Visual digital intermediaries and global climate communication: Is climate change still a distant problem on YouTube?

PLOS ONE Visual digital intermediaries and global climate communication (which actors) and concentration (their distribution) can be crucial. A number of considera- tions may play into making results relevant, including freshness, user satisfaction, and intro- ducing fresh content by serendipity (while perhaps balancing with sponsorship). The evidence challenges common assumptions about the constraints on diversity in typical ranking logics [56]. However, studies also present instances in which particular platform ranking practices privilege large or already popular sources while under-representing others [57], concentrate attention to a select number of sources [58], or skew content due to differences in what high- and low-ranked sources present [59]. The second strand examines the relevance of top-ranked content for differently situated audiences in terms of proximity, which has bearing on issues of local connection. Impor- tantly, this work illuminates systematic variations in the ‘localness’ of top-ranked content (beyond personalisation). Platforms roll out features and moderation technologies at differ- ent pace across locations or in keeping with local regulation [60]. The same arrangements may also produce distinct results in different contexts. In one study, for example, Google Search returned city information with different localness of sources depending on how developed publishing and scientific industries were in the respective regions [61]. In a different take on local connection, another study found that object recognition systems of the kind that

underpin image search performed poorly in recognising household items such as spices from non-Western and low-income communities [62]. These two strands inform our approach to the question of how platforms present climate change to differently situated publics, and specifically the degree to which top-ranked content includes fresh voices and local connection. We specify our first question as follows: (RQ1) What characterises top-ranked results for climate change with respect to diversity and proxim- ity on YouTube Search — and does this differ across region/language search contexts? 2.3 The conditions of visibility across contexts Another important part of what digital intermediaries do hinges on the processes that pro- duce top ranks. If we want to understand the role of digital intermediaries in climate discourse in light of the green ghetto, we need to consider not only whether fresh voices and local con- nection appear in the top results, but also what the conditions of visibility are for this to occur. Here, the workings of the digital intermediary go beyond ‘the’ platform or ’the’ algorithm to also encompass actors such as the users that post content and the audiences that engage with it. Research on how algorithms intervene in the world emphasise the socio-technical dimen- sion [63]. A strong case is made for acknowledging the distributed agencies that converge as assemblages in ranking processes [11], what we here refer to as ‘ranking assemblages’. On this view, ranking processes are embedded in, and are an entanglement of, platform politics [64]

and use cultures [65]. Intensifying the multi-sided tangle, interaction and cues run in sev- eral directions. For example, between algorithms premised on user engagement (among other things), actors that post content and anticipate such curation (among other things), and audi- ences that discover content to like. Consequently, in specific assemblages, some outcomes become more likely and others less so, which circumscribes the conditions of visibility in that context. Taking this starting point seriously implies that it makes sense to consider ranking assem- blages and their conditions of visibility in a comparative perspective. Given that not only plat- forms but also other actors imprint ranking assemblages, we can expect differences across contexts in several tangents. On the one hand, as noted, platform-side doings may in fact vary PLOS ONE https://doi.org/10.1371/journal.pone.0318338 April 8, 2025 5/ 23