Technology, Transparency and Tallying

-->

Edge of a computer screen

Two things stand out in the Kenyan General Election of 2013:  IEBC’s late procurement of both services and equipment related to the election, and the fact that all technology should have been tested and debugged far in advance of the election. The failures that occurred were both foreseeable and preventable. Of note was the failure to plan for backup power. Electricity fails routinely in Nairobi, and is often absent in rural areas altogether. In addition, the cell phones and biometric scanners were not procured until approximately one month before the election, and were most likely not tested sufficiently for either load or other stresses.  Professor Makau Mutua asks whether the collapse of the computer systems during vote tallying was due to incompetence, technological illiteracy, or lack of adequate preparation. The most likely answer is a lack of adequate preparation, combined with a failure to follow good advice. 

Despite the massive technological failures, it can be argued that their main impact was a significant delay in reporting the results, not the integrity of the election itself. Importantly, the physical count of votes was the final and official record of the election. The manual voter register worked well to identify voters at the polling station level. No vote count was finalized at the polling station level without agreement of the presiding officer and political party agents. This process was repeated again as all presiding officers reported their numbers to the reporting officer in full view of political party agents and observers at the constituency level. All marked and unused ballots were locked into the transparent tally boxes with final numbers. Those boxes were tracked from polling station level to the constituency level, and eventually flown to Bomas to ensure that the final vote was correct.

The idea that manual ballots trump electronic systems is widely accepted internationally. A symposium on Voting, Vote Capture, and Vote Counting was held at the Harvard Kennedy School of Government in June 2004 in the wake of the counting failures in the Bush vs. Gore presidential race in the US. At that event scholars indicated that best practices for a secure voting system include a hybrid system that includes paper for audit and an electronic system for speed and flexibility. Ironically, their study noted that digital systems can actually produce more complex failure modes and concluded that paper ballots, carefully tracked through a custody chain remain necessary to ensure accurate voting outcomes. The Kenyan election of 2013 illustrates those conclusions well.

A long day at Bomas waiting for Kenyan election results

I spent a few hours at Bomas of Kenya today, which is where the tallying center for the General Election Kenya 2013 is.

The Bomas tallying center was very quiet. In the time I was there, they only announced results of two constituencies: Rongo (in Nyanza- a Raila stronghold) and Undanyi (on the Coast-voting is mixed). The depressing thing is that hours after I returned home, the news is still reporting these same results, with the addition of Kajiado Central (where Raila won)

My understanding is that the electronic machinery at Bomas really stopped working, so now they are going back to manually verifying the tallies.

The downside is that the technology failed. Another huge downside is that the delay is painful and may be causing wananchi (citizens) to panic. Rumors are flying on facebook.

The upside, and the big upside, in my view, is that the election remains peaceful. People are going to work. Streets are not empty, but they are not full eitheri. The manual technology works perfectly well. And from what I saw, the procedures implemented make rigging extremely difficult.

So, I prefer to see the glass half full.

Results have changed little from my last report. Due to the manual tally occuring in Bomas, this process could be slow. Based on what I have seen, I think they have reports of all the votes that have been cast. What is likely delaying the process is that agents must agree on the vote before they release it to the press.

At this rate, this vote counting could take a week.

I am going to be boarding a plane in a few hours, but I will keep posting my analysis.

~WMB