EXIT POLLS AND DIEBOLD

On Tuesday evening, as actual vote tallies in the presidential race began coming in, television commentators immediately noted that there was a marked difference between the actual vote tallies and the projected vote tallies as worked out in the exit polls. The exit polls were being conducted outside of voting booths across the country by Edison Media Research/Mitofsky International in a national election pool jointly sponsored by the Associated Press, CNN, Fox News, and the three broadcast television networks.

Consistently, the actual votes recorded for Kerry were coming in less than the vote projected by the exit polls. And so, for much of the evening, television commentators batted around the question of how the exit polls could have gotten things so wrong.

As if there might be no other possible explanation.

In elections, people steal votes. That is the way of elections, whether they take place in Kabul or Kansas City. If enough votes are stolen, one way or another, the results of an election can be altered. If you call that a conspiracy theory, you are either very naive, or you're covering up.

Probably the most infamous example of American vote-stealing occurred in the 1960 presidential race, which ended with a 49.7 percent/49.5 percent split in the popular vote, one of the closest on record. Nixon dropped a widely-anticipated challenge of that election based upon allegations of Democratic vote-stealing in Illinois and Texas, but only after outgoing President Eisenhower withdrew his support for such a challenge. Rumors later surfaced that Eisenhower took that position only because of fear that a court challenge might reveal that Republican vote-stealing had given Nixon the electoral victory in other states.

In those old paper ballot days, votes were manually counted at each polling station, one by one, after the polls closed. Usually, the poll workers would open the ballot boxes and empty the ballots out on a table. One worker would read out the vote, ballot by ballot, while another worker would mark down each tallied vote on a sheet. To guard against fraud, candidates would have observers in the room standing behind both the reader and the tallyers, making sure the votes were both called out and written down right. To steal votes at the counting table in those days, you had to get rid of the other candidates' observers. And so, I remember that in the first election I worked as an observer-1966, in Dallas County (Selma), Alabama-two of us got run from the polling station by Jim Clark, the Dallas County sheriff who became infamous a year earlier when he ordered the beating of civil rights marchers on the Edmund Pettus Bridge during the Selma-to-Montgomery march.

Still, paper ballots served American elections well during much of the life of the republic. The big advantage was that, so long as you preserved the ballots and made sure no one tampered with them, you could always go back and count them again if someone suspected that someone had cheated on the first count. In most states, in fact, when the difference between the winner and loser gets to within a certain small percentage point, recounts were mandatory in order to make sure an election was not won or lost by human error.

But that was in the old, paper ballot days.

Somewhere along the line, we decided that paper ballots were too old-timey, too old school, and we needed to move up to the modern age. Why such a move was necessary, I've never been able to figure, but move, we did. First to the mechanical devices like the old punch-card machines–the ones made famous in the 2000 "butterfly ballot" and "hanging chad" Florida election–in which the voter pulled a lever that punched a hole in a card, rather than the voter simply taking a pencil and marking an "x" on the ballot. Then, when we became enamored with computers, we decided that we just had to computerize our voting as well. And so, en masse, counties around the country have been running their elections through computer touch screens. Touch your finger to your favored candidate on the screen, an "x" appears in its place, something happens somewhere inside the guts of the computer, and the voter walks away satisfied that the correct vote has been recorded.

And so, trust in human observers has been replaced by trust in computers. In an age when we see clever hackers run circles around computer programs, entering the most "secure" spaces, we ought to be wary of that.

How do we know that the votes in those touch-screen machines are being correctly counted? Actually, we don't.

In the 2003 California gubernatorial election, I got thrown out of the polling counting station in Berkeley, California–the first time that had happened to me since 1966–when, in my job as a reporter for the Berkeley Daily Planet, I attempted to observe the vote count procedure being used on the Diebold computerized voting machines. Representatives of the Alameda County Registrar of Voters office later said that barring observers from the counting station was a mistake, but the workers who called the Alameda County sheriff's deputies and ordered me out insisted they were doing so on orders from their superiors. Who gave those orders, I've always wondered since then, and why? What were they trying to hide?

In that same 2003 recall election, two minor candidates got close to 40 percent and 30 percent, respectively, of their entire statewide vote total in one county using Diebold computerized voting machines. That widely-publicized discrepancy was caught only because it was so out of whack with the statewide results. But if some clever computer wizard were to secretly program in a shift of say, one-half of one percent of the vote from one candidate to another in each computer voting machine in each precinct across the state, the shift would never be noticed, and it could change the results of a statewide race.

We could guard against such a problem if voters were issued a paper receipt when they finished voting on one of those computerized voting machines. The voter could check the receipt to make sure each vote was listed correctly, and then drop the receipt in a ballot box on the way out, just as they used to drop their paper ballot in the old days. In a dispute, the receipts could be manually recounted to make sure the electronic tally given by the computers was correct. ATM's give such receipts. Computerized cash registers give such receipts. But so far, makers of the computerized voting machines have resisted installing such receipt devices on their machines. One wonders why.

Between 2000 and 2002, Diebold–the most popular computerized voting company–gave $200,000 to the Republican National Committee, and Diebold CEO William O'Dell pledged his support to help George W. Bush win re-election in 2004. One wonders how.

Were the Edison Media Research/Mitofsky International exit polls wrong on Tuesday night, and the votes reported by such computerized voting devices as Diebold right? Or was it vice versa?
   
No accusation, friends. Just a preliminary question, from a curious election observer.