Information Technology in the 2013 Kenyan national election

By the time you read this, I expect that the Independent Elections and Boundaries Commission will have announced that Uhuru Kenyatta won the 2013 national election. While there are other possibilities, including a run-off, that one seems most likely. The election has been controversial, divisive, newsworthy, and often confusing. It’s time now not only to congratulate the Kenyan nation, but begin to review the role of information technology (IT) in its realization.

In principle, the election was fully computerized. While I don’t yet have access to all the technical details, the popular press described a plan for biometric authentication at all polling places, centralized aggregation, and transport in at least some segments by the mobile telephone network. The IEBC itself exposes a nicely-documented application programming interface.

Plans didn’t go smoothly. In the day or two of and after the election on Monday, 4 March 2013, something, and maybe, as so often is the case in real-world systems, several things, broke down. There was talk of hackers, miscalculations, networking problems, and more. Much of the social value of the IT systems was to control the vulnerability to manipulation which made the 2007 presidential election such a tragedy. The technical results were, at best, disappointing.

The IEBC, courts, electorate, and even the major contending parties seem to be succeeding, though, at creating a credible, and even remarkably transparent adjudication and transition.

Until more certain information about the computing systems emerges, most of what has been written remains speculation. It might well be, for instance, that claims “the computer” mistotaled voided ballots by a factor of eight are in fact arithmetic confusions about how the election rules apply in a race with eight candidates.

Electoral computing is a hard problem, though. Kenya set higher standards for itself than countries, such as the United States and Europe, with nominal per capita economic resources ten or thirty times as great. Kenya has a history of tumultuous elections, and its IT design for this one was ambitious. Even from the little that’s already public knowledge, plenty of positive lessons have emerged:

  • Civic society is not 19th-century biology, and ontogeny need not recapitulate phylogeny. Just as Kenya’s economy isn’t bothering to reproduce the old banking or telephone infrastructure of, say, the United States, it can also choose to jump ahead of the “developed” countries in at least parts of its election procedures.
  • Management matters: strong IEBC leadership means that, however the servers were configured, what people will remember from this episode was a peaceful and therefore successful election. IT systems are subordinate to the human-level workflows they support.
  • “Developed” countries can learn from the experience of Kenya, India, and other early adopters of biometric and other innovative authentication methods.

There will be opportunity later to review the low-level technical details of Kenya’s 2013 election. Now is the time for unity and recovery. My favorite summary appeared before the day of the election, when Elko Namlo wrote of “… growing concern that the demand for clich├ęs is outstripping supply”, and foreign journalists would have too little photogenic violence to report.