Edited by Sokhna Fall
Cover image. “Judged by data: when algorithms become invisible juries.” Source: AI-generated image (DALL·E, OpenAI), 2026.
In contemporary digital societies, deviance no longer occurs solely within social spaces governed by human norms and institutional rules. Increasingly, it unfolds in algorithmically mediated environments where visibility, reach, and legitimacy are shaped by computational systems embedded in social media platforms. Digital deviance can therefore be understood as behaviour, expression, or content that deviates from the implicit norms regulating online interaction – norms defined not only by communities or legal authorities but also by technological infrastructures that structure digital communication.
Social media platforms have become central arenas of public discourse, identity formation, and political debate. Yet regulation within these spaces is not performed only through formal moderation or legal oversight. Algorithmic systems curate information flows, prioritize forms of engagement, and silently suppress others. In doing so, they participate in the production and enforcement of social norms at an unprecedented scale, influencing what becomes visible, credible, or marginal in the digital public sphere.
While existing research on platform governance and surveillance capitalism has primarily examined the economic and political power of algorithms, this article conceptualizes algorithmic selection as a mechanism of deviance production, through which digital behaviour is continuously classified, amplified, or suppressed. It introduces the concept of hybrid deviance to capture how deviance in digital environments emerges from the interplay of users, platform governance, and algorithmic systems.
In the digital era, every post, share, comment, or video is filtered, amplified, or hidden by algorithms. Satirical videos, scientific critiques, or political memes may spread widely – or vanish from public feeds – not solely because of their content but because of how algorithms interpret engagement, behavioural patterns, and platform rules. Each click becomes a signal; every interaction a datapoint in an automated system of evaluation. The sociological question thus shifts: who defines deviance when visibility itself is governed by software rather than law?
From Classical Deviance to Digital Deviance
The sociology of deviance has long emphasized that deviance is a socially constructed category emerging from processes of interpretation, labelling, and control. Classic interactionist perspectives demonstrated that deviance emerges through social reactions rather than inherent rule-breaking. As Howard S. Becker argued, social groups create deviance by establishing rules and applying labels to individuals or behaviours that violate them.
Erving Goffman highlighted how stigma and identity are negotiated through social interactions, showing that credibility and reputation are central to social classification. Similarly, Stanley Cohen illustrated how deviancy amplification transforms behaviours into social problems through cycles of public attention and institutional reaction. These frameworks underscore that deviance is relational, performative, and mediated by social processes rather than intrinsic to actions themselves.
In digital contexts, these dynamics are increasingly mediated by technological infrastructures. Social media platform structure communication through algorithmic systems that filter, rank, and recommend content. Visibility functions as a form of implicit social classification: content that is amplified gains legitimacy, while content that is downranked or suppressed becomes marginal or “deviant.” In this sense, algorithms act as silent moral entrepreneurs, assigning visibility, credibility, and legitimacy similarly to traditional social institutions, but at scale and without overt human enforcement. Digital deviance therefore emerges from interactions between users, platform governance, and algorithmic evaluation, forming what we term hybrid deviance.
Algorithms as Invisible Authorities
Algorithms increasingly operate as silent regulators of social behaviour. Unlike traditional institutions enforcing norms through visible laws or sanctions, algorithmic systems govern via visibility and attention. A post may reach millions, while another disappears without explanation. In this environment, visibility signals legitimacy; invisibility implies marginality.
Recommendation systems evaluate interactions – likes, shares, watch time, comments – to predict what content will maximize engagement. Over time, these mechanisms establish implicit behavioural norms, subtly encouraging conformity. Algorithmic infrastructures thus participate in shaping social conduct by structuring attention and visibility.
The authority of these systems is largely invisible. Users rarely understand why content goes viral or disappears. Apparent spontaneity masks complex computational models embedded with social assumptions and platform objectives. Embedded with social assumptions and platform objectives, often operating as opaque black boxes that govern information flows without public accountability.
Moderation, Profiling, and Behavioural Identity
Content moderation illustrates algorithms’ role in defining deviance. Automated systems detect hate speech, disinformation, or violent material but often misinterpret nuance, irony, or cultural context. Consequently, legitimate critique or minority perspectives may be downranked, while sensationalist or polarizing content is amplified. Moderation thus functions as algorithmic judgment, structuring which ideas circulate publicly.
Algorithms also construct behavioural identities. User interactions create profiles that guide recommendations and predict engagement, affecting perceived credibility and legitimacy. Politically sensitive or controversial users may see reduced visibility, while mainstream-aligned content is amplified, generating implicit hierarchies that shape discourse.
These mechanisms produce social and behavioural consequences. Users learn which content succeeds, adapting expressions accordingly. Online behaviour becomes simultaneously expressive and strategic, conditioned by algorithmic evaluation, forming a digital panopticon where hybrid deviance is continuously produced.
Private Power, Public Consequences
Despite being privately owned, social media platforms exercise influence comparable to public institutions. Recommendation, moderation, and ranking algorithms regulate information flows for billions of users, with significant public consequences. During health crises, elections, or social debates, algorithmic curation shapes which perspectives dominate public discussion .
Platforms thus do not merely host communication; they actively structure it. By controlling visibility, algorithms participate in constructing social norms and the boundaries of acceptable behaviour, making digital deviance a co-produced, technologically mediated phenomenon.
Understanding digital deviance
In digital societies, deviance is increasingly hybrid, emerging from interactions between users, platforms, and algorithmic systems that govern visibility and attention.
Algorithms act as silent authorities, shaping the circulation of ideas, credibility, and legitimacy. Through moderation, recommendation, and profiling, they engage in classification, normalization, and control – functions traditionally associated with social institutions.
Understanding digital deviance today requires expanding sociological frameworks beyond human actors to include the infrastructures that mediate communication. In this hybrid digital public sphere, deviance is not only observed and judged; it is engineered through code, revealing the profound ways algorithms shape contemporary social life.