Scaling non-linearly in your mind
Quantifying the spread of EFF & RET disinformation

Quantifying the spread of EFF & RET disinformation

I worked with Ferial Haffajee at the Daily Maverick to quantify and visualise how EFF & RET talking points, often grounded in heresy, permeated discussions around Pravin Gordhan. You can read the full article here.

It included images like these:


  1. Izak Marais


    I found your website via the Daily Maverick article. As someone who has an interest in both politics and data science, I was immediately drawn to it. The article was informative and the visualisations where interesting. However, I think the conclusions drawn from the visualisations could have been more clearly explained.

    For example, the first graph is introduced with “What is immediately clear is that there are three main groups (representing 67% of all users in this data) generating most of the discussions around Pravin Gordhan…”. Perhaps this is immediately clear to someone well versed in reading dense twitter graphs (or viewing the original high quality versions of the graphs) , but I did not find it clear based on the published image. Maybe it is the choices of colours used here (compared to some of your other analyses)… I struggled to discern the red from the grey, let alone clearly identifying three groups. Is there a third colour I am missing apart form red and grey? I also struggled to see much difference between the three different graphs contained in the article. The end result is that the conclusions drawn seemed a bit like tea leaf reading. This is not to say that I think they were incorrect, just that, if I were reviewing this as a peer review article, I would recommend the graphs be clarified a bit more and more room be given to explaining the interpretations.

    Maybe something was lost in translation between the explanation you gave and the article Ferial wrote. Or maybe I am expecting too much scientific clarity from a news site article.

    Sorry to sound overly critical; I already browsed your blog archive and am definitely going to subscribe to your blog. I look forward to future posts.


    1. Superlinear

      Hey Izak. Thanks for the detailed feedback. I agree that things weren’t that obvious in the final article. Unfortunately I had no control over that. My comments that you quote originally referred to a version of that network that showed the community breakdown. This graph showed the clear communities but it wasn’t included in the final article. I try to always explain the data visualisation the first time that it is shown in an article but, again, I had no control of what was finally shown this time around. Thanks!

Leave a Reply

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