Bluesky, Echo Chambers, and the Threat of Social Contamination Online

This is an extract from chapter 6 of our new book, Beyond Woke and Anti-Woke: Explaining the Rise of Social Justice Ideology, which comes out with Bristol University Press on 17 February 2026. If you don’t already subscribe to Thomas Prosser’s work here on Substack, he can be found at The Path Not Taken.

How might we understand the relationship between social media and ideological change? There are two broad categories of argument. On one hand, the ‘context collapse’ thesis sees digital platforms as dissolving the individual through the production of new forms of collectivity and synchronization, effectively absorbing users into its logics. Such platforms are not simply information hubs or tools for collaboration and networking. They also, as the media scholars José van Dijk, Thomas Poell and Martijn de Waal point out, ‘shape the way we live and how society is organised’.¹ As such, they play a central role in the structuring of social reality. The algorithms that sort search results, arrange the order in which user posts appear on platforms like Facebook and X, label content and display AI summary outputs have a way of imposing their logic on user perception. Via a process of ‘algorithmic thinking’, the account of the world they mirror back – and the descriptions they impose on it – becomes a kind of overlay that users end up assimilating and using to model their own behaviour and thought processes, both online and (potentially) offline.²

Van Dijk and co-authors suggest that platforms act as a kind of ‘recalibration laboratory’ in which new ‘new features are constantly tested on users’. To illustrate this, they provide a highly relevant example: the gender identity options provided by Facebook in the mid-2010s. Not only did these enable the company to extract data from their users (which could subsequently be monetised for advertising), but they also ‘actively influenced social norms’ by encouraging users to think of their gender identities in ways that moved beyond the binary.³

On the other hand, social media are seen to produce a new order of hyper-individualism in which all users are given the tools to fashion their own separate online worlds. This is the ‘filter bubble’ or ‘echo chamber’ thesis. According to the ‘bubble’ and the ‘echo chamber’ metaphors, there is an outer shell or membrane that is impassable to certain forms of outside information. Each user bubble is impervious to opposing viewpoints and social realities that might disrupt the user’s sense of their own coherence.

The legal scholar and commentator Cass Sunstein was the first to apply the ‘echo chamber’ metaphor to online behaviour, using it to describe ‘people’s growing power to filter what they see’ combined with the ‘growing power’ of online content providers to in turn ‘filter’ that view for each user, based on personal data.⁴ In 2011, the internet entrepreneur Eli Pariser coined the term ‘filter bubble’. If Sunstein’s critique gestured towards the causes while in practice paying more attention to the supposed effects of information filtering, Pariser’s account sought more grounding in the technical infrastructure underlying bubble and filter creation. The object of Pariser’s critique was a series of then-recent changes to how digital platforms were displaying content. According to Pariser, what ‘the new generation of Internet filters’ then being rolled out at Google, Facebook, and YouTube were doing was tracking ‘things you like – the actual things you’ve done, or things people like you like’ and attempting to match those past preferences dynamically and in real time when displaying new content.

Driven by the commercial imperative of targeting their users with personalised advertising, these companies were, he wrote, effectively starting to act as:

prediction engines, constantly creating and refining a theory of who you are and what you’ll do and want next. Together, these engines create a unique universe of information for each of us—what I’ve come to call a filter bubble—which fundamentally alters the way we encounter ideas and information.⁵

For Pariser, the ‘algorithms that orchestrate our ads are starting to orchestrate our lives’. The result of these processes was the creation of the uncanny ‘digital doubles’ – algorithmically constructed out of our previous online activity – that we encounter whenever we resume our browsing. While we might feel some repulsion at being mirrored back by our screens in such simplified terms, we are still tempted to plunge back narcissistically into old habits. ‘A self built on clicks’, Pariser wrote, ‘will tend to draw us even more toward the items we’re predisposed to look at already—toward our most Pavlovian selves’.⁶

According to proponents of the model, the denizens of filter bubbles and echo chambers are subject to an ongoing, real-time process of radicalization.⁷ Deprived of access to opposing views – or the ‘fact-checking’ voices that might challenge their political convictions – those within echo chambers adopt increasingly extreme versions of their own political ideologies through relentless exposure to what they already believe.⁸

Following Elon Musk’s purchase of (then) Twitter in October 2022, there have been periodic migrations of progressive users to other social media platforms; such phenomena illustrate the logic of the echo chamber thesis.⁹ Initially, Threads and Mastodon were popular destinations, but after the 2024 US presidential election (and in recognition of Musk’s activities on behalf of the Trump campaign) there was a coordinated effort among progressive X users to transfer their posting activities to the Twitter clone Bluesky.¹⁰ Bluesky is often described by its detractors as an ‘echo chamber’ that is especially prone to ‘censorious and doctrinaire’ social justice discourse.¹¹ Part of the reason for the ideological makeup of its current user base is that many early account holders favoured social justice ideology and were able to use Bluesky’s invitation-only membership system to populate the platform with likeminded associates. However, another key aspect of its appeal lies in its comprehensive content moderation and labelling systems.

Replicating norms developed on Twitter in the 2010s, Bluesky gives its account holders the ability to create, share and subscribe to block lists to limit their exposure to other users. These can be very effectively scaled via coordinated action to throttle the visibility of accounts deemed ‘harmful’ by other users and limit their ability to form influential follower circles. In 2024, Bluesky also introduced an open-source content labelling system, extending the use of content warning-style labels beyond the site’s official moderators. This system allows users to subscribe (on an opt-in basis) to a series of third-party labelling services, which have the power to label site content – both posts and other users’ profiles and accounts – in order to ‘warn’ other account holders or enable them to filter it from view entirely.

The labels created by the ten most popular third-party labellers on Bluesky provide an insight into the core concepts operating within the platform’s ideological universe. These include antiracism (‘dehumanizing-rhetoric’ and ‘misogynoir’) and trans rights (‘misgendering’, ‘enbyphobia’, and ‘terf-gc’, the latter of which is defined as being applied to ‘accounts who self describe as being a “Trans-Exclusionary Radical Feminist” (TERF), “Gender Critical” (GC), “Gender Realist” … or accounts with multiple posts which attack trans people or use pejorative terms such as TRA’).¹² In addition to their stated rationale of harm prevention, many of these labels are also oriented towards what the political theorist Graham Harrison calls the ‘combat’ dimension of political ideology.¹³ Bluesky’s moderation principles of ‘hiding’ content, ‘warning’ users and ‘categorising’ networks become ways of justifying the pejorative labelling of accounts or content related to opposing political tribes. The label ‘jesse-singal-follower’, for instance, ‘places a badge on the profile of accounts that follow Jesse Singal, for informational purposes’. A user can only have the ‘badge’ removed by unfollowing Singal’s account – the unstated, though implied, aim of the label itself, via the enforcement mechanism of public shaming.¹⁴

Introducing its new moderation policy, Bluesky celebrated the fact that it was democratising the labelling of problematic users and content – something that happened ‘under the hood’ under orthodox site moderation regimes but could now be brought into the open in the interests of ‘transparency’.¹⁵ However, each of these labels (and their accompanying definitions) also have the effect of rolling out an ideological overlay across every aspect of the site, one that is heavily social justice-coded, given the nature of its user base and most active and committed labellers. The labels pose as a more or less objective information management system grounded in an ethos of safety and the protection of marginalised groups. However, their main function seems to be to that of ideological decontestation. As the philosopher Holly Lawford-Smith puts it, there is in these and similar policies an attempt to ‘settle arguments in advance’ through the implementation of linguistic compliance frameworks that pose as the rules of social reality itself.¹⁶ They solicit recognition for, agreement with and the enforcement of a particular worldview and its associated (and evolving) understandings of what constitutes harm and offence.

Using Bluesky as an example, we could argue that the filtering effects of particular platforms should be regarded more as ideological artefacts and less as inadvertent engineering problems, i.e. something that has somehow ‘gone wrong with’ the way information travels online.¹⁷ Self-sorting into bubbles and filters is a deliberate act and something that can take place across the political spectrum. Moderation policies and practices enable the cocreation of a consensus reality or set of evolving etiquettes, underpinned by the tools of social classification, description, redescription and labelling. While filtering of the sort offered by Bluesky’s third-party labellers attempts to enforce ideological decontestation and consensus, in practice this has proven difficult (if not impossible) for platforms to bring about. Those who find themselves out of alignment with the normative worldviews imposed by ideologically aligned moderation policies are not simply driven off platforms; very often, they are affectively polarised into more intensified forms of ideological opposition. The possible impact of Twitter’s ‘hateful conduct’ policies – which saw many gender-critical Twitter accounts banned in the mid to late 2010s – on the subsequent galvanisation of energies in the gender-critical movement is one example of this.¹⁸

Moreover, it is not clear that Bluesky’s attempts to create a safe ideological space for its users through moderation is proving effective in enabling the site to grow. After enthusiastically migrating to the platform in November 2024, the progressive billionaire Mark Cuban subsequently cooled on Bluesky, commenting on its echo chamber-like qualities:

The lack of diversity of thought here is really hurting usage …The moderation and block tools on here are so advanced, if you see someone you don’t want to see on here, just block them. Don’t attack them. There used to be great give and take discussions on politics and news. Not so much any more.¹⁹

---

New rules and codes have a flipside: rule violation. Social media encourages what Christian Parenti calls a ‘concern with moral and political hygiene that constantly draws a distinction between the politically clean and unclean, friend and enemy, good and evil’. As a result of this underlying fear of moral contamination, ‘certain people, places, books, ideas, and consumer products must be avoided’.²⁰ During the social media era, this logic has had a key influence on the development of social justice ideology.

In Lexi Freiman’s 2023 novel The Book of Ayn, New York City-based novelist Anna produces a satire of the American opioid epidemic and is promptly cancelled for ‘classism’. After she and her book are singled out for castigation in The New York Times, Anna is vilified on Twitter, dropped by her publisher, shunned by her former friends and blacklisted by potential employers. Looking for a way to make ends meet, she finds that even her former high school refuses to consider her application to teach there: ‘all the prep schools were now, at least rhetorically, anti-capitalist’.²¹ Blindsided by the uniformly negative reception of her book, Anna realises that, in her refusal to ‘scroll’ and ‘binge’, she has found herself out of step with mass feeling in the digitally mediated, post-#MeToo and #BLM moment.

Nevertheless, she professes herself baffled at the totality of her cancellation; ‘There were so many new rules—all set by college students paying two hundred thousand dollars for their humanism’.²² Was it now the duty of a novelist to trace all character flaws to primal scenes of childhood trauma? Contemporary satire, she decides, has been reduced to just-so stories about power and privilege. The new rule that satire had to ‘punch up’ in practice meant imposing a simplistic and childish moral framework on art: ‘there had to be a good side, and a bad one’.²³ Surely fiction-writing should be more complex than dividing characters into sympathetic and unsympathetic categories, so that readers could identify with or despise them accordingly?

The Book of Ayn has largely been read by reviewers as a straightforward satire of contemporary cancel culture. It is also, from its very first sentence, a confrontationally scatological one; ‘I could trace my contrarianism back to age three, when I’d painted the walls of my parents’ bedroom with my own poop’. However, the novel’s relentless, pseudo-satiric scenes are not there solely for shock or gross-out value. Freiman uses them to unpack the surreal and Kafkaesque politics of social hygiene that underpins cancel culture. Interpreted more broadly, they show how contamination phobias maintain the boundaries between the ‘politically clean and unclean’ and how morally coded friend/enemy distinctions are promoted by those same logics of coding and standardization that structure social media.

The practice of list making is a concrete example of this and reflects the basic structure of social media, i.e. its replicability via cut and paste and public visibility.²⁴ Readers make public spreadsheets of ‘problematic authors’ and share them on social media or sites like Goodreads. The perceived moral transgressions of authors who appear on these lists can derive equally from ‘harmful incidents’ recorded in their biographies, the supposedly ‘non-inclusive’ or ‘problematic’ nature of their storytelling (in, for instance, writing about ‘sensitive topics’ such as rape or racism in ways that appear ‘off’ or inadequate to the list-maker), or the statements they make on social media. As with boycotts and petitions, there is a certain kind of ‘moral immediacy’ at work with these acts of online critical list-making. They seek to short-circuit more formal kinds of justice, making the mere existence of a record (or a petition or boycott) into its own form of sanction.

Whilst such processes are rooted in social media and social justice ideology, they also reflect the logic of quasi-professional digital communities such as academic Twitter/X, where the online reputation of named profile holders is underpinned by their offline professional reputations and there can be severe consequences (loss of esteem or even employment) for those seen to make discourse mistakes online. Social media profiles are, as Moeller and D’Ambrosio write, typically curated to appeal to the expectations of those unknown second-order observers whom they term ‘the general peer’. To maintain their reputations in these radically public online environments, successful profile holders need to conform to ‘what is seen to be seen as good’, anticipating the ‘ever-shifting moral trends’ taking place in their communities and shaping their behaviour and language to fit.²⁵

Reputation crises are not confined to individuals. Because of the densely interlinked nature of online reputation, a reputational crisis for an academic employed by a particular university can become a second-order crisis for anyone in their network; for their university, their departmental colleagues and their students, all of whom may feel morally implicated by association. The hierarchically mixed nature of these online communities, which may include postgraduate students and/or junior employees eager to make their mark and ‘take down’ established figures who are ‘out of sync’ with contemporary moral sentiments, may also contribute to the intensity of social media reputation spirals.

Viewed in this light, the obsessions with purity and moral hygiene that underpin cancel culture make more sense. The need to be ‘seen to be seen as good’ so as to maintain ‘good standing’ in morally competitive, professionally oriented online communities, the real-time nature of contemporary ‘moral trends’ and their associated, continually updating jargons and the basic conditions of visibility and replicability that hold true for social media content combine to make the statements and connections we make online matter in ways seemingly out of all proportion to the often momentary and offhand nature of their creation.²⁶ Indeed, the more seemingly minor a statement or connection is, the more significant it can be made to appear, as though its casual nature reveals an insight in itself into the user’s moral beliefs at their most unguarded (and therefore ‘true’) level.

1. Van Dijck, J., Poell, T., and de Waal, M. (2018) The Platform Society: Public Values in a Connective World. Oxford: Oxford University Press, 9.

2. Beer, D. (2024) The Tensions of Algorithmic Thinking: Automation, Intelligence and the Politics of Knowing. Bristol: Bristol University Press.

3. Van Dijck, Poell, and de Waal, The Platform Society, 11.

4. Sunstein, C. R. (2018) #Republic: Divided Democracy in the Age of Social Media. Princeton, NJ: Princeton University Press, 6.

5. Pariser, E. (2011) The Filter Bubble: What the Internet Is Hiding from You. London: Penguin, 6.

6. Pariser, The Filter Bubble, 114.

7. O’Hara, K., and Stevens, D. (2015) ‘Echo Chambers and Online Radicalism: Assessing the Internet’s Complicity in Violent Extremism’, Policy and Internet, 7, 404.

8. Kitchens, B., Johnson, S. L., and Gray, P. (2020) ‘Understanding Echo Chambers and Filter Bubbles: The Impact of Social Media on Diversification and Partisan Shifts in News Consumption’, MIS Quarterly, 44(4), 1619–20; Han, B. C. (2018) The Expulsion of the Other: Society, Perception and Communication Today. Cambridge: Polity Press, 3.

9. Radivojevic, K., Adams, D.J., Griffin, L., Felixander, K., and Weninger, T. (2025) ‘User migration in the Twitter diaspora’, EPJ Data Science, 14(36).

10. Lake, S. (2025) ‘Bluesky Is Backfiring’, Fortune, 12 June.

11. McArdle, M. (2025) ‘The Bluesky bubble hurts liberals and their causes’, Washington Post, 8 June.

12. Labels and descriptions listed by the @blacksky.app; @asukafield.xyz; and @skywatch.blue third-party labelling accounts (as of 16 June 2025, the sixth, ninth, and fourth most heavily subscribed labelling accounts on Bluesky). A full list of such accounts and their labels can be found at ‘BlueSky Labellers’.

13. Harrison, G. (2024) The Internet Left: Ideology in the Age of Social Media. Bristol: Bristol University Press, 62.

14. For this label description, see Jesse Singal Watch (@singal.shame.wtf).

15. ‘Bluesky’s Moderation Architecture’ (2024), 15 March.

16. Harrison, The Internet Left, 63; Lawford-Smith, H. (2022) Gender-Critical Feminism. Oxford: Oxford University Press, 139.

17. Nguyen, C. T. (2020) ‘Echo Chambers and Epistemic Bubbles’, Episteme, 17(2): 141.

18. Lawford-Smith, Gender-Critical Feminism, 122–3.

19. Lake, ‘Bluesky Is Backfiring’.

20. Parenti, C. (2024) ‘The cargo cult of woke’, Catalyst, 8(1): 121.

21. Freiman, L. (2023) The Book of Ayn. New York: Catapult, 6.

22. Freiman, Book of Ayn, 3–4.

23. Freiman, Book of Ayn, 62.

24. On social media affordances, see boyd, d. (2010) ‘Social network sites as networked publics: Affordances, dynamics, and implications’, in Z. Papacharissi (ed) A Networked Self: Community and Culture on Social Network Sites. London: Routledge, 46.

25. Moeller, H-G. and D’Ambrosio, P.J. (2021) You and Your Profile: Identity After Authenticity. New York: Columbia University Press, 28.

26. Brubaker, R. (2022) Hyperconnectivity and its Discontents. Cambridge: Polity Press, 62–3.