Who tweets in South Africa? And what are they tweeting about? KYLE FINDLAY looks at how last year played out on the country’s most raucous social media platform, from #paybackthemoney to #FeesMustFall and everything in between.
South Africa’s political landscape is vibrant and raucous. While our politics are not for the faint hearted, it was exhilarating to see so many in our country find their voice for the first time in 2015. Most South Africans who grew up under apartheid were cowed either into submission or an apathetic stupor. Blacks, coloureds, Indians and Asian citizens were taught to know their place, while white citizens were taught not to question the comfort they found around them. Of course there were vocal, and sometimes violent, exceptions to this in the forms of activists across the colour spectrum but, to a large extent, this was the status quo bequeathed to us as 1994 dawned – a politically unengaged electorate.
The status quo has changed. Many South Africans have come of age without the shadow of apartheid looming directly over their heads and they have not been subjected to the same social engineering that their parents were. Instead, they have been taught that they too can be a doctor or president; except, when many critically appraise their surroundings, they conclude this to be an empty promise. They recognise that our society simply is not structured to allow most young millennials to reach their full potential. Unfettered by the past, the new generation is making their dissatisfaction over this realisation known and one of the primary outlets for doing so is Twitter. Twitter is the digital equivalent of a giant public space where anyone can mount a soapbox and attract a crowd to listen to what they have to say. If a message resonates with enough people, movements can form.
For the purposes of this article, I wanted to understand who the main constituencies and ‘orators’ were that led the South African political discourse in 2015. To do this, I merged together all of the datasets relating to political events that I accumulated last year via Twitter’s publically accessible data platform (known as an “API”). The result is a dataset of 2.5 million tweets across 635,000 individual users. Let’s see what this combined 2015 dataset has to say about who the main communities are in the South African political landscape and who the main influencers were leading each community’s conversations. It is these same voices and constituencies who are likely to shape our politics in 2016.
The specific datasets that I’ve grouped together consisted of tweets relating to each of the following topics in chronological order:
- The State of the Nation address (SONA), where the Economic Freedom Fighters (EFF) called for President Jacob Zuma to “pay back the money” on his controversial R200 million-plus housing development at Nkandla that was paid for out of the public purse
- The #RhodesMustFall movement, where students at the University of Cape Town called for the removal of the statue of British imperialist and nation builder, Cecil John Rhodes
- The xenophobic outbreaks across the country where foreigners were attacked for supposedly stealing local jobs and not contributing to the economy
- The release of the Public Protector’s report into the legality of President Jacob Zuma’s R200 million-plus expenditure on his private housing at Nkandla
- The #FeesMustFall movement where students rose up to demand free education as the first step towards redressing the structural inequalities in our society (read my analysis of this event here)
- The EFF march in Johannesburg towards the end of last year
- President Zuma’s firing of then finance minister, Nhlanhla Nene… (read my analysis of this event here)
- …and the subsequent #ZumaMustFall movement that emerged in response (read my analysis of this event here).
Before we begin, there is an important caveat to bear in mind with this data: I did not collect every single tweet relating to every event. Twitter’s public systems limit how many tweets you can get out of it at one time. This is only really a problem for very popular events such as the State of the Nation address and #FeesMustFall where hundreds of tweets were generated every second – a rate too high for Twitter’s public API which means that some tweets were not collected during these spikes in tweet volumes. As a result, these events are surely under-represented in this article. Even so, we can still gain much insight from the data that we do have.
With that out of the way, let’s start by having a look at a timeline of events across 2015. Figure 1 shows the daily volumes of tweet generated during each event. It gives us an idea of the magnitude of each event and when they happened in relation to each other:
Figure 1: Time series for 2015 (20 Jan – 26 Dec 2015) showing daily tweet volumes. The spikes are colour-coded by event.
We can see that South Africans enjoyed a reprieve between June and October last year from the regular battering that we otherwise endured (of course, this doesn’t take into account the events that I might not have collected data for). Aside from this, we weathered a major political event roughly every couple of months.
To get an even better feel for the magnitude of each event, we can put each dataset side by side (see Figure 2). This gives us an idea of the scale and longevity of each event, with implications for the potential impact of each event on the national zeitgeist:
Figure 2: This chart shows us the daily tweet volumes for each event. By putting the datasets side by side, we can see which ones stand out as having generated the most conversation and having lasted the longest. Events that generated tweets at a particularly high velocity such as the State of the Nation (SONA) address, the xenophobic violence and #FeesMustFall are likely under-represented in terms of volumes.
#FeesMustFall was the largest event of the year. It has clearly left a mark on our society given the subsequent coverage and discussions that is spurred on although it remains to be seen what kind of momentum the movement will recapture in 2016.
The second largest event was the conversations around the xenophobic attacks early in the year where foreigners were attacked for supposedly taking local jobs and not contributing to the economy. The vast majority of conversations condemned the attacks. The attacks had the counter-intuitive effect on Twitter of bringing together South Africans in solidarity for our fellow Africans. It was great to see the outpouring of condemnation of the acts and support for the victims of these terrible acts. South Africans showed their deep humanity, proving that we have a huge capacity for reason and empathy.
One could argue that the third largest event was the combined firing of then finance minister, Nhlanhla Nene, and the subsequent calls for President Jacob Zuma to resign under the #ZumaMustFall banner. Many were deeply unhappy with President Zuma’s rash decision and his actions have had a lasting effect on the national zeitgeist (not to mention the economy). Like #FeesMustFall, it remains to be seen what kind of momentum this sentiment will maintain in 2016 but it is clear that ordinary citizens have forced President Jacob Zuma, and the ANC by association, onto the back foot going into 2016. Indeed, if we perform a simplistic analysis whereby we group together all of the events covered in this article into broad themes, we see that three things caused the main South African Twitter storms of 2015: social movements (such as #FeesMustFall and #RhodesMustFall), President Jacob Zuma’s actions, and xenophobia. Figure 3 summarises the cumulative impact of each in terms of the volume of conversation generated, which gives us some hints as to the impact of each on the broader society:
Figure 3: We can categorise the various events into the broad themes based on whether they related to social movements, the actions of President Jacob Zuma and the ANC, xenophobia or other types of events.
The next step is to understand who the main constituencies, or communities, were that led these conversations. By looking at each community’s agenda and seeing which communities were most active during each event, we can get an idea of who steered the political discourse and events in our country.
To understand who the main communities were, I created an interaction network, or ‘conversation map’, by connecting users together whenever they interacted with each other via retweets and @mentions. I then ran a community detection algorithm on this network to find the distinct groups of people interacting with each other more than with other groups. This is what the conversation map looks like for South African politics in 2015 across 2.5 million tweets and 635,000 users:
Figure 4: The conversation map across all 2015 datasets shows some clear community structure with distinct groups discussing the various events. This network represents 635,107 unique Twitter users.
The conversation map shows some distinct community structure with communities consisting of revolutionaries and activists, mainstream media, political parties and international commentators. Here’s a breakdown of the ten communities with the most members:
Figure 5: These are the top ten largest communities in the conversation map based on number of unique users within each community as identified using our community detection algorithm (users can only belong to one community). The chart also shows what proportion of all tweets in our dataset were generated by each of the top communities.
An unprecedented 52% of all tweets in our data came from just two communities. One might characterise these two communities as representing “the establishment” and “the anti-establishment” respectively.
The positions of the user nodes in the network are not arbitrary. User nodes are positioned based on who they interacted with. As a result, we can start to draw some tentative conclusions based on the positions of the communities in the map. For example, the mostly white mixed community was more internationalist in its interactions: it interacted with international news media and influencers more than other communities did. The black middle class was closer to the activist community, which makes sense since most activists, for economic and historical reasons, are likely to come out of the black community, implying a greater overlap between these groups than with others.
Figure 6: Close-up of the largest community in our dataset which included most political parties and mainstream media organisations. It represented 11% of all users but it generated a massive 31% of all tweets
The single largest community (11% of users in our data) arguably represented the establishment. It was made up of political parties and mainstream media. Given that it only represented 11% of all users, the community generated an unprecedented 31% of all tweets. This points to the incredibly engaged nature of the discussions within the community which is unsurprising given that the highly emotive and important topics discussed. Community members cut across races and if there was one community that represented the Rainbow Nation in all its diversity, it would have been this one. The community also included news organisations such as POWER987 News, Times Live, IOL News, City Press, SABC News Online and the Mail & Guardian, amongst others.
I was surprised to see that the algorithm lumped all major political parties – the ANC, DA and EFF – and their mouthpieces together, implying that they are all part of the establishment according to our algorithms. As we’ve already discussed though, the position of individual user nodes is not entirely arbitrary. We do see that the EFF sub-community sits closer to the activist community, which we would expect given the overlap in revolutionary ideologies between these groups.
The official @MyANC_ account generated the most action throughout our entire dataset. It was @mentioned more than it was retweeted, implying that people were talking to it (often when angry about its policies and decisions) more than they agreed with, or consumed, what it had to say. However, it still generated a large number of both @mentions and retweets overall.
Figure 7: Close-up of the second largest community in our dataset which included activists and social movement-related accounts and their followers. It represented 8% of all users but generated 21% of all tweets
I’ve referred to the second largest community (8% of all users) as the “Activists & social movements” community; so named because it included accounts such as @RhodesMustFall, @TheDailyVox (the student-led news organisation that popularised much of the initial #FeesMustFall discussions), prominent individual activists, and their followers. For privacy reasons, I’ve blurred out the names of individuals’ accounts within the blue sub-community that was centred around Wits University and who were the progenitors of the #FeesMustFall movement. The blue Wits sub-community stands distinct within this community from the orange #RhodesMustFall community. I assume that this represents a regional divide between groups (UCT in Cape Town and Wits in Johannesburg; the respective homes of each movement), although the divide might be ideological (I’m not informed enough to say). It’s interesting to see uber-influencer, Khaya Dlanga, on the periphery of this community. In the past he would have formed part of the mainstream community (or perhaps the black middle class community). While he clearly has his own voice that speaks to a broad variety of South Africans, the kinds of things that he said appears to have resonated strongly within this community.
We could continue unpacking the remaining top communities, but in the interest of space and time, I’m going to leave it at this juxtaposition of the mainstream establishment versus the revolutionary anti-establishment.
Let’s finish off with a high level look at who the top conversation leaders were in 2015. We use a pragmatic definition of ‘influence’ that is based on a user’s ability to spur action in others. We quantify this by looking at who had the most retweets and @mentions.
So, who then were the top political conversation leaders within South Africa in 2015? Based on our definition of influence, and within the limits of the data I had to work with, the top influencers were unsurprisingly made up of a varied mixture of news organisations, politicians and activists.
Here then are the top twenty accounts that led the South African political discourse on Twitter in 2015 (I’ve included their Twitter profile picture and descriptions for easy reference):
Figure 8: The top twenty conversation leaders, or ‘influencers’, across our entire dataset for 2015 based on how many retweets and @mentions they received. Their Twitter profile pictures and descriptions are shown.
From this list, we can see that the major political parties are well represented. The ANC’s official @MyANC_ account had the most combined retweets and @mentions of all users in our data. This is a surprise since, back in 2011 when I started researching South African politics on Twitter, the ANC had almost no presence on Twitter at all. They’ve clearly upped their game in the intervening years.
The EFF beat out the DA for second place thanks to the prolific tweeting of former banker, Sentletse Diakanyo, who acts as the EFF’s unofficial voice on Twitter. He plays a close second to Khaya Dlanga (who also makes it into the top twenty) when it comes to appearing in every South African Twitter dataset that I’ve ever looked at. Diakanyo’s fiery, no-nonsense and often controversial comments strongly resonate with many South African Twitter users.
Rounding out the list we have several activists out of Wits University whose tweets were widely picked up on during the #FeesMustFall marches in November 2015. The official Rhodes Must Fall account also makes an appearance in our list implying that it has true traction.
So, what can we take out of this brief analysis? A few things:
As South Africans, we’ve become accustomed to weathering a major social upheaval roughly every few months. In 2015, it became the new normal, which doesn’t give the South African psyche much respite between repeated psychological batterings.
A few events stood out in terms of magnitude. To the extent to that the volume of Twitter discussion around each topic gives us an indication of that topic’s impact within the broader South African society, President Zuma’s State of the Nation address, the xenophobic attacks and the #FeesMustFall movement had the largest individual impacts, followed by President Zuma’s firing of then finance minister, Nhlanhla Nene, and the subsequent call of #ZumaMustFall.
There is a clear divide within the South African political discourse between the mainstream establishment and the revolutionary anti-establishment. The mainstream is anchored around large accounts belonging to major political parties such as the ANC, DA and EFF, as well as news organisations, while the revolutionary, anti-establishment communities are more diffuse, with few clear centres of gravity. This is a pattern that we often see with highly engaged, de-centralised movements and points to a ‘healthy’ community that will likely persist over time. The two largest communities demonstrated healthy levels of engagement, arguably acting as a counter-balance to each other, which bodes well for the country in my opinion. It will be interesting to see the extent to which one worldview wins over the other, although I am hoping that the two will engage with, and mould, each other to evolve our society towards a better, more equitable place for all.
Hopefully this article has given you some insight into who the main players are – the communities weighing in on the discussions and the individuals leading the conversations. South African politics are more vital than they have been in years and much of it plays out on Twitter. The 2016 state of the nation address is around the corner. I can’t wait to see how the South African Twittersphere reacts to events in parliament on the day. We live in interesting times and Twitter allows us to track history as it unfolds giving us insight into the seeming chaos around us. DM
Daily Maverick © All rights reserved