- 20/10/2019
- Category: Digital Intelligence Lab EN
Author: Paska Darmawan
First published: October 20, 2019. Last update: April 27, 2021
On the 23rd of September, students and other protesters all across Yogyakarta marched down the Gejayan street to voice out their concern against several problematic bills that were on the edge of being passed by the Indonesian parliament. Activists were using social media to spread information about the protest. They posted jargons, demands, and digital posters using the hashtag #GejayanMemanggil (translation: Gejayan is calling) to persuade people to join the protest. As a result, there was a high turn out of people joining the protest even though the first tweet was only posted the day before, on September 22, 2019 at 00:15:56 local time.
The use of Twitter for mass mobilization has been well discussed since last decade. There are multiple cases where Twitter plays a huge role in mobilizing people, such as the #Occupy movements (Gleason, 2013; Penney and Dadas, 2014; Tremayne, 2014), Arab Spring (Streinert-Threlkeld, 2017), and youth protests in South Africa (Bosch, 2017) and Austria (Maireder and Schwarzenegger, 2012). Twitter makes it much easier for protest organizers to mobilize people due to its design as a social networking site (SNS). Twitter limits the number of characters in a post, which forces people to compose short and direct posts. Twitter also allows users to forward a public post easily by retweeting it to their followers. These two features facilitate quick and massive diffusion of messages, which may attract support from other communities that are not directly affiliated with the protest (Boyd et al., 2010; Recuero et al., 2019).
In these previous cases, we can see that it took a while for the organizers to raise public’s awareness and convert people’s interest into active participation in real life. However, the Gejayan protest successfully garnered more than 40,000 tweets under its #GejayanMemanggil hashtag within 24 hours, which led to a massive protest despite the short notice. This raised a question on how information about the protest was spread rapidly on Twitter.
To answer that question, this research has analyzed tweets containing the hashtag #GejayanMemanggil that were posted on September 22, 2019. In total, there are 48,574 tweets collected for this research. Social network analysis (SNA) was conducted to identify the clusters of conversation by using the modularity algorithm and to measure the influence of each user by using their in-degree centrality. Qualitative content analysis (QCA) will also be used to analyze the topics of conversation within the hashtag.
The Rapid Spread of #GejayanMemanggil
Based on the collected data, the first post mentioning the hashtag was tweeted on September 22, 2019 at 00:15:56 WIB (Western Indonesia Time). Data retrieval for tweets using #GejayanMemanggil prior to September 22 was attempted, but the search returned no result. The first tweet was posted by @mahasiswaYUJIEM, but it did not gain many responses from other users. As of October 9, 2019, it only has 2 retweets and 1 reply.
Note: The username has changed to “@lalidalanurip” as of April 27, 2021
The hashtag, however, started to blow up when the same account posted a giveaway for those who retweet and/or like their post. The post, and other subsequent posts in the thread, also contain digital posters about the protest. As of October 9, 2019, the post gained 784 retweets and 750 likes.
Note: The username has changed to “@lalidalanurip” as of April 27, 2021
Looking at the number of tweets over the hours, it can be seen that the hashtag started gaining traction after noon. The post count had begun to gradually increase until around 9 p.m., where it peaked with a total of 6,756 tweets.
Figure 1. Hourly Post Count of Gejayan Memanggil
From noon until midnight, there are a total of 16,190 tweets, of which more than 82% are retweets. Therefore, it is important to analyze the tweets that are being retweeted the most within this time span. Based on the data in Figure 2, most of these tweets encourage people to join the protest. This explains how Twitter became a medium for people to spread information about the protest to a wider audience.
Figure 2. Major Clusters within #GejayanMemanggil Network
Visualization below helps us understand how information spreads from one user to another. Interaction network within #GejayanMemanggil was analyzed by using dynamic network analysis. The graph shows that there are several key accounts whose tweets reach greater audience in less than 24 hours as their tweets got retweeted and, to a much smaller extent, replied to. The visualization is color coded based on their modularity class, which will be explained in the next section. One important thing to underline is how this hashtag is driven not only by a few influential users with high number of followers (e.g. BukuMojok, BerdikariBook), but also by other users whose reach of their tweet far exceeds their follower count (e.g. panjipnjk, tempelan_kulkas)
Figure 3. Dynamic Graph of #GejayanMemanggil Network
The Presence of Separate Agenda
The users were grouped by using modularity algorithm (see Blondel et al., 2008) to identify different clusters within the whole network. As a result, seven major clusters were identified with coverage of more than 5% each. In the visualization, smaller clusters and outliers have been filtered out for clarity.
From Figure 2, it is shown that most of these clusters have similar shapes. Almost all of them are quite centralized, with only one key account that acts as the source of information, represented by the bigger size of their circle and label. However, the graph shows that cluster #4 is more decentralized and has quite different structure compared to the other major clusters.
Figure 4. The Structure of Cluster #4
Looking at Figure 4, cluster #4 has a few main users who do not have significantly higher degree (number of connections) than other users in the cluster. This confirms the previous finding that cluster #4 is more decentralized and has different structure compared to the other major clusters, whose network is dominated by one key account.
Table 1. Top Account and Top Post from Each Major Cluster
The difference between cluster #4 and the other clusters can also be identified by analyzing the top tweets from each cluster. Table 1 shows that most of these top tweets have direct relevance with the protest by either declaring their participation, explaining about the urgency, or attaching visual media related to the protest. On the other hand, the top tweet from cluster #4 by @Anggraini_4yu brings in issues related to religious sentiment and the re-emergence of PKI (the now-defunct Communist Party of Indonesia), which are not directly relevant with the demands of the protest.
Figure 5. Most Frequently Mentioned Words in Each Cluster
To further see the different contents, all tweets were grouped based on their cluster. The most frequently mentioned words in each group were then identified and visualized by using word cloud. As we can see, most of these clusters contain similar words, except for cluster #3 and #4. For cluster #3, most of the tweets contain words like “giveaway”, “goodluck” [sic], and “beruntung” (lucky) as many of these tweets are in response to the giveaway post from @mahasiswaYUJIEM. The other frequently mentioned words within cluster #3 are still relevant with the protest. On the other hand, words that have no direct correlation with the protest, such as “Kamboja” (Cambodia), “PKI/Komunis” (communist), Al-Quran, and Muslim become the most frequently mentioned words in cluster #4. This finding further confirms how users in cluster #4 are trying to push another agenda by riding on the hashtag.
Conclusions
The findings suggest that activists and student-run university accounts played a huge role in spreading the information. They became key opinion leaders by posting information about the demands and the technical details of the offline protest. A smaller number of users that were not in direct affiliation with the protest organizers also gained a huge level of in-degree centrality by posting memes or sharing screenshots from their lecturers who give students permission to join the protest. There were also some political influencers who attempted to free ride on the hashtag and push another agenda that is unrelated to the protest. However, the number of these intruders were small, mostly because of the short time-span between the start of the hashtag and the offline protest. This indicates that quick mobilization through social media might prevent other parties from hijacking the movement. Further research with data from other cases and/or different time span is encouraged to verify this finding.
