Connecticut Performs Pre-election Audit Of Memory Cards
In what may be a program unique to Connecticut, UConn ‘s VoTeR Center performs a pre-election audit of the memory cards for the state’s Premier AccuVote Optical-Scan tabulators. This is a smart program that should be mirrored across the nation.
The University just released a report of their audit of the memory cards used in the state’s August primary election. The cards were programmed by LHS Associates which also provides voting systems, programs memory cards, and does other election work for the states of Connecticut, Maine, Massachusetts, New Hampshire and Vermont. UConn received 185 memory cards directly from LHS. Ten of those cards or 5.4% had nothing but “noise” on them. How many bad cards were sent out to polling places in CT? How many bad cards were sent to polling places in the other states LHS services? ...
http://waronyou.com/forums/index.php?topic=1531.msg4135;topi...
It's not the cards that are the main problem
It's the central tabulator and the procedures for auditing the count.
If there is a paper trail of the actual votes, then the cards can truly be audited, else the whole 'audit' is a sham.
If the vote can be edited at the central tabulator, it matters not, if the individual precincts didn't printout and seal their vote counts. (along with hard copies of the individual votes to check against the printout totals)
If no one but the central tabulator can reproduce the vote from anything but data blips, then they can make up whatever vote results they want.
There are four, and only four rules absolutely necessary to secure the vote:
1)hard copy verifiable printout given to each voter, who then stuffs into a ballot box.
2)printout with cross check hand count certification at each precinct in full public view conducted by poll workers.
3)central tabulator that is NOT EDITABLE by ANY MEANS. This includes no ability to edit final results as well as individual memory cards. The central tabulator MUST be hooked up online with real time internet posted updates, as well as hard copy printouts which will be certified and verified against hand counted totals from individual precincts. (as the central tabulator plugs in a memory card, the update appears on screen as well as online, ASAP - on a public and mirrored site (likely with an RSS feed))
4)OPEN SOURCE SOFTWARE - if the process of counting votes as well as the procedure is required by law to be public, then that would include the algorithms used by the computer software. There is no need for tens of thousands of lines of code, nor the ability to 'subtract' votes. (though canceling a vote, for a mistake and allowing a re-vote should be possible - sort of like an eraser on a sheet of paper - though only one valid hard copy should be allowed to be stuffed into the ballot box, and all cancels/revotes should have a paper trail.) The software should be able to be inspected on a machine by machine basis before and after the election. (this is possible with a simple text based parser looking for changes in the code) Errors, backdoors, and flaws will be caught by people reviewing and testing the software in the public domain who have no vested interest in the outcomes. (major benefit of opensource collaboration)