A Robot Watched Raila’s Debate Performance, Here’s What It Found

From our previous analyses of the 2017 Kenyan Presidential debate, it was abundantly clear how big a deal President Kenyatta’s absence at the debate was, with many Kenyans left disappointed about it. However, there was also a lot of discussion about NASA presidential candidate, Raila Odinga who showed up for the debate solo. Kenyans from both sides of the political divide of course went ahead to have their own debate about his performance. As humans we possess inherent biases around almost any topic and that’s why two Kenyan voters can watch the same debate performance and come out of it with extremely different opinions of how Raila handled himself and the questions that were asked of him.

But what if we could get an unbiased view of how Raila handled the debate? What if we could bring in an impartial party who could look at and derive conclusions from the data, and nothing but the data? We have attempted to tackle this question by making a computer watch the debate using our proprietary video analytics technology. Our robot watched the debate, capturing a video frame at periodic intervals and matching the faces to transcripts of the debate to infer emotion. The robot uses computer vision technology executed on a series of high speed shot frames to categorize emotions of the subjects in the frame either as sad, happy, angry, neutral (stone face or no emotion), contemptuous, surprised or disgusted.

Below is a visualization of what our robot returned after its analysis. We have collectively categorized all emotions except happiness and neutral as negative emotions.

But before you look at the results, take a minute and consider, where do you think his weak points and strong points were?

As the graph indicates, our algorithm recognized a lot of happiness from Raila from the very beginning of the debate, perhaps joyous his opponent not showing up may mean an easier debate. Some of Raila’s happiest moments during the debate included:

  • “12: Asked where Kenya stands regionally
  • “45: Asked about statement that could have threatened national unity mainly targeting people from outside Kajiado
  • “57: Asked about Kenyans generally voting along tribal lines
  • “63: Asked about how he would handle the corruption in his government
  • “90: Asked about the name calling that goes on between the presidential candidates

There were also several moments where Raila got rattled in his responses and this seemed to elicit negative emotions from him:

  • “18: Criticism of national institutions namely the IEBC and the use of the military in trying to influence the election
  • “36: Asked about a meeting in 2008 about importation of maize
  • “60: Asked about tribe-inclined decisions such as people winning nominations with his direct endorsement and hiring practices during his tenure as Prime Minister
  • “72: Asked about the issues he might face in his time in office
  • “84: Asked about the Lamu Coal Plant project and its effects on the environment

From the data we can see that Raila maintained an overall upbeat mood throughout the debate. Whether he performed strongly or not is a matter of your opinion, which of course brings us back to our partisan prejudices. See you on election day :)


While we used this technology on the debates we believe its primary use lies in Market research applications such as ad testing, helping us deliver valuable insight as to what emotions people go through at the point of experience. Our software can currently identify Gender, emotions and age(currently in beta testing). If you are interested in working with us contact us at team@odipodev.com for more information.