Monday, November 15, 2004
Exit Poll Results Differed from Actual Election Results
Editor's Note How could the exit polls in this year's presidential election have diverged so drastically from the results that election officials and the media announced?
Professor Steven Freeman, a statistician at the University of Pennsylvania, offers a disturbing answer. Looking at the exit polls and announced results in Ohio, Florida, and Pennsylvania, he concludes that the odds against such an accidental discrepancy in all three states together was 250 million to one.
Read the article at http://truthout.org/unexplainedexitpoll.pdf.
After reading the article, I recalled articles published several months ago regarding how easy it would be to modify the program in a computer that tabulates the results of touch-screen voting. The equipment used in several States for such voting was supplied by the Diebold Company. Officials of that company reportedly contributed generously to the campaign to reelect George W. Bush.
I don’t say that I believe that the software was tampered with or that the results of the tabulation of votes were illegally altered to favor Bush. I merely submit the following as a hypothesis.
We know that Diebold’s software is proprietary and the source code was not available to county registrars and Secretaries of State for independent verification and audit. All that could be done was to check the machines and verify that votes for candidates Bush, Nader, Kerry, and others were recorded and stored correctly. We also know that it is easy to prepare a subroutine that will operate only on a specified date. An infamous example is the Michelangelo Virus that operated only on a particular date in March.
Suppose a routine was prepared that would count incorrectly the votes for Kerry such that, say, every 40th Kerry vote was given to Bush. Suppose also that the routine became operational only on Election Day, November 2. Suppose also that the routine deleted itself on November 3. The result: 2 ½ percent of Kerry’s vote is given to Bush, enough to change the outcome of the election in a few States. Checking the voting system before Election Day or after Election Day would not have revealed the swindle. Even checking it on Election Day might not have revealed the false count. One would have to insert forty (or whatever number the programmer had chosen) Kerry votes to see the effect.
Of course, this is just speculation and opinion, not fact. I don’t assert that it happened. I only say that it could have happened and would explain the discrepancy between the exit poll results and the actual count of votes.
Professor Steven Freeman, a statistician at the University of Pennsylvania, offers a disturbing answer. Looking at the exit polls and announced results in Ohio, Florida, and Pennsylvania, he concludes that the odds against such an accidental discrepancy in all three states together was 250 million to one.
Read the article at http://truthout.org/unexplainedexitpoll.pdf.
After reading the article, I recalled articles published several months ago regarding how easy it would be to modify the program in a computer that tabulates the results of touch-screen voting. The equipment used in several States for such voting was supplied by the Diebold Company. Officials of that company reportedly contributed generously to the campaign to reelect George W. Bush.
I don’t say that I believe that the software was tampered with or that the results of the tabulation of votes were illegally altered to favor Bush. I merely submit the following as a hypothesis.
We know that Diebold’s software is proprietary and the source code was not available to county registrars and Secretaries of State for independent verification and audit. All that could be done was to check the machines and verify that votes for candidates Bush, Nader, Kerry, and others were recorded and stored correctly. We also know that it is easy to prepare a subroutine that will operate only on a specified date. An infamous example is the Michelangelo Virus that operated only on a particular date in March.
Suppose a routine was prepared that would count incorrectly the votes for Kerry such that, say, every 40th Kerry vote was given to Bush. Suppose also that the routine became operational only on Election Day, November 2. Suppose also that the routine deleted itself on November 3. The result: 2 ½ percent of Kerry’s vote is given to Bush, enough to change the outcome of the election in a few States. Checking the voting system before Election Day or after Election Day would not have revealed the swindle. Even checking it on Election Day might not have revealed the false count. One would have to insert forty (or whatever number the programmer had chosen) Kerry votes to see the effect.
Of course, this is just speculation and opinion, not fact. I don’t assert that it happened. I only say that it could have happened and would explain the discrepancy between the exit poll results and the actual count of votes.
