At this point the analysis reaches its most counterintuitive claim: that resistance does not necessarily destabilize systems that circulate harm, and often stabilizes them instead.
In systems where violence persists as conserved flow, opposition is not external to the process. It is metabolized within it. Outrage, investigation, activism, moderation, and critique become dissipation channels: ways the system sheds pressure without altering its internal configuration. Resistance absorbs energy that would otherwise force redesign. It converts moral urgency into labor, time, affect, and attention, none of which require the system to internalize cost.
This explains a persistent empirical puzzle: why systems facing continuous, well-documented condemnation often remain unchanged.
The answer is not hypocrisy or bad faith. It is that resistance operates downstream of the invariant.
To see this clearly, consider the role of outrage. Outrage is an affective escalation that signals violation. In moral terms, it is oppositional. In signal terms, it is amplificatory. Platforms register it as engagement. Media register it as attention. Institutions register it as pressure, but only so long as that pressure does not crystallize into enforceable obligation. As long as outrage remains expressive rather than compulsory, it can be accommodated.
Activism follows a similar pattern. Awareness campaigns, exposés, petitions, and documentation efforts generate visibility and urgency. They are necessary for recognition. But in the absence of conversion mechanisms, they also perform a stabilizing function: they demonstrate that “something is being done.” Responsibility is displaced onto those doing the work. Institutions can point to civil society activity as evidence of concern while deferring structural intervention.
Moderation labor exhibits the same dynamic. Content reviewers absorb exposure. Trust-and-safety teams process volume. Burnout cycles through human nodes while the system persists unchanged. Harm is managed at the periphery so that core design remains intact. Exhaustion becomes a pressure-release valve.
Even investigation, arguably the most structurally oriented form of resistance, often fails to break the invariant. Investigators trace networks, archive evidence, and map flows. But unless attribution closes at a node capable of being compelled, investigation produces knowledge without consequence. The system survives scrutiny because scrutiny does not automatically force redesign.
None of this implies that resistance is futile or morally compromised. It implies that resistance that does not alter attribution topology is absorbed as maintenance.
This is why resistance appears from all polarities. Actors who otherwise disagree (platforms, activists, regulators, commentators) occupy different positions on the same constraint surface. They rely on the same invariant: that responsibility remains unattributed across the full causal chain. Their disagreements are orthogonal, not antagonistic. They push against each other without severing the surface itself.
The system only becomes unstable when polarity collapses: when condemnation no longer dissipates, when transgression no longer escalates, when amplification no longer externalizes cost. That collapse requires architectural responsibility internalization. And that is precisely what the system is structured to prevent.
If resistance is metabolized as dissipation, then participation in resistance does not confer innocence. It does not place the actor outside the system. It places them inside a stabilizing loop unless their action forces attribution closure. This is not a moral indictment. It is a positional fact. It marks the boundary of the old explanatory frame.
The reader can no longer imagine themselves as an external critic confronting a malfunctioning system. They are inside a system that uses critique as fuel unless critique carries a conversion mechanism with it.
At this point, a common reaction is withdrawal: cynicism, silence, disengagement. This too is stabilizing. Silence reduces load. The system continues.
The alternative is narrower and more demanding. It does not involve louder condemnation or purer motives. It involves identifying and forcing the specific points at which cost can no longer be externalized: where attribution must close or the system must restructure.
The next section names that boundary explicitly. It identifies the invariant the system cannot tolerate being made explicit, and why naming it removes the last procedural escape routes without offering consolation.
Wounds in Distribution
Cruelty, once enacted, does not remain local. It propagates.
In networked environments, harm is no longer confined to the site of production or to the bodies directly subjected to it. It becomes distributable, reproducible, and asynchronous. The wound detaches from origin and enters circulation. This chapter specifies that transition operationally: from cruelty as an act to cruelty as a distributed system effect.
From Event to Circulation
Classical accounts of violence treat harm as an event: bounded in time, space, and agency. A body is injured. A perpetrator acts. A witness observes. Response mechanisms (responsibility, repair, enforcement) are anchored to that moment.
Networked cruelty breaks this model.
Once recorded, encoded, and transmitted, the act persists independently of its initial conditions. The recorded suffering functions as a transmissible signal. Each replay, repost, crop, or transformation reproduces exposure to the same harm-content without repeating the original act. The representation of the wound becomes persistent within the system.
This is a systems property. Distributed architectures transform singular inputs into durable outputs wherever replication pathways are cheap and resilient. Under such conditions, harmful media behaves like a persistent payload within a network: it persists, mutates, and routes around suppression.
The Logistics of Harm
Distribution introduces logistics. Harm is routed, cached, mirrored, indexed, and reassembled across platforms, channels, and devices. Each layer increases distance between producer and downstream effect while expanding reach.
Three properties define distributed cruelty:
Detachment from authorship
The original actor is no longer necessary for continuation. Propagation becomes a function of replication affordances rather than ongoing production.
Latency and recurrence
Harm reappears at unpredictable intervals, affecting new audiences long after the originating event. The system enables recurrence of exposure without re-initiation.
Context collapse and ranking selection
When removed from circumstances, the material loses interpretive friction. What circulates is no longer “what happened” but “what sustains engagement.” Recommendation and ranking systems preferentially surface content that reliably drives interaction, without consistent sensitivity to harm semantics, thereby amplifying extreme material once it proves attention-efficient [148][149][150].
These properties convert suffering into infrastructure-compatible content.
Secondary Victims and Tertiary Harm
Distributed cruelty generates victims beyond the original target.
Primary harm is inflicted on the animal subjected to violence.
Secondary harm can impact exposed viewers and moderators through trauma load, stress injury, and desensitization effects.
Tertiary harm emerges systemically as norm erosion, normalization of spectacle, and degradation of civic trust [151][152].
These layers are not incidental in outcome terms. In cruelty networks, downstream distress functions as an observable indicator of effect. Circulation volume can function as a proxy metric of impact within these systems: proof that the act has forced attention and produced impact beyond the immediate victim.
The Inversion of Witnessing
Historically, documentation of violence is framed as bearing witness: recording in order to expose, remember, or prevent recurrence. In distributed cruelty systems, that function can invert.
Visibility does not reliably produce accountability. Under engagement-optimized metrics, exposure can increase distribution. Outrage and denunciation can be processed by platform systems as high-signal engagement, indistinguishable from endorsement at the level of ranking. The witness can become a relay.
This creates a structural bind for ethical actors. Engagement risks propagation, and disengagement risks abandonment. Distributed cruelty exploits this bind by forcing interaction within systems that are indifferent to intent.
Distribution as Power
In distributed systems, power is not located solely in decision-making nodes. It resides in routing capacity, amplification potential, and resilience to disruption.
Cruelty networks do not require centralized leadership or coherent ideology.
They require: infrastructures that optimize for engagement without semantic harm comprehension, payment and hosting pathways compatible with pseudonymity, and social environments where status is accrued through transgression.
Within these conditions, cruelty becomes a low-cost, high-impact signal. Its distribution demonstrates reach, dominance, and effective immunity to consequence.
Why Takedown Fails as a Primary Strategy
Conventional responses focus on removal: delete the content, ban the account, disrupt the channel. These measures can be necessary, but they are structurally insufficient as a primary strategy.
In distributed systems, takedown targets instances rather than persistence conditions. Content is redundantly stored. Accounts are disposable. Communities reconstitute via invitations, forks, mirrors, and backup archives. Removal is routinely outpaced by replication. In some contexts, enforcement pressure can increase cohesion and adaptive sophistication by confirming internal narratives of persecution or “proof of impact” [153][154].
Effective intervention therefore requires upstream constraints: throttling amplification pathways, disrupting monetization, and altering incentive gradients that reward circulation.
The Wound as System Memory
Distributed cruelty functions as system memory. It encodes what persists, what is rewarded, and what can be done without consequence.
Over time, this memory reshapes behavior. New participants learn through exposure rather than instruction. Norms are not declared: they are inferred from what remains available, replicable, and socially rewarded.
This is how cruelty scales without evangelism.
Implications for Responsibility
When harm is distributed, responsibility fragments. No single actor appears decisive, yet the outcome is stable and reproducible. This diffusion is routinely exploited to deny accountability.
Platforms blame users. Users blame culture. Culture blames technology.
Distributed systems are designed. Incentives are chosen. Defaults are set. When cruelty propagates reliably, it is because the system’s configuration permits and rewards propagation, whether by intent or by negligent governance. Responsibility therefore attaches not only to acts, but to architectures that enable persistence.
Reframing the Problem
Cruelty in distribution is not primarily a moral failure of individuals. It is a failure of containment.
The operative question is no longer “Why do people do this?” but: “Why does the system allow this to spread, replicate, and reward itself?”
Until that question is addressed at the infrastructural level, distributed cruelty remains adaptive, resilient, and insulated from outrage-based responses.
Information Failure Modes in Networked Violence
Elijah Morgan
© 2026 Elijah Morgan
This work is licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).
License: https://creativecommons.org/licenses/by/4.0/
DOI: https://doi.org/10.5281/zenodo.19599231
Version 1.1
Published 2026-04-15