Canonical Definitions & Usage Constraints
The following four terms are designated as technical terms within Information Failure Modes in Networked Violence. They are not rhetorical metaphors, moral descriptors, or stylistic flourishes. They operate as analytic primitives. Their meanings are fixed as follows.
- Useless Machine
Canonical Definition
A useless machine is a socio-technical system that exhibits continuous procedural activity (movement, signaling, throughput), absorbs inputs (reports, complaints, moderation actions), produces outputs (notifications, removals, dashboards), all without producing proportional change in the underlying harm dynamics it nominally governs.
A useless machine is not broken. It is functionally stable. Its defining property is that it converts concern into motion while preserving the conditions that generate concern.
It describes an equilibrium state, not a failure event.
Usage Rule
Throughout the manuscript, useless machine refers only to systems that address harm procedurally without altering amplification, monetization, or selection dynamics.
It is not a synonym for “ineffective policy”, a critique of individual moderators, a metaphor for bureaucracy in general.
Operational Test
If the system can process infinite reports while harm persists unchanged, it qualifies. - Systemic Risk
Canonical Definition
Systemic risk refers to the foreseeable probability that a system’s normal operation will generate, amplify, or stabilize harm at scale, independent of individual user intent.
In this manuscript, systemic risk is architectural, probabilistic, and persistent. - Selection Effect
Canonical Definition
A selection effect occurs when a system differentially amplifies certain behaviors or content types, not because they are intended, but because they perform optimally against the system’s success criteria.
In this manuscript selection effects are mechanistic. They operate through ranking, recommendation, visibility, and reward. They explain why cruelty persists without needing motive.
Scope Clarification
Selection effects do not rely on claims of algorithmic bias in the moral sense. They do describe emergent outcomes of optimization.
Cruelty is treated as a selected outcome under specific configurations of engagement optimization, low friction replication, and weak sanction.
Usage Rule
Selection effect must always be paired (explicitly or implicitly) with the optimization target (e.g., engagement, retention), the filtered population (users, content types), the observed outcome (escalation, normalization, circulation).
Operational Test
If changing the optimization objective would plausibly change the prevalence of cruelty, the phenomenon is selection-driven. - Duty of Care
Canonical Definition
Duty of care denotes a design-level obligation borne by system operators to anticipate foreseeable harms arising from system operation, implement proportionate mitigation measures upstream of harm propagation, and prevent predictable amplification where risk is known.
In this manuscript, duty of care is structural, and independent of intent.
Scope Clarification
Duty of care is violated when harm pathways are documented, mitigation is technically feasible, and yet architectures remain unchanged.
It is not defined by User agreement, Post-hoc moderation, or “Good-faith” procedural effort.
Usage Rule
Every claim of duty-of-care failure must be traceable to a known risk pathway, a controllable design parameter, or an absence of proportional intervention.
Operational Test
If a platform could reasonably reduce harm by altering amplification or monetization, but does not, the duty is unmet.
Scope Clarification
Systemic risk applies when harm is reproducible across cases, removal of individual actors does not collapse the phenomenon, the system’s design parameters predict recurrence. It explicitly excludes isolated criminal acts, one-off moderation failures, cultural explanations detached from infrastructure.
Usage Rule
Every invocation of systemic risk must be mappable to design features, incentives, or governance gaps, compatible with regulatory categories (e.g., DSA systemic risk, UN due diligence).
Operational Test
If harm persists under enforcement pressure and reappears via substitution, the risk is systemic.
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