Faculty Perspective: Technology in Kenyan Elections and Lessons for Moving Forward

Dr. AdomdzaKenya's Supreme Court annulled the nation's 2017 election results, setting a historic precedent for democracies in Africa. Ashesi lecturer, Dr. Gordon Adomdza, reflects on the tech problems that led to the annulment, and lessons to bear in mind moving forward.  

This is an interesting case to reflect on, because the voting was largely peaceful. The Independent Electoral and Boundaries Commission (IEBC), relying on digitally collated results, announced Uhuru Kenyatta as the winner by a margin of 1.4 million votes. In response, opposition leader Mr. Odinga alleged that IEBC’s systems were hacked using passwords stolen from the murdered commission's IT manager, Chris Msando. 

To properly collate all the results, digital data was normally verified against hand-collected data – on Forms 34A - which were filled by presiding officers to tally votes at polling stations. However, it appears that there were some transmission problems. Some one in four polling stations didn’t have strong mobile phone coverage, while in some areas, there was no mobile data coverage at all; all of this delayed the transmission of the collated results. 

So, perhaps it is not surprising that in annulling the results, Chief Justice David Maraga said the IEBC “had committed irregularities in the transmission of results". The delay in transmission may have led to the irregularities alluded to. This was not the first time Kenya’s election body had faced accusations of flaws in its transmission system and technology.

What these developments point to is that technology has become very important in elections across parts of the continent; however, it must be accompanied by well-planned user and systems behavior. For the next round, the IEBC could consider adopting some design processes. They could develop the step by step journey of a collated data from polling station as it travels to the national collation center, for example (essentially, the journey map of a collated data point). Mapping the journey out will enable IEBC to identify all the loopholes in the transmission process and fix them.

User behavior is also essential for the success of processes and systems. Hence, the IEBC can work with the relevant stakeholders such as polling agents of the different parties, to map out their mental and operational space in order to understand and anticipate potential positive and negative behaviors and respond to them appropriately.

Attention could also to be paid to behaviors that do not depend on technology. Polling officers may have the technology available, but can fail to use it professionally. Anticipation of such behaviors will be the first step to preventing them and potentially dealing with them. Most system planners will do these analyses, but most of them fail to visualize their analyses in order to more effectively capture all potential loopholes and close them. 

A few years ago, Mr. Kenyatta declared himself a “digital president”, casting Mr. Odinga as ‘analogue’. In fact, the president is said to have employed the services of Cambridge Analytica to provide market research and consulting on how to win elections in a country where 88% of the population have access to the Internet through their phones. With that level of Internet penetration, Kenyan citizens themselves can track, monitor and provide collation at the polling center level alongside the processes run by the IEBC. In Ghana, we saw radio stations and citizens play such a role by providing real-time information on collated data and also displaying it on their websites and social media platforms, making it almost impossible to tamper with the results. Such an effort will ensure that the IEBC data is socially-verified when it is announced.

If neither side boycotts the election (as Mr. Odinga has indicated he may), the next round of voting can still depend on technology to be effective. However, due to the vulnerabilities of digital data today, and the present mistrust the electorate has in electoral systems, collation will still have to involve analogue preservation of the data. After all, we have just witnessed the analogue man stop the digital man in his tracks, even if temporarily.