CTVotersCount Visits the UConn VoTeR Center

On June 4th we visited Dr. Alex Shvartsman and his team at the UConn VoTeR Center, toured the lab, and observed the testing of memory cards and the Diebold AccuVote-OS. Dr. Shvartsman and some of the staff

We have applauded the VoTeR Center’s academic work on evaluating the security of the AccuVote-OS and other voting machines <read> <read> along with their programs to independently inspect memory cards. <read> <read>. We have have also highlighted the Center’s post-election audit reports with some criticism based not on UConn’s work but on the lack of our confidence in the integrity of the data provided <read>

UConn’s reports on Diebold vulnerabilities complemented those of other researchers around the country, confirming the same vulnerabilities and adding additional vulnerabilities. The computer program to independently test memory cards represents excellent work. We are aware of no other researchers and no others state with a similar capability. This makes possible the unique program in Connecticut to test memory cards. We would like to see the program expanded to pre-test 100% of our memory cards before each election prior to delivery to election officials. The testing continues to highlight quality control problems in the programming of the cards by LHS Associates, and the failure of election officials to rigorously follow pre-election testing procedures.

In the lab we saw the equipment the team attaches to a modified AccuVote-OS machine that allows a variety of tests and experiments. By adding some code to the AccuVote-OS a copy of all the data on a memory card can be loaded onto an attached computer for analysis. For instance the post-election data on the card can be compared to the data originally used to load the card to make sure the election definition was not changed before or after the election. Pre and Post-election testing can be performed to make sure no stray data and no unauthorized program is on the card (There is a small report generating program on each card – it should be the same on every card – fortunately it is a very short program and can be evaluated for risks by the VoTeR Center with a very high degree of confidence)

We learned from Dr. Shvartsman that the recent Presidential Primary post-election memory card testing report will be completed shortly. We understand it will indicate problems similar to those reported from the November election. In addition testing revealed instances of a duplicte memory card being copied from another card which were not expected. While we were in the lab the team brought up that memory card log and we could see that the log indicated that the copying occurred on the same day as pre-election testing – even though it is possible to change the date and time on the AccuVote, the log is maintained in order so it is quite likely that the copying was done at that time rather than when the cards were originally prepared. (We look forward to the full, official memory card report and more details on this copying) .

Another capability of the setup is to scan a ballot and have a list of votes counted displayed on the attached monitor. We discussed the issue of ballots with questionable markings and confusion amoug election officials concerning which marks the scanner will normally read. The team quickly performed a test where we made small marks withing an oval, testing a ballot untill the vote was counted. In our limited test, a small dot was not sufficient, but a small vertical line either horizontal or vertical was sufficient to be detected and counted. This result was consistent with coalition audit observations and audit statistics – that most marks by voters within or through the ovals are counted.


Leave a Reply

You must be logged in to post a comment.