So with all the recent news on Kenya’s elections and subsequent (concurrent?) riots, it crosses my mind that there is a great need for transparent+verifiable, reliable, and low-cost voting systems in developing countries. I know that a lot of people at UC Berkeley and other universities are working on e-voting, both on the implementation/computer science side, and on the policy side. But of course much of this work is targeted towards federal United States voting requirements, which entail all sorts of things like audit trails, voter anonymity (both to protect individuals, and to prevent them from selling votes), as well as probably lots more esoteric accessibility requirements, not to mention our weird electoral system and lots of absentee ballots.
Imagine if Diebold or Pitney-Bowes came up with a low cost electronic voting system that could not be compromised (okay, given, that’s still a hard problem) and could be used by governments of developing countries to hold reliable elections.
In Kenya the results were delayed for several days as the voting was completed. In the meantime Kenyans rioted. Then the results were announced – the incumbent President Kibaki had won. And then the EU Electoral commission declared dissatisfaction with the vote tallying process: ““We believe that… the ECK… has not succeeded in establishing the credibility of the tallying process to the satisfaction of all parties and candidates,” chief EU observer Alexander Lambsdorff said in a statement.“. More Kenyans rioted – and now the latest news report says 30 have burned to death in a church set on flame by a mob.
[an aside: i imagine americans in a similar situation might object, but would be too apathetic to do anything as energetic as rioting. perhaps it has to do with our low voter turnout... and i guess al gore did the nice thing and conceded rather than waving a banner and accusing the bush family of voter fraud]
So problem #1: it takes a long time to manually count votes. And problem #2: auditing capabilities are limited – although some auditing can be done (apparently one district reported 110% turnout), only the incumbent party gets a useful “assurance” of the tally’s credibility. Both of these problems seem easily solvable by information technologies. For example one could use DTN+Tierstore to collect the votes and publicly publish them, decentralizing the counting and publication of the votes such that an attacker (in the Bruce Schneier technical sense) would have to compromise many units. But to prevent ballot stuffing, each vote would need to be accompanied by an individual identifier like a thumbprint (is it possible to distinguish a thumbprint from other fingerprints?) or signature, which could also be published on a per-district level, disassociated from the voter tally of the district. Use of computing equipment also assures the speed of tallying so the result could take hours instead of days.
But however, feasible/interesting this may seem as a research project, I anticipate a few possible glitches. Firstly, one would have to collaborate with the government, which may not actually want fair elections. Then (as a PhD student) one would have to be quite lucky to have the election during the course of the PhD, and really, aside from testing, they would have only one try per country every 4 (5? 10?) years. And there’s no guarantee that even with a verifiable and transparent election people won’t riot anyways. Although it seems that election fraud is the cause of this spate of violence, I would surmise to guess that a key underlying cause would the ethnic division, and the corrupt elections were just an excuse for such outward displays of frustration, angst, and wanton destruction. Um. And I wouldn’t want to be in Kenya during that disputed election period… But I think it’s still perfectly feasible to 1) do research on election systems in developing countries and 2) design a novel software+hardware solution motivated by an understanding of election systems in these countries, along with a model for how this solution could be adopted by various countrie, all without doing a full-scale deployment.
So if anyone is looking for a research topic… take this one and run with it.