-- WarrenPearce - 22 Aug 2016

How is the topical space of COP21 shaped on Twitter and YouTube?: From issue climatisation to issue drift

Group leaders: Nicolas Baya-Laffite, Warren Pearce. Team members: Donya Alinejad, Laura Candidatu, Gabriele Colombo, Melis Mevsimler, Domitilla Olivieri, Tiago Salgado.

Contents:

	*  Introduction 
	*  Research Question 
	*  Datasets 
	*  Twitter: User-curated issue climatisation 
	*  YouTube: algorithm-curated issue drift 
	*  Conclusions 
	*  Discussion 
	*  References 

Introduction

The Rio Declaration in 1992 brought about two global policy agendas: climate change and biodiversity. Of the two, climate change has become the most visible, mobilising the international community of states and civic society. Climate change is a ‘wicked issue’ (Grundmann, 2016; Hoppe, Wesselink, & Cairns, 2013; Incropera, 2015). On the surface it is very simple, arising from technical information regarding the effect of increasing greenhouse gas emissions on the Earth’s climate. However, wicked problems require “profound understanding of their integration in social systems, their irreducibly (sic) complexity and intractable nature” (Prins et al., 2010). Bringing together technical information with social systems is a political process, and has led to NGOs and other actors strategically cross-framing issues with climate change. This is known as issue climatisation (Aykut, Foyer and Morena, forthcoming). The high profile COP21 meeting in Paris, 2016, provided an important strategic communication opportunity for NGOs to climatise certain issues. Online platforms provide space for climatisation dynamics to unfold. In this project we illustrate these dynamics on Twitter and YouTube, and how they co-exist with separate platform effects of issue drift.

Research Question

How is the topical issue space of COP21 shaped on Twitter and YouTube through dynamics of issue climatisation and issue drift?

Datasets

We utilized the existing COP21 Twitter dataset available through the TCAT-DMI tool, which includes all tweets containing the string ‘COP21’. We extracted YouTube data via YouTube Data Tools, querying the ‘Related To’ function in order to discover the relationship between the top 50 ‘most viewed’ videos containing COP21 in the title or description, and the top 50 ‘most relevant’ videos to COP21 as ranked by YouTube. The ‘Related To’ function refers to those videos that YouTube suggests to be related to the current video being viewed (before taking into account a YouTube user’s viewing history). These are displayed down the right-hand sidebar on the desktop version of YouTube and are titled ‘Up Next’ (see Figure 1). These videos are referred to as ‘Up Next’ for the remainder of this report. YouTube datasets were extracted on 6/7/2016.

Figure 1: screenshot illustrating YouTube videos ‘up next’ subsequent to one currently being viewed, as suggested by platform algorithm.

Twitter: User-curated issue climatisation

During COP21, a wide variety of Twitter users helped to shape the topical issue space on the platform through tweeting, hashtagging and content sharing. In doing so, they often inscribed their own causes into COP21 on-line space. By analysing those hashtags that most commonly co-occur with the search term 'COP21', we can gain an insight into this issue climatisation. In particular, many of the concentrated bursts of activity in displayed in the bottom half of the chart represent concerted efforts to publicise issues connected to climate change: for example, cities, human rights, climate justice, renewable energy.

Figure 2: Prevalence of hashtags co-occurring with 'COP21' in tweets published Nov 29th-Dec 15th 2015.

During the project, we produced user-hashtag maps to investigate which users were using particular hashtags (e.g. see slides 6 and 7 here), which illustrated the Babel-like nature of Twitter - a multitude of chattering voices each with their own take on COP21. We focused on the role of video in shaping the issue space, noting that this topic is comparatively under-researched in the climate change communication literature (Schäfer & Schlichting, 2014). In particular, we compared the top 400 YouTube videos that Twitter users tagged as 'COP21' with the top 400 videos presented on YouTube as being most relevant to COP21. The comparison was done using the Triangulation tool. Only 31 videos appeared on each list, an overlap of 7.8%. This suggests that curation methods matter; in this case, the user-curation of COP21 videos carried out through Twitter provided a very different set of results to the algorithmic curation of COP21 videos carried out on YouTube.

YouTube: algorithm-curated issue drift

To gain further insight into how the COP21 issue space is presented on YouTube, we analysed the most viewed videos on YouTube containing COP21 in the title or description, using You Tube Data Tools. By far the most viewed video (as of 6/7/16) is " Koko the gorilla is the voice of nature at COP21" with 1,614,777 views (the second most commented video, from the official COP21 channel, has 570,929 views). The video was uploaded by Noé, a small biodiversity NGO based in France. The Koko video has been by far their most successful video; their next most viewed video upload had only 3.568 views (as of 6/7/16). The video is a short (1’15”) campaign spot focused on strategically framing biodiversity as a cross-cutting issue that needs to be included in the COP21 negotiations. The strategic choice of the NGO is clear: instead of picking up a scientist as a “nature’s advocate”, Koko the gorilla was presented as not only as a more charismatic, but also more concerned porte-parole of nature, who would more successfully enroll larger audiences into supporting their cause: not just “climate change”, but “includ[ing] “the preservation of biodiversity in the Paris Agreement”. On the face of it, the Koko COP21 video was a huge success, both in terms of their own outreach efforts and in terms of making COP21 and the Paris Agreement ‘visible’ to wider audiences beyond issue advocates for climate change. But how was COP21 subsequently presented to YouTube users who viewed the Koko video? In theory, Koko presented a new entry point into understanding COP21. ––– one would expect to see other videos recommended on COP21 or climate change. The YouTube platform presents videos related to that being currently viewed down the right-hand sidebar of the screen under the heading ‘Up Next’ (see Figure 1). If the user lets the video play out, then YouTube subsequently ‘autoplays’ the next related video. The algorithm(s) that YouTube uses to determine these relevant videos is a key means by which the platform shapes the content that is visible to the user. These algorithms are designed to maximise the 'watch time' of YouTube users, yet we know little about how they work (Gielen & Rosen, 2016). They are proprietary and opaque. Extracting lists of these ‘Up Next’ videos can help to illuminate how the algorithms work. In particular, the Koko video provides a stark example of the platform effects of YouTube. Below is a network diagram of the 50 most viewed YouTube videos, showing how they are related to each other on the platform. That is, once you start watching one of these videos, which other ones will the platform present you with as being related?

Figure 3: 50 most viewed COP21 videos on YouTube as of July 8, 2016, and their relationship through the ‘up next’ algorithm.

Each node represents a video. There are three clusters, broadly representing the official, ceremonial parts of the conference (blue), activism (green). Nodes size is related to the number of views for each video on YouTube. What is interesting here, is that Koko is a very popular video related to COP21, but is not a route into the rest of the network. So although Noé was very successful at creating viral content, it was unsuccessful as a trampoline into the issues of biodiversity and climate change. On YouTube, the Koko video only led users to other Koko videos. Despite the popularity of the video (showed by the size of the node), YouTube does not relate it to other ‘most watched’ videos on COP21. A similar phenomenon is observable with the 50 videos identified by YouTube as ‘most relevant’ to COP21 (the default search setting on YouTube). Both networks are visible below. Figure 4: YouTube ’s ‘Up Next’ algorithm detaches the most popular COP21 video from other COP21 content.

In short, the ‘Up Next’ algorithm betrays the NGO’s trampoline strategy for inscribing biodiversity in the COP21 conversation. So if viewers of the Koko video are not subsequently pointed towards more videos about COP21 and/or climate change, what are they directed towards? The simple answer is: more Koko videos. Figure 5 shows the ‘Koko universe’, the videos that YouTube suggests to viewers of the Koko COP21 video. None of these videos are related to COP21.

Figure 5: network of 50 related videos to ‘Koko the gorilla is the voice of nature at COP21’ (seed).

Taken together, these networks provide a clear picture: that while the Koko video was hugely successful in its own right, it proved to be a cul-de-sac for exploring the COP21 issue on YouTube. The algorithm’s logic of attention-keeping through topical drifting privileges the Koko element, rather than the COP21 element

Conclusions

On Twitter, users played a key role in shaping the COP21 issue space by linking specific issues to the broader topical issue of climate change that were the subject of the negotiations. Figure 2 provides an overview of the ebb and flow of this issue climatisation during COP21. A comparison of YouTube URLs linked to COP21 on Twitter and YouTube provide an insight into the effects of each platform in shaping the topical issue space. On Twitter, the prevalence of user curation facilitates issue climatisation, as Twitter users link their own concerns to COP21 (see also Pearce, Holmberg, Hellsten, & Nerlich, 2014). When searching for ‘COP21’ on Twitter, one is only exposed to tweets containing that term. On YouTube, algorithmic curation is more dominant, particularly through the presence of ‘Up Next’ videos that seek to maximize the ‘watch time’ of users rather than facilitate issue exploration. The impact on issue spaces of different curation methods was visible through the distinctive nature of COP21 videos on Twitter and YouTube. The overlap between ‘most tweeted’ YouTube videos within COP21 tweets, and the ‘most relevant’ COP21 videos according to the YouTube platform was only 7.8%. YouTube ’s algorithmic curation was particularly evident in the case of the most viewed COP21 video, featuring Koko the gorilla. Even though this video was a successful example of issue climatisation in its own right, linking biodiversity to climate change, YouTube users were subsequently directed to other videos about Koko rather than videos about COP21 or climate change. This is an example of issue drift.

Discussion

If the emerging literature in social media climate communication is to provide meaningful insights, it must ground ‘big data’ analysis in the “key contexts which give social media posts their meaning” (Pearce, Brown, Nerlich, & Koteyko, 2015). The findings from this project suggest that these contexts should include the effects on communication that are inherent in the social media platforms themselves. More radically, one might query the role of YouTube (one of the world’s most popular websites) in issue exploration and public education. Can we dream of a plug-in that repurposes YouTube from a distraction machine to an issue exploration machine?

References

Gielen, M., & Rosen, J. (2016, June 23). Reverse Engineering The YouTube Algorithm. Retrieved from http://www.tubefilter.com/2016/06/23/reverse-engineering-youtube-algorithm/

Grundmann, R. (2016). Climate change as a wicked social problem. Nature Geoscience, 9(8), 562–563. http://doi.org/10.1038/ngeo2780

Hoppe, R., Wesselink, A., & Cairns, R. (2013). Lost in the problem: the role of boundary organisations in the governance of climate change: Role of boundary organizations in the governance of climate change. Wiley Interdisciplinary Reviews: Climate Change, 4(4), 283–300. http://doi.org/10.1002/wcc.225

Incropera, F. P. (2015). Climate Change: A Wicked Problem: Complexity and Uncertainty at the Intersection of Science, Economics, Politics, and Human Behavior. New York, NY: Cambridge University Press.

Pearce, W., Brown, B., Nerlich, B., & Koteyko, N. (2015). Communicating climate change: conduits, content, and consensus. Wiley Interdisciplinary Reviews: Climate Change, 6(6), 613–626. http://doi.org/10.1002/wcc.366

Pearce, W., Holmberg, K., Hellsten, I., & Nerlich, B. (2014). Climate Change on Twitter: Topics, Communities and Conversations about the 2013 IPCC Working Group 1 Report. PLoS ONE, 9(4), e94785. http://doi.org/10.1371/journal.pone.0094785

Prins, G., Galiana, I., Green, C., Grundmann, R., Korhola, A., Laird, F., … Hulme, M. (2010). The Hartwell Paper: a new direction for climate policy after the crash of 2009. Retrieved from http://eprints.lse.ac.uk/27939

Schäfer, M. S., & Schlichting, I. (2014). Media representations of climate change: a meta-analysis of the research field. Environmental Communication: A Journal of Nature and Culture, 8(2), 142–160. http://doi.org/10.1080/17524032.2014.914050