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 YouTube has been accused of being both too lax and too censoring of actors and content (in different contexts) [76]. If global search results on climate change follow the patterns from English/US studies on mainstream climate search terms, we would expect them to be rela- tively conservative: they would be primarily characterised by ‘usual suspects’ and classic visual themes. Previous research on US/English samples finds that news media channels are promi- nent, and that they dominate among those showing classic distancing themes such as the ‘dis- aster impact’ frame [69,70]. We can expect continuities across regions on this point. Climate change is precisely the type of issue for which YouTube elevates authoritative sources such as news media, and research on global climate journalism shows that while there are region- specific trends, there is continuity across countries in common news frames, including the distancing frame [12,14]. Regarding the user-led channels, studies of YouTube Search indicate that they also appear in top results, and that they tend to be more popular [69,77]. However, it is less clear what to expect from them with respect to local connection in content. A qualita- tive study with focus on ‘gain’ and ‘loss’ frames observed that user-led climate channels tend to re-inscribe legacy media repertoires into their videos [78], but the study on the disaster impact frame did not note user-led channels using this frame in top results [69]. 4 Method

This study pursues an approach of quantitative description [17] and large-scale observation [11,66] to examine what YouTube Search presents in top rankings and the underpinning pro- cesses with respect to climate change. For our first question, about platform presentation, we consider the channel actors and visual content in top ranked results. For the second question, on the underpinning processes, we consider the platform, channel, and audience sides of the ranking assemblages. 4.1 Delimitations As we detail in the following sub-sections, we make several key delimitations in order to focus platform-level processes in this context. Chief among these include the following: • We analyse only the visual dimension of YouTube and its videos, and do not consider other aspects such as the audiovisual elements and user commentary. • We look at API results, not at what specific users would see when searching (e.g. based on their search history or other criteria that YouTube might be using but not corresponding to specific API settings). • We focus on a keyword that has been shown to have a mainstream profile on YouTube search, ‘climate change,’ [70], since we expect green ghetto discourse to characterise main- stream communication. For this reason, in this study we have not used other keywords from the literature, such as ‘climate hacking’ or ‘chemtrails’, better suited to the study of specific aspects within climate change and/or specific groups of producers and audiences. Practically, using a general keyword also simplifies the translation to other languages,

which is an important factor given the world-wide scope of our research. • We examine only one kind of distancing theme (in its arctic rendition), which is selected to give us leverage on the comparative analysis of relevance in terms of freshness and local- ness. • We approach the image data as still images, and do not consider their sequence since we are only looking at which videos show this type of content. • Finally, although we are working with a sizeable dataset, our observations are by design based on small sample sizes, since we focus on the videos that the platform ranks at the very PLOS ONE https://doi.org/10.1371/journal.pone.0318338 April 8, 2025 7/ 23