ORG's report on the May 2008 elections in London has just been published. (Thanks for the alert, Glyn). It comes at a timely moment for me, since I'm scheduled to draft a case study on evoting at some stage in the next few weeks, for a forthcoming Open University course on ICT. I'm also hoping to get together the main framework of a response to the government's consultation, nominally on choice of election day. But buried deep within said consultation are apparently throw away questions on evoting:
"Question 4 Do you think that greater access to remote voting (whether through traditional postal voting or by electronic means) should be made available alongside weekend voting? Should such arrangements be explored even if polling day were not moved to the weekend? Please explain why.
Question 5 What do you perceive to be the benefits and the drawbacks of remote e-voting?"
The ORG report on London is a clinical indictment of the reality of using evoting in a live election, even when the project managers of that election do a commendable job within the constraints that they are operating to, and it concludes that:
"There is insufficient evidence available to let independent observers reliably state whether the results declared in the May 2008 elections for the Mayor of London and the London Assembly are an accurate representation of voters‘ intentions."
The good folks at ORG also make a number of recommendations:
ORG‘s position is that e-counting obscures the workings of elections from voters and candidates. Mitigating this risk in order to sufficiently enhance the transparency of e-counts could well be more expensive than sticking with manual methods. ORG has received comments that suggest that e-counting is inevitable and that opposing these technologies is a Luddite view. ORG disagrees, and considers it telling that a significant proportion of those concerned about voting and counting technologies are computer scientists and professionals, who are normally the most enthusiastic adopters of new technology.
The political climate is still in the shadows of the chaotic May 2007 e-count in Scotland, and the electoral timetable is likely to preclude the deployment of computers in elections for the next two years. For the moment, therefore, ORG recognises that elections administrators may be turning away from experimenting with e-counting technologies in statutory elections. However, ORG suspects that in two years' time these deterrents may have faded and legislators may feel eager to experiment with e-counting again. ORG therefore makes the following recommendations for improved practice in e-counting below, and refers any legislators tempted to reopen the Pandora‘s box of e-voting to the conclusions and recommendations of ORG‘s May 2007 elections report.
Recommendation 1: A full cost-benefit analysis of electronic counting at the London elections in May 2008 should be performed by London Elects. The analysis should be set against a properly-costed manual count of similar scope. London Elects should also cost the following recommended enhancements to the electronic count, including:
o A statistically significant live manual audit on count day, or some other effective means, accessible to the layperson, of monitoring votes that are counted as valid
o A comprehensive, independent audit of all source code deployed on e-counting systems, made publicly available before the elections
o Improved record-keeping facilities at the ballot box verification stage
o Improved transparency around the contractor‘s service management desk
o System-designed assurance that the voter‘s paper ballot remains the ballot of record so that, for example, paper ballots can easily be retrieved by CROs wishing to ascertain the intention of a voter where this is not clear from the scanned image of a ballot.
Recommendation 2: Time should be given for formal consultation — at national and local levels — prior to the approval of e-counting being used in an election.
Recommendation 3: Administrators should remain committed to long lead-in times for procurement and implementation of election technology. Based on the experience of London Elects, ORG revises this figure upwards from one year (as recommended in ORG‘s May 2007 report) to 18 months as a suitable application and implementation timetable."