Tweeting Against Climate Change

NGO’s use of Twitter to push their agenda in the public sphere

Team Members: Jeroen de Vos - Milan van Vugt - Isabelle van der Ende - Massimo Airoldi - Alberto Cossu - Christian de Bruijn - Christiaan Ate Paauwe -
Esta Kaal.

Introduction

COP21

From November 30^th to December 12^th, world leaders, researchers and specialists came together to discuss measures to make global warming come to hold during the 21st session of the Conference of the Parties to the United Nations Framework Convention on Climate Change, also known as the COP21 or the CMP11 (we will further refer to this event as the COP21). The COP21 was hosted and presided by France and was held in the Le Bourget exhibition centre in Paris. The strict goal of the conference was to come up with a resolution to the threat of global warming and find solutions as to how the global warming could be kept under 2ºC. On December 12^th the COP21 presented a report in which new global goals were described. In the report it was pledged, for the first time, to try and keep the global warming under 1,5ºC in order to save the small island states that are threatened the most by global warming and the rising sea levels it causes (“More details about” n. pag.).

During the climate top a range of NGOs and activists put effort in contributing to the on-going discussions with Greenpeace being one of them. Greenpeace is an international operating non-governmental organization (NGO) that states that: “A green and peaceful future is our quest” (“Our story” n. pag.). To achieve this future Greenpeace organizes campaigns to change current attitudes and behaviour. The goal of these campaigns is to protect and conserve the environment and promote peace. Currently one of Greenpeace’s main campaigns focuses on: “Catalysing an energy revolution to address the number one threat facing our planet: climate change” (“About Greenpeace” n. pag.). During the COP21 Greenpeace hosted several side events to educate people about green energy and to underline the importance of it. Spokespersons for Greenpeace have stated that in general they are very content with the outcome of the COP 21 conference, although they still feel that more should be done in the future. Kumi Naidoo, who was Greenpeace’s International Executive Director at the time of the COP21, communicated that view in a blog post on the Greenpeace website (n. pag.).

To get a clearer view on the topics that were covered at the COP21 conference, we have listed the key subjects of the conference per day. These topics are visualized in Table 1, along with side events hosted by Greenpeace’s as to get a better grasp on what exactly the organization has done during the COP21 (for a more elaborate idea of the topics covered by the COP21 and Greenpeace’s events, see Appendix I).

Table 1: Key topics per day of the COP21 and Greenpeace special side events.

30/11	1/12	2/12	3/12	4/12	5/12
Science Water Biodiversity	Forest Agriculture Carbon African Renewable Energy Initiative Greenpeace petition on forest fire.	Farmers’ Day Resilience Oceans Coastal ecosystems Technology Knowledge	YOUNGO Day Buildings Transport Energy efficiency	Ocean Day Education Green economy Fossil fuel Greenpeace renewable energy	District and Buildings Energy Efficiency GHG emission

6/12	7/12	8/12	9/12	10/12	11/12
South-South cooperation	Energy: efficiency, renewable, sustainable Finances Emissions gap report	Cities Business Innovations Carbon Air Greenpeace’s investigation on climate deniers (fossil fuel)	Forest Landscape Innovative technology Finances	Education Climate knowledge Human rights Human Mobility Finances	Mountain environment Closing press conference

Current research

The current research is initiated by Soenke Lorenzen, the research manager of Greenpeace International, and aims to examine the sub debate focused on renewables and green energy that Greenpeace set out to communicate on social media during the COP21. Lorenzen provided us with an overview of the focus points of Greenpeace’s communication during the COP21 (Wiedmann, Loeffelbein, Kaiser & Philippe). These focus points made it possible for us to focus our research on specific terms and agendas pushed by one of the most important NGOs involved in the renewable energy debate. In context of the COP21 Greenpeace’s communication narratives focused on: green actions, renewable energy, human rights and human movements. To convey these messages Greenpeace employed several slogans within their communication. These were:

• “Rise up for renewables”
• “Nothing stops climate change faster than our actions”
• “100% renewable energy for all by 2050”
• “100% renewables: A promise for peace”
• “Emissions anywhere affect people everywhere”

These slogans were translated into hashtags for Greenpeace’s communication on Twitter. The hashtags Greenpeace employed on a regular basis during these two weeks were the following: #go100percent, #go100percentre, #renawables, #COP21, #climate, #climatechange, #indigenous, #justice, #peace, #humanrights, #fossilfuels, #climatemarch, #marchforme, #actionsforclimate. There were several other NGOs that made use of the same hashtags to contribute to the integration of the green energy narrative into the larger COP21 narrative.

Within this research we will focus in particular on the broader question of the role of NGOs in the energy related debate reflected on Twitter during and after the COP21. The project will be informed by NGO stakeholders who will provide the meta­narratives developed and disseminated as part of their communications strategy. These stakeholders have defined their research needs that include an analysis into whether these narratives have successfully made their way into new media sites and platforms, and if so, whether they were primarily shared by actors within, what is dubbed the NGO "bubble", or whether some actors were migrating this conversation into the larger COP21 narrative on social media (Lorenzen n. pag.).

Research Questions

This research is build around one particular question, tapping into Greenpeace’s (and other NGOs’) desire to know what influence they had on the larger narrative on Twitter about COP21 in general and the energy debate in particular. In addition, this research focused on mapping the presence of NGOs in the sub narrative about renewables and green energy, hence our research question was:

To what extent was the effort of NGOs to situate the renewable energy debate within the larger narrative about COP21 successful and visible on Twitter?

In turn, this question has been operationalized into two sub questions. These questions allowed us to focus on different parts of our main question to eventually accurately formulate an answer to our research question as a whole.

RQ 1: To what extent is the energy debate visible and integrated within the larger COP21 narrative?

RQ 2: Which NGOs are part of the top influencers in the energy debate?

At the beginning of the process we expected to find a sub narrative about renewable energy within the larger narrative that had place on Twitter surrounding the COP21. However, we were not sure if that narrative was very visible and expected it to not be very connected to the larger COP21 narrative, as it seems to address a very specific sub focus of the climate top. As to the second part of our research question, we expected to find a lot of NGOs heavily invested in the sub narrative about green energy. We also expect to find Greenpeace, which is a world renowned NGO, to be very present in this sub debate and for it to be a large influence in connecting the renewable energy sub narrative to the larger narrative about COP21.

Methodology

Twitter

For this research we will draw upon Digital Methods for our methodology. Digital Methods is a term coined by Richard Rogers, who is the founder of the Digital Methods initiative, and refers to an internet-related research method whereby the web is approached as being a large dataset (Rogers, “Digital Methods for” 1). One of the main principles of Digital Methods is to build upon existing, dominant devices to extract relevant data from the web and subsequently turn that data into “indicators and findings” (Rogers, “Digital Methods” 3). Within this research we aim to repurpose data extracted from the social networking and micro blogging website Twitter and subsequently turn them into indicators and findings.

Debanalization of Twitter

In order to give meaning to our data set we have to debanalize Twitter and give meaning to the data we extracted from the platform. Research focussing on the Iran elections in 2009 already showed that Twitter data can very well be remodelled to tell interesting stories (Rogers, “Debanalizing Twitter” 5). 2009 was also the year that Twitter changed its slogan from “what are you doing?” to “what’s happening?”, emphasizing Twitter’s change from a banal social network to a more newsworthy medium on which current events as (natural) disasters and elections can be followed in detail (Rogers, “Digital Methods for” 19). According to Rogers the change in tagline was not only caused by a shift in how twitterers used the platform, but he also dubbed it a “nudge” to the Twitter users to comply the information sharing character of the platform (Rogers, “Debanalizing Twitter” 4). This shift in interpretation of Twitter’s characteristics is also the root of a new research approach to the platform. It can now be studied as a way to follow online conversation regarding offline events (Rogers “Debanalizing Twitter” 4). This is made relatively easy, because of Twitter’s built-in means of analysis. Collections of tweets can easily be organised and interpreted based on retweets, followers/followings, mentions, hashtags, replies, and shortened URL’s (Rogers, “Debanalizing Twitter” 7). This debanalization of Twitter and the real time character of it make it a very suitable platform to study the online public narrative on that surrounds the COP21.

Program and anti-program

A second interesting aspect as to chose Twitter as a medium for this research is about a specific mode of interaction that is possible on the social networking site due to the large dependence on hashtags for conversation ordering and labelling. Rogers dubbed this interaction “program versus anti-program” (“Foundations of” 5). The idea is that there is a keyword, a hashtag in this case, that represents one side of the debate. This keyword or hashtag is dubbed the program. In this research Greenpeace set out to promote the green energy program by using specific hashtags. It is very possible though that lobbyists who support fossil fuel based energy set out to fight Greenpeace’s program by launching a so-called anti-program. In this anti-program hashtags that convey an opposite view are employed (Rogers “Foundations of” 5). Within the current research this insight in program versus anti-program can really clarify different sub debates in the energy discourse on Twitter.

Authority on Twitter

The last motive as to why we ought Twitter to be a suitable platform to focus our research on, revolves around authority. Individuals, political parties, NGOs and researchers all have a say in the public debate on Twitter. Twitter itself does not categorize the different types of profiles; all tweets are seen as equal by the platform. However, over time some tweets become more influential than others, as do the people who tweeted them.

If a profile is followed by a large number of users it can give a first insight in the influence of the profile. However as Anger and Kittl state: “followers are without doubt important as they amplify the content, but quantity does not equal quality and a small audience of engaged users is worth more than a large audience of less active users” (4).

One way to measure the engagement of a user is by looking to the directed traffic it generates. We can do so by mapping the extent the user was mentioned either through addressing a message, retweeting its message or replying on the message. As for retweeting Haewoon Kwak, Changhyun Lee, Hosung Park, and Sue Moon have found that a retweeted tweet will reach an average of 1000 people, no matter how little followers the sender of the original tweet has. Moreover, retweeted tweets will almost always be retweeted by 2nd, 3rd and 4th parties, expanding the reach of the tweet even further (600).

These different ways to measure and indicate the ‘authority’ of twitterers will help us to indicate who is really influential within the COP21 debate as a whole, but even more important, within the sub discourse around green energy. As emphasized earlier, all these different ways to measure the authority of twitterers are built-in to the Twitter-platform and are thereby naturally an essential part of our dataset.

Dataset

The data we used in this research was gathered with the DMI-TCAT tool, which is a tool developed by the Digital Methods Initiative to gather and analyse tweets (Borra & Rieder 266). The tool relies on the Twitter API and Borra and Rieder therefore stress that the tool is bound to the limitations and possibilities this API brings forth (267). For this research the DMI-TCAT tool was instructed to collect all the tweets containing either “#COP21” or “COP21”. The gathering of the data started on November 29th 2015. The tool was still collecting tweets at the time we started our research. Within our research we focused on a specific time period beginning at the start of the conference (the 30^th of November) until a week after the conference (19^th of December) to capture how the discussion changed over time and ended in a new equilibrium after the actual event. Within this time period the TCAT-tool collected a total of 4.006.316 tweets.

The TCAT-tool allowed us to explore this dataset of tweets in several ways. Because the data set was too big to compute the entire database at once, we decided to focus on four particular days. Hereafter we will explain the different decisions we made starting the research process and which datasets we exported and studied with the help of the TCAT-tool.

Dates

The COP21 conference lasted from the 30^th of November until the 12^th of December. These begin and end dates were also the days that people tweeted the most about the COP21, as is visible in Figure 1. Therefore we choose to focus part of our analysis on those dates.

Figure 1: The amount of Tweets (blue line) from 29th of November until the 23rd of December. The two biggest peaks in Twitter activity are on the 30^th of November and the 12^th of December.

We also decided to focus on the 7^th of December since on this day the central topic of the COP21 debate were green and renewable energy sources (illustrated by Table 1 and Appendix I). To get an idea about the energy debate surrounding the COP21 after the conference, we decided to analyse the collected tweets of the 19^th of December as well, which is exactly one week after the final report of the conference was presented.

Renewable energy sub narrative

To determine if and how the energy narrative was situated within the larger debate about COP21 we exported several specific datasets from the TCAT-tool (Borra & Rieder). All these datasets were then either explored within Gephi (Bastian, Heymann & Jacomy), or studied and processed within Excel.

Top hashtag streamgraph

In a way to grasp the presence of the renewable energy debate within the debate as a whole, we employed the associational profile function within DMI-TCAT. This TCAT-tool output let us focus on the shifts in hashtag associations. To really get an overview of the “trending” hashtags within the COP21 debate we focused this part of the research only on the top fifteen hashtags that were used in association with COP21. By exploring the frequency of the top fifteen hashtags associated to COP21 of each day of the conference we were able to trace the evolution of the different facets of the discourse surrounding COP21 over time. In particular, this provided us to see the relative relevance of the energy related debate in relation to the whole COP21 debate on Twitter. This part of our research was not specifically focused on our four chosen research dates, but covered all the dates of the conference, so to not miss relevant information and to provide a context for the dates we later zoomed in on. We visualized this data in a streamgraph. The streamgraph was created online with the Raw data visualization tool created by members of the Density Design Research Lab (Caviglia et al.).

Top energy related hashtags

To get a first grip on the content of the data and to get an idea about the larger debate that surrounds COP21, we exported the data files that showed us the most used hashtags for our focus dates within this research. We then manually tagged the specifically energy related hashtags to explore how these hashtags are situated within the larger debate in terms of usage frequency. The following table, Table 2, shows how the various energy-related hashtags evolve over time in terms of usage.

Table 2: Usage frequency of the hashtags that were associated with the green energy debate.

Energy related hashtags	30-nov	07-dec	12-dec	19-dec	Total usage
keepintheground	1509	1045	1945	0	4499
auspol	1283	463	1345	103	3194
renewables	1159	695	841	32	2727
dkgreen	1117	333	750	6	2206
energy	584	955	288	112	1939
fossilfuels	963	257	666	12	1898
solar	845	523	340	50	1758
cleanenergy	823	244	331	7	1405
green	791	251	303	28	1373
carbon	346	668	243	75	1332
coal	628	218	297	58	1201
renewableenergy	258	576	117	14	965
co2	336	267	218	24	845
fossilfuel	355	221	262	4	842
go100percent	245	386	108	9	748
fracking	326	171	189	33	719
reenergise	136	515	58	3	712
nuclear	233	327	96	23	679
cleantech	224	308	96	8	636
water	211	294	81	12	598
divest	269	150	166	3	588
pollution	237	218	107	21	583
sustainable	221	210	85	20	536
renewable	211	127	71	12	421
oil	187	57	69	18	331
transport	117	120	38	3	278
emmission	45	123	6	0	174
nonukes	106	27	30	9	172

Based on this list we made an animation in the shape of a word cloud that would show the priority of the hashtags on every specific date. The hashtag that was used the most on that date has the largest font size, whereas the hashtag that was used the least has the smallest size. The font size differs with 4% in a range of 232-400 per cent of the original size of 22pt.

Co-hashtag network analysis

To determine if and how the energy related sub debate was part of the larger narrative around COP21 we exported co-hashtag network analysis from DMI-TCAT for all of the four studied dates. This kind of network provided us the opportunity to determine how the energy related debate was situated within the whole conversation around COP21. The networks were processed In Gephi and formed with the Forceatlas2 algorithm. To declutter the networks we set the degree range setting to 30 for the three days during the COP21. For the 19^th of December we adjusted that setting to 15, because there were less tweets send and therefore there was less data available. When we adjusted the degree setting to 30, we lost almost all nodes in the network. The node size within these networks is based on how often the hashtags are used.

The hashtag COP21 was automatically included in the network, but since the condition to be in the network was that the hashtag was used together with #COP21 all of the nodes in our networks were connected to it and the COP21 node stood central in our network. We therefore decided to delete this node from all four of the co-hashtag networks to get a better view of the modularity and connections between the other nodes.

Subsequently we studied the networks and coded the different nodes that were energy related (1) or not energy related (0), based on Table 2. This added dimension to the dataset enabled us to give the energy related and non-energy related terms different colours within the network. This colour dimension made it possible to clearly identify the sub narrative about energy and how it was positioned within the larger debate about the COP21 on the specific dates.

Renewable energy sub narrative top influencers

The second part of our research question is focused on the visibility of NGOs within the energy debate and how successful they were in pushing this debate outside of the NGO and experts “bubble” in which it normally seems to reside. To answer this part of our research question we exported additional datasets from the TCAT-tool.

NGO ratio analysis

Using the TCAT-tool we have generated a list of the users who were mentioned most often on the four researched dates. The 100 users who were mentioned most on each date were manually researched. Subsequently we categorized them into eight different groups. We had set up some categories before we started categorizing, but adjusted these as we found necessary while analysing. Eventually the categories we used where the following:

• Activists
• Governmental Organizations
• NGOs
• Public Figures
• Political Figures
• Press
• Researchers
• Other

Based on the results we calculated which percentage of the top of the overall Twitter debate about COP21 seemed influenced by each type of category. Based on this calculation we could distillate a ratio that would represent the influence of the NGOs in the top of the debate.

Top influencers within the energy debate

In order to get insight into the top 500 mentioned users within the energy related debate, the TCAT-tool was used to look into which parties were ascribed the most authority of the day the energy debate was on the agenda (December 7^th); and on December 12^th, the day the conference report was presented. These @mention user graphs give us insight in the most mentioned actors of the days. These days, the tool was queried with our list of energy related hashtags. Adding this query provided us with a network in which only the top mentioned people that are involved in the sub narrative concerning renewable energy. The query we utilised within DMI-TCAT was the following:

“renewableenergy OR renewable OR fossilfuels OR nuclear OR green OR cleantech OR coal OR keepintheground OR renewables OR co2 OR emmission OR carbon

OR pollution OR reenergise OR sustainable OR fossilfuel OR fracking OR divest OR auspol OR solar OR oil OR dkgreen OR nonukes OR energy OR cleanenergy OR water OR transport OR go100percent”

The user mention network was compiled for further analysis in Gephi. Users occurring in this most mentioned list are actually based on tweets containing at least one of the above keywords. Therefore one can interpret the top mentioned users to have a certain authority in this sub debate (Anger & Kittl 4). Within Gephi the number of mentions determined the node size and the colour was derived from whether the node was one of the larger NGOs coded in the overall discussion (as specified in Appendix II). We did this in order to compare the effort of NGOs presence in both the sub debate concerning renewable energy and the overall COP21 discourse. The Forceatlas2 algorithm has been used together with stronger gravity and LinLog mode to show the top 500 mentioned users all in one sphere while showing the connectedness of the actors.

Findings

Renewable energy sub narrative

In order to answer the first part of our research question, focusing on the reach of the energy debate outside of the so-called “bubble” of experts, we looked at the discourse about COP21 on Twitter in several ways.

Top hashtag streamgraph

To get an initial idea of the debate on Twitter surrounding the COP21 we visualized the top fifteen hashtags that are used in combination with COP21 in a streamgraph. This graph presented below in Figure 3. Within the streamgraph we marked the hashtags that are specifically related to the debate around energy (as specified in Table 2) in a bold green. The other hashtags are visualized in a lighter green.

Figure 3: Streamgraph of the top fifteen hashtags associated to COP21 on Twitter. The energy related hashtags are visualized in a bold green, whilst the non-energy related hashtags are visualized in a lighter green.

By looking at the top fifteen hashtags associated to COP21 in each day of the conference, we can argue that energy-related hashtags are quantitatively marginal in the overall discourse. This visualization then indicates that, although energy-related hashtags are present, on ‘busy’ Twitter days (visualized here in terms of hashtags used by the length of the graph over the y-axis) they seem to be overshadowed by more general and popular hashtags in usage frequency (e.g., climate change etc.). However the absence of the energy related hashtags on most days does not mean that the debate surrounding energy was not present, it only indicates the hashtags associated with this debate were not among the most used hashtags.

Top energy related hashtags

The screenshot of our animation in Figure 4 is supposed to give insight in the evolution of the energy related hashtags on the four dates we examined. The animation shows us how topics regarding renewable energy (as was the focus of Greenpeace) grow on the 7^th of December, compared to the two prior dates, whereas the hashtags concerning fossil fuels where the most used on the last day of the conference. From this animation it is not clear however in what context these fossil fuel hashtags were used. The animation does however tell us that the 7^th was the day where the discourse about renewable energy was the most popular. This is not that surprising, as we know on December 7^th the renewable and green energy debate was a central topic of the conference (See Table 1 and Appendix I).

<iframe width="560" height="315" src="https://www.youtube.com/embed/KagjuTqbH2Q" frameborder="0" allowfullscreen></iframe>

Figure 4: The animation of the top energy related hashtags and their evolution between November 30^th and December 7^th, 12^th and 19^th. For the full video see: https://www.youtube.com/watch?v=KagjuTqbH2Q

Co-hashtag network analysis

By visualizing the hashtags that users frequently used together in a network graph we hoped to complement the insights that the streamgraph in Figure 3 provided us with. Co-hashtag networks are also able to give us a lot more in depth knowledge of the specific energy related debate on Twitter and how it is situated within the whole COP21 debate.

The co-hashtag networks are shown in Figures 5, 6, 7 and 8. In these networks we marked the energy related hashtags in green and the non-energy related hashtags in pink. What is visible in all the networks is that there are a few, very general, hashtags in the centre of the network graphs. These hashtags are used the most, as becomes clear from the node size, and are also used with all the different side debates concerning the COP21, as is pointed out by the central position of the nodes.