Social media, social unfreedom

1 Introduction

Perceptions of social media’s moral status have changed remarkably in the past years. The troubling features of these spaces might have been evident since their inception. But it took a string of democratic shocks in several countries (mainly, the Brexit referendum and the election of Donald Trump) for state and civil society actors to realize more clearly the threats that the platformization of self-expression and interaction pose to social life. Now, a broad consensus seems to exist about the need to tame such a process and the conglomerates profiting from it. Some of the ideas towards this revamp include, inter alia: breaking up the market-dominating firms; preventing algorithmic amplification of falsehoods and hate speech; regulating content-moderation mechanisms; mitigating biases affecting machine learning systems; establishing stricter limits to how data are collected and users’ attention is captured, and; making platforms truly knowable to users and policymakers. These proposals assume that social media can either produce or enable some form of injustice to users. Those concerned with privacy, freedom of expression, and dignity seem to presume that forms of injustice spawned by the platforms are ultimately a legal matter, and so should be addressed as such. Anti-trust scholars seek to restore the importance of structural power inequality in debates around market dominance, while advocates for greater transparency appeal to the idea that users should be able to make – individually and collectively – autonomous decisions on how to use and regulate platforms. While not necessarily in conflict with each other, these views are hardly consensual concerning how to define the different instantiations of injustices that social media are involved with, and usually remain silent or vague about whether and how such instantiations relate to each other.

This essay provides a comprehensive and hopefully coherent conceptualization of what injustice means in the context of social media, exploring its different facets and how they are uniquely linked. It does so through the lenses of recognition theory, leveraging insights from Axel Honneth’s 2014 book, Freedom’s right. In his attempt to update Hegel’s Philosophy of right, Honneth posits that the modern conception of justice involves the institutionalization of three forms of autonomy: legal freedom (embodied in basic subjective rights), moral freedom (exerted through rational self-legislation and determination), and, centrally, social freedom (the mutual realization of our affective, economic, and political aims with others). In Honneth’s account, these forms of autonomy are deeply entangled with one another. Legal and moral freedoms allow individuals to autonomously understand and devise their aims, making possible the social fulfilment of such aims – an achievement which would, in turn, promote subjective rights and self-legislation practices. Pervading all these relations is the idea that we must mutually recognize one another as equally capable subjects, without whom the realization of our own aims would be impossible. These concepts and associations are useful for reassessing social media’s troubles – but in ways that both challenge and enrich Honneth’s theory, as this essay points out.

The core of our argument is that social media’s sociomaterial structures are intentionally designed to deny users’ legal and moral freedoms, and that such denials depend on, and spawn, users’ social interaction. These platforms thus seem to generate a form of institutionalized legal, moral, and social unfreedom (i. e., injustice), whose dynamics mirrors Honneth’s model – but in symmetrically inverted fashion. That is, the violation of users’ rights and self-determination does not stifle but rather enables (and is enabled by) a form of social life underpinned by forced forms of mutuality and the exploitation of recognition experiences, as we will explain.

Social media injustice stems – even if indeterminately – from how social media are materially structured. These spaces can be defined as digital platforms (Gillespie, 2010) in which sociality is mediated by interface functionalities supported by a sprawling infrastructure of code and devices (Plantin, Lagoze, Edwards, and Sandvig, 2016). In this essay, we are interested in corporate platforms that operate on the basis of datafication – the “transformation of social action into online quantified data” enacted by “real-time tracking” that allows for “predictive analysis” by machine learning algorithmic systems (van Dijck, 2014, p. 198). Datafied social media platforms are structured by what we have called elsewhere “algorithmic visibility regime” (Magalhães and Yu, 2022). In this regime, how users are read by platforms (an input of sorts, produced by an ever-increasing set of dataveillance tools) informs semi-automated decisions about how and what users see on the platforms (the ranking of contents on end-user interfaces, often an output of machine learning algorithmic systems). Of course, users do make their own decisions about what to click, block or follow, while platforms constantly make unilateral (and increasingly automated) content moderation decisions. Yet, these tweaks and interventions only generate more data to be processed by this sort of regime. Therefore, even if that entanglement is not perfectly closed, it remains the main control logic of datafied platforms – which have long kept this management of visibilities itself invisible to external actors. Such an algorithmic visibility regime can largely explain the sort of generalized social media injustice we theorize in this essay. If said mechanisms of computational reading materialize the denial of users’ subjective rights (legal freedom) and deliberate opacity of the regime erodes reflexive self-regulation (moral freedom), the exploitation of users’ data will construct (quite literally) certain social relations – the gist of platforms’ social unfreedom.

First and foremost, this essay contributes to critiques of corporate datafication systems. We are not the first to consider how platforms undermine freedom, of course. In addition to an oceanic literature on how digital technologies erode rights, an emerging body of work has richly theorized the ways in which self-determination is undermined by technologies and devices that are minutely designed to maximize the amount of time we spend looking at, clicking on, and thinking about them, “engineering” profitable actions, aims, and subjectivities (Couldry and Mejias, 2019; Williams, 2018, p. 8; Zuboff, 2019). The idea of using Hegelian ethics and recognition theory to make sense of these technologies, although rarer, is not entirely new either. Some have offered empirically rich studies on how these digital spaces mediate intersubjective recognition (e. g., Newlands, 2022). Others have considered whether datafication transforms the struggle for recognition into a competition for “reputation”, fought in a “gamified public sphere” populated by armies of micro “celebrities”, whose autonomy is curtailed by the existential need to be “liked” and fears of being “cancelled” (Davies, 2021; see also Amoore, 2019; Chun, 2021).

We aim to offer another layer of conceptual clarity to these approaches, using Honneth’s theory to name different types of unfreedom as well as systematize the associations between them. Furthermore, to these extant discussions, we add not only an argument centered on the nexus between datafication, recognition, and justice, but also a formulation of the seemingly radical transformation of the social nature of autonomy ushered in by datafication technologies. To say that social media are structured by three self-reinforcing forms of unfreedom helps us to see the immense distance that exists between these social spaces – important to billions of people – and the sort of modern ethical aspirations Honneth describes. As we mention in the conclusion, even the most apparently ‘radical’ reformist proposals seem too timid to truly fix datafied corporate platforms.

Second, we hope to contribute to more general debates around the Honnethian version of recognition theory, which, while somewhat updated in Freedom’s rights to reflect the advance of digital technologies, still heavily neglects the role played by things and artefacts (Deranty, 2006). This essay aims to point to previously unnoticed ways in which emerging computational systems materialize puzzling social and recognitional relationships between individuals and institutions – not the state, as is usually assumed by Honneth, but novel corporations that have emerged as a new dominant type of social actor. Moreover, we demonstrate how social media platforms exploit recognition strategically, without openly denying it. In this way, we dialogue with those who have found fault with Honneth’s critical thought for downplaying the instrumentalization of subjective experiences of – and discourses around – recognition by illegitimate forms of power (Coulthard, 2014; McNay, 2008).

This article evolves as follows. The first section outlines Honneth’s theory of justice, focusing on his definitions of legal and moral freedoms, and explores how these freedoms are structurally denied by social media.^[1] Then, we describe the associations between such lack of autonomy and a new form of social unfreedom. In the conclusion, we briefly consider what building a just platform would demand, from a recognition theory view of justice.

2 The legal and moral unfreedoms of social media platforms

The ethical nature of recognition has long been central to Honneth’s thought. His first major book, The struggle for recognition (1992), advanced a sort of developmental view of the construction of autonomous subjects, which depended on them developing self-confidence through love, self-respect through rights, and self-esteem through social solidarity. These elements are still present in Freedom’s rights but in quite different ways. The most important novelty arguably regards Honneth’s turn to the question of justice, which has come to be understood, after modernity, as necessarily about freedom, as he argues. It follows that a just society is the one “which protects, fosters, or realizes the autonomy of all members of society” through social norms and practices – through “institutions”, in Honneth’s lexicon (Honneth, 2014, p. 37).

Although the claim that justice involves designing institutions that can provide universal access to individual freedom is hardly new, what sets Honneth’s (and Hegel’s) account apart is the way in which they conceptualize freedom in the first place. Honneth offers substantial critiques of negative and reflexive views of freedom without rejecting their role in his own expansive account of social freedom – which is realized in three different interactional spheres: personal relationships, market, and the public political sphere. Injustice, however, entails the “unnecessary exclusion from or restriction on opportunities to participate in social processes of cooperation” (Honneth, 2014, p. 161). Yet, blatant injustice does not exhaust the ways in which unfreedom is produced, for certain “pathologies” may prevent individuals from “adequately grasping the significance” and “purpose” of cooperative practices and norms (p. 161–162). Pathological social practices, in this context, are those which stem from, and perpetuate, a distorted or myopic view of freedom.

This section first explains Honneth’s definition of negative/legal freedom before demonstrating how social media platforms structurally deny these freedoms. Then, a similar argument is developed in relation to moral freedom. The section ends with a consideration of recognition and the sort of subject implicit in these unfreedoms.

3 Social media, social unfreedom

So far, the essay has explored the notions of moral and legal freedoms and explained the ways in which social media platforms’ materiality is premised on the need to structurally deny these kinds of autonomy. Legal and moral unfreedoms are not impervious to social life, of course. In fact, they can only be understood as social phenomena. However, in Honneth’s narrative, these autonomies inevitably hinge on the need to guarantee individuals the capacity to ultimately reject external forces – either interventions into one’s private space, property, and expression, or normative demands that cannot withstand rational scrutiny. When legal and moral freedoms are respected and supported, individuals are expected to develop autonomous aims. Which exact aims these are is irrelevant: if those freedoms are properly institutionalized, aims will be based on mutual forms of recognition.

Yet, merely understanding and developing authentic and justifiable aims is necessary but insufficient, since proper aims must be realized through social practices, as Honneth posits, and such realization is simply impossible without others. Social freedom is a way of conceptualizing this interactional attainment of one’s reflexively constructed aims. Accomplishing one’s aims through interactions with others is, of course, a way of having one’s subjectivity recognized. However, this interaction can only be understood as the realization of individual freedom when it does not deny others’ aims – when recognizers are themselves understood as realizing their autonomously developed desires. This ontological complementarity entails much more than tolerating or valuing others’ individualities: It involves seeing “the aims of the other as the condition for the realization” of one’s own aims (Honneth, 2014, p. 89). We can only be free with others. Importantly, this sort of mutual recognition is not a chance event or an incidental occurrence; nor is it the necessary consequence of human interaction. In fact, it emerges from, unfolds in, and must be supported by interactional spheres of social norms and practices. In these “institutions of recognition”, the realization of one’s aims is not only tolerated but actively desired by others. These institutions are constitutive of freedom since they both teach individuals which “classes of behavioral expressions” can involve them in said complementary realization of aims and enable them to experience their freedom as a social phenomenon (Honneth, 2014, p. 96). Without such institutions, one’s aims would be impossible to achieve, regardless of how reflexively formulated these aims had been. In Zurn’s (2015, p. 160) useful example, the key reason why it was incredibly difficult for any woman living in the early eighteenth-century Europe to pursue a professional life, or even want to do so, was that the role of “professional woman” was not an “intelligible” category.

In Honneth’s definition, justice depends on granting universal access to three institutions – or “spheres” – of recognition. The first one, the sphere of personal relationships, includes three kinds of human associations that, in their modern form, presuppose mutual recognition processes: friendship, romantic relations, and family. Each one of them is wholly founded in the fact that one can only exert one’s desire to love others when others also are willing to love them, a mutuality that had to be made possible by modern social norms. The second sphere is that of the market. Here, Honneth engages with the difficult assertion that capitalism is founded on the promise of enhancing production to meet people’s material needs and enabling social recognition via labor. He does not argue that today capitalism delivers these promises but that our attempts to humanize it reveal the general assumption that capitalism is founded on an expectation of mutual autonomy. Lastly, there is the public sphere, in which a democratic will is formed and collective self-governance unfolds. Modern democracy hinges on a profound cooperation, he says: If a community decides to govern itself, its members can only do so with one another, even if this agreement involves complex representative arrangements. As these three spheres are based on mutual recognition, they cannot develop “pathologies” but “misdevelopments”, “anomalies whose sources must be sought elsewhere, not in the constitutive rules of the respective system of action” (Honneth, 2014, p. 242).

Social media platforms can, to different extents, encompass relations that in principle appear typical of these three spheres. Platforms have long been understood as spaces in which we primarily engage with friends, lovers, and family, or make new friends, meet previously unknown relatives, and fall in love with strangers. Overtime, many of these platforms have also enabled commercial (such as Facebook Market, for instance) and labor relations (think of the industry created around the production and management of social media content and pages). Also, of course, a large portion of the informational and communicative processes on which democratic will formation depends have now migrated to spaces provided by the platforms. At the same time, some might argue that Honneth’s theory of institutions of recognition cannot simply be applied to social media platforms, since, in his historical reconstruction of such institutions, they in principle comprise symbolic norms and non-material practices rather than a particular organization. As such, the possibility that social media can give rise to their own forms of social freedom might appear absurd. Nevertheless, arguably social media are not just the central social space for billions of people, but their computational processes and design are also involved in the creation of new social interactions today. As argued above, to be a “friend”, “worker”, and “citizen” on platforms depends on their structures, which give rise to distinctive sets of norms and practices partially configured by the platforms.

It is thus clear that platforms enable specific – and extraordinarily relevant – social roles and forms of interactions. Yet the mere existence of such sociality is obviously not enough to configure social freedom. For if individuals experience these positions and relations as “imposed”, if they cannot feel “the mutual supplementation of their actions as an ‘objective’ realization of their own freedom that is desired and strived for”, they are not participating in institutions of recognition (Honneth, 2014, p. 237). And it seems clear that users do not reflexively accept the social roles imposed on them by the platforms. Honneth’s theory has taken us thus far. However, the reality of the associations between social life and freedom can hardly be captured by his model alone.

Indeed, social media appear to engender the very antithesis of social freedom, whereby the relationship between legal, moral, and social freedoms is operationalized in a symmetric but inverted manner to what was described in Honneth’s work. Linking up the points made in the previous sections, it is possible to argue that the imposed violation and distortion of users’ subjective rights, in conjunction with the violation and distortion of users’ ability to understand, reflect, and reject such imposition, informs a sort of “engineered sociality” (van Dijck, 2013), on which platforms rely to generate profits. In the previous section, we discussed how platforms aim to “insinuate themselves within ‘the self’s needs, desires, and other choices’” (Couldry and Mejias, 2019, p. 171). Closely related to this is the attempt to manufacture not only aims but also actions – what Zuboff (2019) names “behaviour modification”. But if her focus remains on one-off actions, much more complex forms of engineered practices are exchanges and relationships between users. Arguably, social media’s main innovation, as a social space, concerns not precisely their ability to automatically curate content, but their data- and probabilistic-driven use of content to trigger interface-level interactions through newsfeed appearances, suggestions of ‘friends’ and accounts, visual organization of chat messages, etc. These interactions would be unthinkable without dataveillance and the disregard for ordinary self-determination explained before. Therefore, instead of fostering legal and moral respect to help constitute individuals who can formulate their aims autonomously and then providing them with the social means to realize such aims (as in Honneth’s model), platforms deny users’ legal and moral respect with the goal of imposing their own corporate aims – aims that can only be realized socially, according to the material limits and processes controlled by platforms themselves.

A different facet of this particularly forced sociality concerns the fact that users relate to each other not only through interface actions but also through mass data associations. The sorts of high-powered, global machine learning systems underpinning mainstream social media are trained with the data from a large number of users. That is, automated decisions that affect one individual are necessarily dependent on the data collected from and about other individuals, someone who that first user may never have engaged with on the interface level. This represents a rather different definition of mutuality and dependence, one that is realized through computational means and devoid of any assessment by the actors who are forced into cooperating with one another. Who understands or is even aware of how exactly their own actions (say, flagging a video of a ‘follower’ as ‘hate speech’) can help create an action that impacts on other, potentially unrelated users (say, someone whose video is months later taken down by AI content moderation algorithms as ‘hate speech’)? Most likely, not even the platforms’ designers and programmers could explain the exact ways in which these systems emerge (Burrell, 2016). Such arbitrary data cooperation further illustrates the idea that social media appear as a mirror image of Honneth’s model: users seem to (unwittingly) construct the structures of their unfreedom together.

In saying that social media’s material structure enables a form of social life that is inherently unfree, we run the risk of oversimplifying how individuals’ experience such sociality. The Gordian knot of most theorizations of datafication’s moral troubles is the enormous gap between the utterly negative image painted by critics and the much more ambiguous impressions of being a user of platforms. There are several reasons for this gap, such as the opaque nature of social media, the users’ feeling of being disempowered by firms’ unaccountable power, the indirect and often abstract ways in which said harms to autonomy surface, for instance.

But an important and seldom discussed reason lies precisely within intersubjective recognition. When platforms enable social interaction, they also enable multiple channels whereby users can have their worth, adequacy, and usefulness discovered, acknowledged, and approved. These recognitional actions (positive reactions such as ‘likes’, textual compliments, retweets, live communication) might appear to be small and insignificant. But the more central platforms become to social life, the more important these actions become to users.

Even when recognition is informed by the type of unfreedoms theorized above, it might still be experienced as genuine respect and engender forms of proper cooperation able to foster real mutual recognition. There are innumerable examples, from everyday acts of gentleness to democratic changes brought along, which have been realized and organized on and through datafied social media. This is hardly mysterious. Individuals, their aims and behaviors are not solely informed or defined by their abusive relationship with platforms. We remain associated with other state, media, educational, and civic institutions, with which we might develop ethical relationships and that are only tangentially influenced by corporate datafication. The perhaps more puzzling conclusion is that social media do not explicitly aim to stop individuals from feeling recognized. The opposite is true. Actions of genuine recognition that occur on social media are all datafied and, as such, become involved in, and processed by, the same expressions of unfreedom this essay has described. Indeed, their massive affective importance to users likely makes them especially powerful datapoints in how platforms try to entice us never to leave their domains. It is the constant expectation of being somehow heard, that mainly makes platforms so compelling to us. In this sense, the perception of recognition becomes not (only) a tool of emancipation but a piece of platforms’ vast injustice machine. This stands in contradiction to Honneth’s body of work, which asserts the moral inviolability of recognition. But it resonates with older critiques of the political uses of recognition (Coulthard, 2014). The intractable ambiguity and indeterminacy of this recognition does not transform what would have been a primordial moral event into a commodity (something that is hardly new), but into a form of capital, that can be used to generate (profitable) products and interactions. Social unfreedom then both enables and is enabled by legal and moral freedoms.

Finally, what does it mean to say that social media’s unfreedom is ultimately instantiated not by discrete violations of users’ rights and aims, but by social relations that, while being dependent on the systemic denial and manipulation of those rights and aims, are often felt as rewarding? This formulation helps us understand how social media injustice is normalized and affectively justified. When social life comes to embody unfreedom, the nature of disrespect might become extraordinarily difficult to ascertain and confront. Analytically, this conclusion also suggests that the sort of unfreedom discussed in this essay is not a mere deviation from modern freedom, a moral transgression that can be easily understood and tackled as such. It is not that platforms simply disrespect our privacy, prevent us from refusing their norms, and reject our attempts to realize our aims together. They have reorganized the very meaning of these forms of freedom, and of their connections: a much more fundamental transformation of moral life seems to be afoot. Whether datafied platforms engender a departure from Enlightenment’s assumptions of injustice is a hypothesis that would demand much more elaboration. It seems obvious, however, that social unfreedom does not correspond to a new normative model of an ethical sphere. In the conclusion, we briefly consider the question of what it would take for a platform to be just, from a recognition theory perspective.

4 Conclusion

This article has explored how Axel Honneth’s theory of justice can help us understand the ways in which social media platforms, and their algorithmic forms of visibility control, can produce new forms of injustice. We explored Honneth’s theory and demonstrated how it can help clarify the denial of subjective rights (legal freedom) and self-determination (moral freedom), and illuminated the ways in which these unfreedoms make possible (and are also enabled by) social relations that might be said to configure a form of social unfreedom – even when such unfreedom is experienced as intersubjective recognition.

At this point, we want to briefly turn to the possibility of how social media can foster social freedom. What we offer here is not a blueprint but a speculative vision that condenses, in a simplified form, some normative direction and questions. The crux of Honneth’s reworking of Hegel’s Philosophy of right is the apparently simple yet far-reaching conclusion that respecting and fostering legal and moral freedoms is necessary but still insufficient for realizing freedom and producing justice. In the context of this article, this amounts to saying that even if platforms truly changed their visibility regime to abolish acts of surveillance, managed to develop a scalable system to counter mis-/disinformation, and made their operation fully transparent for users, policymakers, and researchers – measures that would be considered radical today – they would still be not just. At best, they would be spaces for users to better understand and formulate their aims, learn how to respect others’ rights and moral capabilities. To be truly conducive to justice, platforms would have to live up to their promise of being an authentic community-building space of interaction.

Given space restraints, it would be impossible to provide here a detailed suggestion of how platforms ought to be designed to generate this ethical outcome, or at least embody the Honnethian idea of justice in their particular design decisions. Instead, we propose to focus on an aspect that is closer to the third, political sphere of social freedom, because, while the three spheres constitute a “complicated web of reciprocal dependencies” (Honneth, 2014, p. 587), the political sphere possesses a clearer governing role over the other two. We propose that the first step to turn platforms into institutions of recognition would involve inviting users to decide, through social media, what social media platforms ought to be and how these spaces ought to produce and govern visibilities. It is important to differentiate our proposal from, say, the idea that elected politicians should use the force of governments to turn current corporate platforms into institutions of recognition. For what matters is not simply the outcome but also the decision to use the potentialities social media’s structure has allowed (e. g., giving a voice to millions of individuals simultaneously, conveying information to both multitudes and specific groups), to enact mass deliberative processes, in which individuals’ desires to govern and create their own social media can only be realized collectively. Through these processes, individuals, as key stakeholders, would be able to decide on how exactly social media’s functionalities should be designed to structure spheres of social freedom.

There are surely innumerable difficulties with this path. Yet the perhaps most important one is that no current corporate social media are designed to enable this sort of radical democratic experiment. This echoes a doubt that hovers uncomfortably over Honneth’s theorization: Social freedom depends on mutual recognition norms and practices – but how do such norms and practices emerge from a world in which they do not exist? Honneth does not acknowledge this problem explicitly, but his normative reconstruction suggests that what once seemed to be small, non-important events – say, isolated workers striking – had, over time, enormous consequences – say, the institution of labor rights. In the case of social media platforms, some of the most promising ideas are being entertained by scholars and activists who directly confront the ownership and governance of these spaces – consider, for instance, platform cooperatives in Brazil (Observatório do Cooperativismo de Plataforma, 2021) and public service platforms in Europe (Fuchs and Unterberger, 2021). This is not to say that any of these models necessarily embody social freedom. Yet, there are alternatives we can follow for the realization of social freedom.