EVT/WOTE Conference, Montreal

8/19/2009:  Back from a week away from the Internet, I’ve updated the links below and highly recommend the sensation of the conference, the debut of “Plaudits for Audits”:

<video and lyrics>

Friends, raise your joyful plaudits to post-election audits
Where we count some votes by hand to check the work of the machines.
It might sound esoteric, or tiresomely numeric
But democracy’s at stake, so let’s make sure those counts are clean.


Monday and Tuesday, the EVT/WOTE Conference was held in Montreal.  This tends to be a highly technical conference on potentential voting technologies, security and vulnerabilities in current technology, and related projects.  For me it is a mixture of new information relevant to voting, interesting technical articles, a time to reflect, and to connect with others involved in causing voting integrity.

Joe Hall and Ben Adida have provide photos and summaries of day 1.  You can see me in the white striped shirt in the front row in the 1st picture from Joe. <Joe Day 1>. <Joe Day 2>  Ben covers a bit more of the details in his two posts <Day 1 AM> <Day 1 PM> <Day 2>  (I’ll update this list with day 2 reviews when they become available)

All the papers are now available from the USENIX site <here>

There were several highlights for me, several of which will provide fodder for more extensive posts in the near future:

  • Larry Norden highlighted the issues associated with voter registration systems and the potential for improving this area.  Our current systems are expensive,  error prone, and disenfranchising.  Surprisingly voter registration is more than half the cost of election administration – more than all the other costs combined:  Equipment, training, ballot printing, auditing, and election day activities.
  • One theme was the potential for cryptography to provide secure and auditable elections without paper records.  Here I was surprised at all the activity and claimed potential.  I am open but not yet convinced.  Proponents claim it can be done, others point out challenges still to be addressed.  There would be a lot of reading for me to follow all the details of the proposed processes to begin to understand the potential risks and value.  Perhaps in a few years we will have an agreed upon alternative to paper ballots.
  • Representative Rush Holt made the case for his bill.  He is one of four or five scientists in the Congress (he has a Ph. D. in Astro Physics).  Several years ago Congress eliminated its Office of Technology Assessment which provided analysed technology independently for the Congress.  He pointed out that the OTA might have avoided some of the problems with HAVA and with bio fuels.  It says to me that most members of congress are clueless that they are clueless about the need for scientific analysis.
  • Much has been made of a paper on the potential to hack voting machines that use return-oriented programming <read>.  While technically significant, to me it is just another confirmation that its almost impossible to trust software – we must assume that any system can be hacked.
  • Last on the aganda, but not least a paper by Joe Hall and several others reviewing “Risk Limiting” audits <read>.  As an attendee I have had access to this paper for several weeks and have been a party to several lively discussions of its implications and conclusions.  It essentially, successfully challenges the assumptions behind previous papers on risk limiting audits and calls for much more rigirous statistical methods of analysis if audits are to claim exact levels of statistical confidence in election integrity.  I will have much more to say on this in the future.  Sufice for now to say that we are are on much more solid ground proposing post-eleciton audit laws to select ballots or districts based on specific fixed audit percentages or tiered audit percentages than laws specifying statistical confidence levels – unless and until statisticans can agree on how to compute confidence levels and realistic levels ballot counting.

Leave a Reply

You must be logged in to post a comment.