Search Engines in Polarized Media Environment: Auditing Political Information Curation on Google and Bing Prior to 2024 US Elections

Concerns about (affective) polarization in liberal democracies have been particularly acute in recent years, marked by the rise of populism worldwide. While the exact scope and measurements of polarisation remain debated by scholars, there is ample evidence that politics is being increasingly perceived as a highly polarising subject. Such a perception is, to a large extent, shaped by the evolving political and media environments that amplify partisanship and contribute to “a self-perpetuating cycle” (Wilson et al., 2020: 223) of polarization. In our paper, we discuss how new types of algorithmic gatekeepers, in particular search engines (SEs), can affect (perceived) polarization in media environments, using as a case study the 2024 presidential elections in the US. Our interest in SEs is attributed to them increasingly shaping media environments worldwide due to high user trust and intense use for different forms of (political) information-seeking behavior. By curating information sources in response to user queries, SEs determine what information individuals are exposed to on various political topics, including elections. Unlike more traditional gatekeepers, SEs’ curation treats users in a differentiated manner to better adapt its outputs to users’ needs (e.g., based on their language or location). Not only does it often lead to unequal information exposure (Makhortykh et al., 2024a), but it also can amplify the users’ perception of a media environment as polarized, thus contributing to societal polarization. The relationship between SEs and polarization is of particular importance in the context of elections. Earlier studies (Epstein & Robertson, 2015) have indicated that SEs’ curation can affect voter preferences by shifting users’ voting preferences. Under these circumstances, skewed exposure of certain groups of voters to information sources with a particular ideological slant can benefit certain candidates as well as increase (perceived) polarization among voters. However, whether differentiated curation on SEs leads to such outcomes and how it is affected by different curation factors (e.g., user location of changes in source relevance) remain unclear with existing studies (e.g., Unkel & Haim, 2021; Trielli & Diakopoulous, 2022) producing contradictory observations. To address these gaps, we examine how the two largest Western search engines, Google and Bing, curated political information prior to the 2024 US elections. Using virtual agent-based algorithm audits, we examine how the presence of specific types of political information sources is affected by user- (the use of slanted vs. non-slanted queries) and system-side factors (user location and time-based changes). Specifically, we simulate the activity of multiple agents (N=120) searching on a monthly and daily basis for election-related information from a selection of Democrat, Republican, and battleground US states. We find that search engines tend to prioritize left-leaning media sources in relation to all types of election-related queries. Secondly, we find that the ideological slant of search engine outputs aligns with the ideological slant of the search queries. The major implication of this finding is that search engines’ information curation actively mirrors the partisan divides present in the US media environments and has the potential to contribute to (perceived) polarisation within these environments.