Posted by Paul Fidalgo on August 18, 2009
Our senior analyst Terry Bouricius has just authored a report on the accuracy of the ballot counting in Aspen's municipal elections this year. You can see the whole report here (it's pretty short, actually), and here is our press release on Terry's findings.
* * *
Report: Widely Used Voting Machine Missed 0.4% of Ballots
Aspen (CO) Analysis Illustrates Need for Redundant Vote Counts and TransparencyProving the value of transparency and redundancy in ballot-counting, an independent rescanning of ballots cast in the May 5, 2009 elections in Aspen (CO) showed that the voting machines used for the initial vote count entirely missed 11 (0.4%) out of the 2,544 ballots cast. The ballots were initially counted with Pitkin County's Premier (formerly Diebold) AccuVote optical scan voting machines, one of the most widely used optical scanning systems in the country. The error was discovered when TrueBallot, Inc., employed under contract with the city of Aspen to re-tally the elections at a central location, rescanned all ballots using a commercial off-the-shelf scanner and discovered the erroneously disregarded 11 ballots.
FairVote recommends that measures be taken to ensure that the errors of the Premier machines are not repeated. "We have seen that when elections are tabulated using non-proprietary, commercially available systems and equipment, the results have proven to be more verifiable than with closed, proprietary systems," said FairVote senior analyst Terrill Bouricius, author of a report on the findings. The report also notes that when the Humboldt County (CA) Election Transparency Project conducted a re-scan of AccuVote ballots using commercial scanners in 2008, 197 dropped ballots were discovered. "It is troubling to know that errors by machines made by one of the largest providers of voting equipment in the United States could go undetected because there is typically no transparent mechanism used to recheck the results. We need to build redundancy into our election tallies."
Bouricius's report includes specific policy recommendations for the city of Aspen as well as broader lessons learned, applicable to elections across the U.S., regarding the recording of individual ballot records, manual audits and public release of raw data from elections to allow independent tallying of the results. Notably, the election in question included Aspen's first use of instant runoff voting, a ranked voting system, which did not in any way contribute to any of the errors discovered, but did lead the city to do the central scanning that uncovered the error.