Abstract
Axel Bruns’ Are Filter Bubbles Real? questions the increasingly popular application of the terms filter bubble and echo chamber to contemporary debates about digital media. Drawing upon the author’s earlier research into social media and news curation, it contends that the emphasis on technological biases in political communication has occluded the much more important role of human actors in fostering political polarisation. From the outset, Bruns debates the existence of filter bubbles and echo chambers and expertly argues that any narrowing of belief is offset by the capacity of Internet technologies to expose people to a wide range of beliefs, opinions and social groups.
The provocative question in the title aligns with other works in Polity Press’ Digital Futures series, such as Will the Internet Fragment? and Should Robots Replace Teachers? Of course, the answer to such questions can never be a simple yes or no, as they depend on many present and emerging social and technical factors. Despite its brevity and the strength of the polemic, Bruns’ erudite close reading of the literature successfully demonstrates that the application of the terms relies too much on unclear definitions, the seductiveness of the metaphors and speculative reasoning without empirical proof.
Although the book tackles many of the technical aspects of academic writing on filter bubbles and echo chambers, it traces their popularity outside of the academy, which is most clearly indicated by reference to Barack Obama’s farewell speech in 2017 in which the president cautions against technologically facilitated political polarisation and homophily. Bruns also tackles the empirical research and contends that most studies rely on small samples and do not consider exposure to information from other sources. Access to information may narrow on one platform while remaining thoroughly open on other platforms. Due to this platform breadth, even people with very partisan claims remain aware of opposing points of view and, indeed, often critique and cite them.
The book makes some very broad claims in the first half – a necessary feature of a polemic – but develops a very nuanced argument as it progresses. Bruns rejects fragmentation, where individuals are politically isolated by media, and talks about other more common types of marginalisation such as ‘clustering’ and ‘follower networks’ (p. 72) which are mainly driven by human activity rather than by automatic, algorithmic selectivity.
A significant feature of the book is its attempt to provide a rigorous definition of filter bubbles and echo chambers, something that is definitely missing in the popular work of Eli Pariser and Cass Sunstein. Sunstein (2018) glosses over the technical definition of echo chambers, in works such as Republic: Divided Democracy in the Age of Social Media, in a rush to investigate the implications. In Bruns’ technical analysis, the echo chamber is defined by types of connectivity, principally in social media networks, underpinned by confirmation bias and often homophily. In contrast, the filter bubble results from the iterability of communication, in which selective exposure and personalisation algorithms exclude outsider perspectives and confirm insider opinion.
The definitions and accompanying critique are definitely needed in academic debate; however, the definitions are too hard, with echo chambers fully excluding outsiders and filter bubbles excluding any information that does not fit the personalisation profile. Bruns states explicitly that the ‘research we have encountered shows simply no empirical evidence for those information cocoons in their absolute definitions, especially in a complex, multi-platform environment’ (p. 95). Imposing a hard definition on such complex processes and then stating that the actual use does not meet the definition is entirely to be expected. Despite the spatial implications of the metaphors, bubble and chamber, the common use of the terms often only suggest a narrowing of beliefs and behaviours in some communicative activities. Bruns’ definitions remain valuable if they are softened and do not rely on the absolute exclusion of information or groups.
Although Bruns insists on hard definitions, he actually applies some other media concepts quite loosely in the interest of his polemic, arguing that the popular and academic invocation of filter bubbles and echo chambers is tantamount to a moral panic underpinned by technological determinism. Despite Obama’s speech and some disquiet among journalists, this limited concern about algorithmic filtering is not accompanied by sufficient public concern, and therefore does not meet Goode and Ben-Yehuda’s (2009) widely applied definition: ‘if the media obsess about a particular issue or condition which does not generate public concern, then, according to Cohen’s model, we do not have a moral panic on our hands’ (p. 25).
Similarly, the charge that the filter bubble argument’s dependence on algorithmic processes ‘represents a dangerous slide towards a technologically determinist understanding of the contemporary media environment’ without proper consideration of ‘human agency’ (Bruns, p. 24) does not properly represent Pariser’s argument. Pariser’s popular book, The Filter Bubble, lacks technical specificity but he does not claim that these personalisation algorithms fully preclude human agency. He always argues that we use apps and filters for convenience, and that the more we do so, the less we have control (Pariser, 2011, p. 213). The argument is not technologically determinist because users do not necessarily cede all control to search algorithms – we can always scroll past the first five results. The main issue, not really discussed in the book, is the degree to which web users turn to the black boxes of social media and search technologies to manage the current surfeit in information, and in doing so, cede control.
Although these are somewhat minor quibbles in response to an excellent book, they raise questions about the relationship between media concepts and empirical methods, with some concepts suggesting broad communicative tendencies – moral panic, technological determinism, filter bubbles and echo chambers – without necessarily providing clearly testable concepts.
