"Open Source" used in VT election, Cuyahoga fudges '04 recount
Like Ping, I also finished a submission to EVT on Monday evening... and my hands (RSI) still hurt! Naveen has said that all the rage over in CS are the Kinesis keyboards which have little "bowls" that your hands sit in; I'll have to save some money to get one of those (wouldn't it be nice if Berkeley had an RSI-prevention equipment matching program!).
I realized one thing while penning the 16-pager I submitted to EVT: in interdisciplinary forums, those who have citation-heavy writing styles actually have less space in which to write. I tend to write in a legal style where every single assertion must be cited to a source via a footnote. This results in some of my "pages" being half footnotes, just like a law review article. I guess, we should either develop some sort of citation to text volume metric and specify that you get more pages if you have more citations... or, I suppose, I could learn to write differently!!!
Anyway, in other voting news...
"Open Source" used in mayoral election in VT?
As your resident open source geek and researcher, I'd like to point out a curious article that came past my screen recently:
Successful Public Election Joins Diebold, Free [as in beer] Software
http://www.govtech.net/news/news.php?id=99062
It appears that Burlington, VT, used DESI's AccuVote-TS terminals to record votes in a mayoral race and then used a disclosed source[1] application developed by Voting Solutions to do the calculations necessary for instant runoff voting (IRV). The code, called ChoicePlus, was disclosed and posted for review along with vote data (the code includes a proprietary Borland library which is licensed under terms that would preclude releasing the code under an OSI-approved license).
It appears that they used the resident DESI code to record the votes (and I'm not sure how if DESI's code doesn't support IRV in the interface and data structure) and then this guy's code to do the tallying of the IRV votes.
Of course, there's little chance a vendor or election official could do this in any of the 39 or so states that require federal certification as the Technical Data Package and such that would have to be submitted to the testing authority is highly proprietary; that is, they couldn't get federal certification for this kind of set up as I seriously doubt that DESI would cooperate enough such that all the information and accouterments would be in place to get certification.
[1] This is different from "open source"... I can elaborate.
Cuyahoga county did not follow post-election audit procedure
David Wagner sent the following article out:
Workers accused of fudging 04 recount http://www.cleveland.com/election/plaindealer/index.ssf?/base/isele/114432450773490.xml&coll=2
While there doesn't appear to be any evidence that these election workers intended to change the outcome of the election, they deliberately side-stepped post-election manual audit procedures. This is very important with computerized election machinery as frequently the only way you can tell that there has been some sort of malfunction or malfeasance is by comparing the machine count to a separate count. With VVPATs, this is especially important because any error could either be in the recording of the VVPAT or the electronic record. The fact that the election workers pre-selected the polling places to recount based on the fact that they wouldn't be discrepant is very disturbing. It also unfortunate that these election workers now face both misdemeanor and felony counts related to not following state election law to avoid the additional work that would have been required. Of course, if elections and election administration was properly funded, they would have had the staff to make following the law a cinch.