Reading The Vote
The ideal and perfect vision of exactly how ballots should be counted came to the world's eyes when television networks in the 2004 election showed, over and over again, the Democratic and Republican volunteers in Florida holding ballots up to the light to check the holes. When the sides agreed on a ballot, that vote was counted. Many ballots were put in the "unsure" box for later review.
Elections are contests between adversaries. Each adversary should be able to verify her vote count from the ballots. For party members, a party representative at each polling place can confirm the count. Independents and write-ins cannot afford representatives at every polling place, nor can they have a presence at every counting.
As my vote is scanned now, the scanner just reads the black spots where I mark the ballot with a grease pencil. Storing this data is cheap. This stream of data - all the votes cast - could easily be sent to any candidate for her own precinct by precinct review. This would let anyone who has doubts about the vote count confirm for themselves, quite inexpensively, how the ballots were evaluated.
All that is missing is a printout telling me, the voter, how my ballot was seen by the system.
In California, a law was passed requiring all voting machines to generate just such a voter-verified paper trail. In early voting in San Diego on touch-screen machines, however, voters are complaining that the paper scoots by so fast that they can't really read it and verify the accuracy. Hmm.
In Florida and Missouri, early voters have discovered that the touch-screen voting machines in use there suffer from what the company calls "slippage". The invisible map of the touch-sensitive spots on the screen slowly slips out from under the image of the buttons. Press the button image for one candidate and you press the zone map hot spot for another.
On his evening news show last night, Lou Dobbs showed video of a candidate visiting a site where these machines are stored to check them out. She pushed her own button on one of the machines and the camera panned to the response from the machine showing that her competitor's name had been selected.
A touch on a Democrat's button image can get counted as a vote for the Republican.
Most will notice. For the small percent who fail to notice this, the totals will be changed.
The Democrats are now onto the scam. Either the machines will get pulled because the creep is a serious bug - a system variable is being overwritten by the program - or they will remain in use and will be checked for creep by aware voters.
The poll workers have been shown how to reset the system variable, although it is unknown whether this resets the counters as well. In a Palm Pilot, it would zero them.
The poll workers say that they often have to reset the machines, and some get so bad they have to be taken out of service. They reset the machines only when people complain. Democrats will complain.
Given that there is a bug that overwrites system memory, it's anyone's guess as to what else that bug may do, given enough time. Bring down a machine, perhaps.
Apparently ISO 9000 compliance was not a part of the specification for these machines. Maybe someday it will be. Flaws discovered become flaws corrected.
Elections are contests between adversaries. Each adversary should be able to verify her vote count from the ballots. For party members, a party representative at each polling place can confirm the count. Independents and write-ins cannot afford representatives at every polling place, nor can they have a presence at every counting.
As my vote is scanned now, the scanner just reads the black spots where I mark the ballot with a grease pencil. Storing this data is cheap. This stream of data - all the votes cast - could easily be sent to any candidate for her own precinct by precinct review. This would let anyone who has doubts about the vote count confirm for themselves, quite inexpensively, how the ballots were evaluated.
All that is missing is a printout telling me, the voter, how my ballot was seen by the system.
In California, a law was passed requiring all voting machines to generate just such a voter-verified paper trail. In early voting in San Diego on touch-screen machines, however, voters are complaining that the paper scoots by so fast that they can't really read it and verify the accuracy. Hmm.
In Florida and Missouri, early voters have discovered that the touch-screen voting machines in use there suffer from what the company calls "slippage". The invisible map of the touch-sensitive spots on the screen slowly slips out from under the image of the buttons. Press the button image for one candidate and you press the zone map hot spot for another.
On his evening news show last night, Lou Dobbs showed video of a candidate visiting a site where these machines are stored to check them out. She pushed her own button on one of the machines and the camera panned to the response from the machine showing that her competitor's name had been selected.
A touch on a Democrat's button image can get counted as a vote for the Republican.
Most will notice. For the small percent who fail to notice this, the totals will be changed.
The Democrats are now onto the scam. Either the machines will get pulled because the creep is a serious bug - a system variable is being overwritten by the program - or they will remain in use and will be checked for creep by aware voters.
The poll workers have been shown how to reset the system variable, although it is unknown whether this resets the counters as well. In a Palm Pilot, it would zero them.
The poll workers say that they often have to reset the machines, and some get so bad they have to be taken out of service. They reset the machines only when people complain. Democrats will complain.
Given that there is a bug that overwrites system memory, it's anyone's guess as to what else that bug may do, given enough time. Bring down a machine, perhaps.
Apparently ISO 9000 compliance was not a part of the specification for these machines. Maybe someday it will be. Flaws discovered become flaws corrected.
posted by Dan McIntyre at 12:50 PM
0 Comments:
Post a Comment
<< Home