violence, that public safety and public health programs use, to design violence intervention programs leveraging community peers, many with past gang-involvement themselves.
Two points are particularly made clear from the internet banging phenomenon: (1) social media makes the participation of others continuously visible, and (2) social media has the power to amplify emotive messages, including those that can instigate violence through affective-to-visceral behavioral pathways.
Overall, this short review shows that social media has distinct power, more consequential than other media formats, to shift both individual and collective thresholds of violent behavior, based on common media metrics of exposure and engagement alone. Two points are particularly made clear from the internet banging phenomenon: (1) social media makes the participation of others continuously visible, and (2) social media has the power to amplify emotive messages, including those that can instigate violence through affective-to-visceral behavioral pathways. Taken together, this helps connect the social media algorithmic conditioning of hostility, aggression, and violence to further degradation of vulnerable, troubled communities. This may have long-term implications of lowered thresholds for these communities to both accept violent action and inaction in response to aggression.
Discussion and Conclusion
On August 26, 2026, a landmark judgment directed Meta to pay $18 billion toward settling a social media addiction lawsuit, part of which will require the implementation of strong child-focused safety measures. We argue that this legal decision carries not just a punitive financial obligation for one social media company, but that it should be considered a sector-wide inflection point to re-examine algorithmic harm from the broader perspective of individual as well as societal wellbeing. This is also an opportunity for multisectoral stakeholders in technology safety, community resilience, law enforcement, and national security to collectively review cognitively safe rules of social media use. This may have particular implications for the design of content delivery, to avoid unintentional manipulation leading to violence and nihilism in society.
Children may be particularly vulnerable to unintended algorithmic manipulations, and as the future of the communities they belong to, it is important to address the social media harms of today starting with the highest-risk communities. In many ways, the logic that applies to young gang-involved persons also applies to children who find belonging in online violent extremist spaces (e.g., 764, black-pill communities) and who go on to commit mass shootings or stabbings. A common thread across these extreme, degraded communities is the casual acceptance of violence and nihilism, which in the case of gang-affiliated youth can come from expectations of low life expectancy. Future studies should also examine changing trends in accelerationism, traditionally associated with fringe groups, and whether a more diffuse, less extreme version of accelerationism is making its way to child and youth populations in troubled but more mainstream communities.
The ultra-optimization of social media algorithms for high performance, tracked in terms of exposure and engagement, may unintentionally support violence discourse.
This essay further supports two considerations of long-term societal degradation: (1) the existence of long-standing, inescapable, systemic structures of violence (e.g., measured in terms of gang presence, weakened social cohesion among others), and (2) lowered thresholds of permissible hostility, aggression, and violence in a focal community, with instigators having zero-value thresholds (e.g. measurable in terms of homicide, non-lethal violent incidents among others). Additionally, for beefing parties located within the same or a nearby geographic area, the potential for an online-to-offline ramp is very high, with implications for widening, retaliatory violence and spiraling, communitywide social tensions.
There is distinct value in studying proxy violence-mediated pathways of accelerationism, as presented in this essay. Most notably, it offers a demonstration of how the medium itself can be the primary vector for violence, rather than the specific involved parties or the specific involved messaging. The ultra-optimization of social media algorithms for high performance, tracked in terms of exposure and engagement, may unintentionally support violence discourse. For social media platforms in the eye of stormy conversations on user harm, efforts to roll back some of these harms can signal commitment to societal resilience and cognitive security, rather than reactionary compliance. It can be a good start to have collaborative, multisectoral conversations around measured/measurable social media metrics. More in-depth studies are needed on these digitally and algorithmically mediated, affective-to-visceral behavioral pathways in extreme, degraded communities.
This suggests a need to examine cognitive warfare not only as a contest over what and how populations react to information, but also as an adversarial process capable of shaping what individuals and communities feel and ultimately do.
The more immediate concern is that hostile states and non-state actors may deliberately exploit these existing vulnerabilities rather than manufacture them from scratch for cognitive warfare. The 2024 Southport riots in the UK, centered around a rumor that an asylum seeker was involved in a mass stabbing targeting a children’s dance party, resulted in 1,804 arrests and 302 officers being injured by January 2025. It appeared fueled by a social media algorithmic mediation involving about 30,000 X mentions across 18,000 accounts within a day and about 27 million impressions on amplifying posts from the account Channel3Now, demonstrating “the role that virality and algorithmic recommendations can play” even as authorities speculated over Russian involvement in what could not even be ascertained as a disinformation campaign. Elsewhere, in Ukraine, smaller-scale, foreign-influenced, foiled instances of mass school-violence have involved the recruitment of teenagers via Telegram and TikTok to conduct the attacks. These instances show how digital platforms can connect grievance and identity, financial incentive, and social media algorithmic manipulation to behaviors that culminate not merely in changed beliefs, but in arson, sabotage, riots, and other forms of physical violence-as-a-service.
Yet this study suggests the importance of looking inward, preventively identifying vulnerabilities in the highest-risk communities both online and offline, and overall adopting a societal resilience approach.
This suggests a need to examine cognitive warfare not only as a contest over what and how populations react to information, but also as an adversarial process capable of shaping what individuals and communities feel and ultimately do. While something like the weaponization of domestic gang rivalries poses a distant threat, local communities and security forces may need to consider the mediating role of foreign actors in existing adversarial dynamics both physical and digital (e.g., the placement of pig heads in Paris mosques by foreign actors to stoke inter-communal violence). Possible gang-related risks include direct gang recruitment for sabotage of embassies and existing money-laundering or courier services offered by foreign criminal actors with wide access to a community’s gang ecosystem. Discussion of cognitive warfare often remains distant from the environments in which its effects may ultimately be felt, with a greater focus on abstract whole-of-society outcomes such as erosion of public trust and long-term degradation of political and institutional legitimacy. The actual response furthermore fixates on the external adversarial involvement, which tends to be bogged down by attribution uncertainty leading to uncertainty in authoritative response. Yet this study suggests the importance of looking inward, preventively identifying vulnerabilities in the highest-risk communities both online and offline, and overall adopting a societal resilience approach.
A better understanding of the algorithmic structures – behaviorally shaping action or inaction around violence – can inform cognitive strategies to keep communities safe.
Specific to digitally and algorithmically mediated affective-to-visceral pathways, when we suggest that more research is needed on degraded, alienated, violent, or otherwise vulnerable online communities, we are talking about gamified recruitment, gig-based financial incentives, anonymous social networks, and constructed identities that may create behavioral architectures that facilitate, accelerate, inhibit, or redirect movement toward violence. Understanding those architectures offers protective, defensive value against cognitive warfare that blends into the ordinary machinery of online life, particularly that of communities already primed for harmful action. Cognitive security strategies could identify where digital environments intensify affect, normalize violence, lower behavioral thresholds, or create pathways from online participation to physical action.
We recommend that cognitive security practitioners treat digitally and algorithmically mediated, affective-to-visceral behavioral pathways as a serious dimension of the evolving cognitive warfare threat. This is aligned with other increasing calls for cognitive security to go beyond the limits of “messages and nonviolent methods” to embrace the full scope of political warfare, and with the acknowledgment of social media manipulation as an analytical domain in cognitive warfare operations. A better understanding of the algorithmic structures – behaviorally shaping action or inaction around violence – can inform cognitive strategies to keep communities safe. Ultimately, our more present concern is how hostile nations and non-state groups can intensify this visceral form of cognitive warfare, exploiting behavioral and psychological vulnerabilities in degraded, troubled communities. We hope that this article serves as an eye-opener to the real, albeit distant, threat of targeted cognitive manipulation with viscerally violent effects.
