Scaling non-linearly in your mind
Twitter’s response to #OccupyLuthuliHouse

Twitter’s response to #OccupyLuthuliHouse

The #OccupyLuthuliHouse protests this past week were perhaps the most visible manifestation yet of the split within the ANC between President Jacob Zuma’s powerbase and those who believe that he is steering the party (and the country) in the wrong direction.

I started collecting tweets on the day of the protests and manged to amass 30,551 tweets generated by 10,317 different Twitter users. These volumes made it a pretty mid-sized “social event” on Twitter compared to some of the other datasets that I’ve collected. So, from a Twitter perspective, the protest was a marginal success. It definitely generated buzz but not nearly as much as some other events, such as #FeesMustFall or pretty much any of President Zuma’s individual scandals (including various Constitutional Court hearings and similar). Part of this might be due to the fact that the ANC is often under-represented on Twitter compared to the DA, the EFF and Woke Twitter who often take to the platform to air their grievances against the status quo (i.e. the ANC-led government). Since this protest came from within the ANC itself, it might not have had the same depth of pre-existing social network inter-connectivity to tap into in order to boost the presence of this event on Twitter.

To set the context, here’s a look at the hourly tweet volumes that I collected by searching for terms like “#OccupyLuthuliHouse” and “Luthuli House” during the period of the protest:

2016-09-10-luthuli-house-01The question that immediately arises is what proportion of the conversation was in support of the ANC status quo (i.e. President Zuma and his current administration) and what proportion supported the protesters’ call for change? We can get a very rough idea of the volumes of for and against conversations without delving into a deep analysis of the contents of each tweet by looking at how often the two most popular hashtags, #OccupyLuthuliHouse and #DefendLuthuliHouse, appear in the data. As the chart below highlights, usage of the #OccupyLuthuliHouse hashtag dramatically outweighed its counterpart. Note that a couple of important caveats to this simplistic analysis apply: 1) both for and against Twitter users might have used the #OccupyLuthuliHouse as a common way of labeling their conversations, and 2) I did not explicitly search for #DefendLuthuliHouse in isolation so this count does not include tweets that might have used the hashtag without also mentioning Luthuli House by name so its presence in the data is likely to be somewhat under-represented. [EDIT: I somewhat tweaked the wording of this paragraph to take into account how the data was collected]


When I map the interaction network of users @mentioning and retweeting each other, we get the below picture (I haven’t included any usernames in this network map for privacy reasons). What’s notable about it for me is that it’s very diverse with little clear community structure. I generally expect to see a DA-liberal community, a mainstream media community, a more decentralised, socialist leaning Woke community, and an EFF community (often straddling the line between mainstream media and Woke communities) dominating in South African datasets of this type. In this case though, the communities are far more diverse with none dominating the narrative, which makes it difficult to tease apart the main players. The EFF is in there and I think that much of Woke Twitter is too, as are the various media houses in a fractured form. What also stands out though is the lack of a clear ANC community, perhaps for the reasons that I mentioned at the beginning of this post. Please also keep in mind that these are my own quick descriptions for each community and should be taken with a pinch of salt given the brevity of this analysis.


This chart shows the above communities in another way by plotting the proportion of users in each community versus the proportion of tweets that each community generated. We’d naively expect a community to generate tweets in proportion to the number of members that they have and thus fall onto the 45 degree diagonal line in the chart. Where a community was particularly vocal and generated more tweets than expected, their bubble appears above the line, and vice versa. From this chart we can see that the EFF & POWER987 community (POWER987 seems to be their preferred news source) was very vocal as usual, as was the main Protest Supporters community in pink. For the remaining vocal communities (which I have given names based on the prevalence of particular news houses’ tweets within each community), it’s probably a fair assumption to assume that they were similarly passionate about the protests.


Finally, let’s have a look at the most popular tweets overall within this dataset.  I’m not going to split them up by community in this post as I have in the past mainly because I’ve already spent enough of my Saturday morning writing this 🙂

So, in summary, what can we take out of this dataset? Well, many things I’m sure, but I’ve only had the time to write up a surface-level analysis here. What stands out for me is that there was overwhelming support for the protests from most of the communities and constituencies that we’ve touched on above. In addition, there appeared to be a lack of a cohesive pro-ANC counter-narrative and, to the extent that it was able to coalesce around the #DefendLuthuliHouse hashtag, their comments were clearly drowned out by commenters supporting the protests.

Twitter isn’t the real world. It’s not representative, but my years of researching it have lead me to believe that it plays an important role as a barometer of the direction that the broader public sentiment is leaning and the winds are clearly blowing against President Zuma’s administration (as we’ve known for a long time now). This was just the latest manifestation, except this time it was different because the call came from within the ANC itself and the usual vocal Twitter voices were forced to sit on the sidelines and watch things happen…

Re-blogs of this post: BizNews | iAfrikan

[Banner image source]

Leave a Reply

Your email address will not be published. Required fields are marked *