How Technology Failed to Fix Kenya’s Election

A Kenyan voter holds a presidential ballot at a polling station in 2013.

A Kenyan voter holds a presidential ballot at a polling station in 2013. Jerome Delay/AP file photo

While the 2013 elections were largely violence free, that was no thanks to technology.

It’s perhaps the most common story in all of government technology: A challenge arises; new technology seems to offer the perfect solution; but something happens between concept and execution that makes that technology seem more like a culprit than a savior and that leads people to complain the old analog solution might have worked better.

That interference could come from a delayed procurement, miscommunication between different vendors, a lack of testing or training before launch or a host of other factors.

This December 2013 report from the U.S. Agency for International Development describes more than a dozen such interferences that foiled the international community’s attempts to use technology to improve outcomes in Kenya’s March 2013 elections.

The international community was deeply concerned those elections would spark a return to the mass protests and ethnic violence that gripped Kenya following its disputed presidential election in 2007. That violence did not materialize despite allegations of voter fraud and a result that second-place finisher Raila Odinga challenged before Kenya’s Supreme Court.

The technological interventions aimed at securing that peaceful outcome were largely a mess, however, as outlined by the report titled USAID Support for Kenya’s 2013 Elections: Rapid Assessment Review.

(Full disclosure: One of the report’s co-authors, Johanna Wilkie, is a friend and former graduate school classmate of mine).

A nationwide system to link biometric data with voter registrations was plagued by corruption allegations, canceled and then reinstated, the report said, which delayed multiple other technology procurements.

That biometric registration system worked well on its own merits on Election Day, but a corollary system to match biometric registration data with people who actually showed up at the polls “failed on a massive scale,” the report found.

A separate system designed to rapidly transmit ballot results from polling locations to the public and the media -- a system many considered vital for reducing electoral uncertainty and mitigating the chances of post-electoral violence -- also failed on a large scale, a result of lengthy procurements, poor training and miscommunication, the report found.

“Technology is a tool, not a panacea,” the report stated in a list of recommendations. “Serious cost-benefit and feasibility analyses should be undertaken before committing to support new technology; local low-tech solutions often may be the most appropriate option.”

The full report is available here