ECREA 2021

Beyond the Bubble: A Critical Review of the Evidence for Echo Chambers and Filter Bubbles

Axel Bruns

• 7 Sep. 2021 – paper presented at the European Communication Conference, online

In the context of the resurgence of populist and illiberal political leaders, there is considerable debate about the extent that contemporary online and social media platforms contribute to these developments by placing users in information cocoons – ‘echo chambers’ (Sunstein, 2001) or ‘filter bubbles’ (Pariser, 2011) – that prevent exposure to ideologically diverse news and information. Such concerns have given rise to renewed calls for further action by platform providers to prevent the emergence of such information cocoons, and for additional regulatory interventions that force providers to take such action.

Despite the considerable public attention to these questions, however, actual empirical evidence for the existence of echo chambers remains limited and unreliable. This is partly due to the severely underdetermined nature of these concepts: the authors who introduced these terms, Sunstein and Pariser, never provided any firm definitions for their concepts, and much subsequent research similarly treats them as self-explanatory and interchangeable; for instance, one recent article speaks of “filter bubbles (aka ‘echo chambers’)” (Orellana-Rodriguez & Keane 2018: 78).

Further, the limited analytical scope of many empirical studies means that they might demonstrate considerable ideological homophily in particular communicative practices within specific spaces on selected platforms, but that it is unwarranted to extrapolate from such strictly localised observations to broader communicative patterns. Conversely, larger-scale studies have produced contradictory results that fail to support the severe concerns expressed by advocates of echo chamber and filter bubble hypotheses: Williams et al. (2015), for instance, find evidence for both cross-ideological debate and ideological segregation within Twitter-based debates about climate change, depending on whether they examined the follower, @mention, or retweet networks amongst participants. However, if users involved within these debates are only occasionally segregated along ideological lines, and otherwise interact with each other freely, they cannot possibly be caught within inescapable information cocoons that limit their access to a diversity of views.

To address these shortcomings, this paper proposes new definitions for ‘echo chambers’ and ‘filter bubbles’ that clearly distinguish these concepts from each other and support operationalisation in empirical research. Using these definitions, it critically reviews a selection of recent studies that claim to show evidence for the existence of echo chambers and filter bubbles, finding considerable limitations to their broader applicability. Overall, we reveal popular and scholarly concerns about echo chambers and filter bubbles to have been severely overstated: although phenomena such as ideological homophily and interest-based clustering amongst participants are readily observable in many social media contexts, the complex structure of mainstream platforms and the serendipitous nature of interactions in social media spaces mean that these rarely give rise to genuine information cocoons that prevent users from enjoying a healthy information diet and could therefore generate the severely damaging effects envisaged by Sunstein and Pariser. Drivers for current tendencies towards populism and polarisation must therefore be sought elsewhere: if support for illiberal and extreme political actors increases even in spite of the diverse information encountered by social media users, then the causes for this are more likely to be psychological than technological.

Video of the Presentation

Presentation Slides