Response to Salon's piece on the elections

Readers should not become pluperfectly pissed off at Salon's Farhad Manjoo, author of two articles debunking, or semi-debunking, the arguments many have offered in favor of the "vote fraud" thesis. Salon is invaluable, and Manjoo has done much good work.

(Incidentally, when he interviewed me vis-a-vis the bulge-gate affair, this same writer asked if "Joseph Cannon" is my real name. An hour passed before I realised how odd it was to be asked that question by a guy named Farhad Manjoo.)

Even so, I think Salon got much wrong. They kindly published my letter of rebuttal -- which I will also place on the record here.

* * *

I am sorry that some readers have offered unkind and even suspicious remarks toward Farhad Manjoo. At times like these, the earwig of paranoia besets even of the best of us.

But fact, not paranoia, compels me to disagree with Manjoo's conclusions. The best place to find these facts can be found in the new study by Dr. Steven F. Freeman of the University of Pennsylvania.

At least Manjoo tries to deal directly with the problematical exit poll data, which most debunkers ignore. Unfortunately, he neglects to tell readers that exit poll discrepancies are the only bulwark we have against electronic vote tampering. That's why such disparities are viewed as a serious matter by international election observers, and why similar disparities in the (formerly Soviet) Georgian elections led to the ouster of Shevardnadze.

I have scoured the net, and I have yet to see a single story indicating that exit poll discrepancies undervalued the Kerry vote in any state. All the evidence I have seen indicates only the Bush vote was undervalued. Certainly, the pattern in the battleground states is beyond dispute. Dr. Freeman reports that the chances against the degree of undervaluation in just three states (Florida, Pennsylvania, and Ohio) is 250,000 to 1.

Dr. David Anick of MIT came to a similar conclusion.

Manjoo falls into the trap of referring to "early" exit polls as though this factor explains the undervaluation of the Bush vote. Why? Common wisdom used to hold that Republicans voted early, and Democrats voted late. If that adage no longer holds true, we should expect the exits to give us errors favoring both Bush and Kerry. The fact that only the Bush vote was undervalued in the battleground states offers legitimate grounds for suspicion.

Manjoo writes: "As Mark Blumenthal, the Democratic pollster who runs the blog Mystery Pollster, notes, exit polling data is not re-weighted all at once -- it's done live, as the results come in, in different precincts at different times."

If Blumenthal is accurate (and if Manjoo cannot prove that all Democrats have suddenly turned into early risers) then the early polls are more likely to give us an objective look at the electorate. If that is the case, then Dr. Freeman (who used the latest available exit polls) should have relied upon earlier data, which showed an even more noteworthy undervaluation of the Bush vote.

Incidentally, Freeman considers the exit poll/actuals shift to be significant, not "very slight." I think we should defer to his expertise, unless Manjoo can bring forth a counter-argument from an expert with better credentials.

Regarding the apparent optical scan/e-vote discrepancy based on Kathy Dopp's information: I was impressed by her chart at first. Within 24 hours, I had to concede much territory after receiving an education on the "blue dog" phenomenon in certain small Florida counties. More recently, I came across arguments attempting to debunk the debunkers -- some believe that those blue dogs voted in greater numbers than one would have predicted. I am willing to leave the question for others to thrash out, since Dopp's chart now strikes me as something of a diversion.

The main issue, it seems to me, concerns the central tabulating system. It handles results from both optical and electronic counties, and it can be hacked by a bright teenager.

Manjoo can't explain why exit polls should undervalue the Republican vote consistently -- state after state, election after election. Neither does he address the question of why these polls proved accurate in the past, or why they are accurate to a remarkable degree in other countries. He asks us to wait until the exits are studied in a probe that is "currently underway." Alas, we had similarly skewed exit poll results in 2000 and 2002. When are studies of those elections going to reach us? If we wait months or years for studies of the 2004 election to be performed, the public's attention will be elsewhere when the results come out (if they come out). Few will double-check any problems. Most will fall for any story that sounds good superficially.

This is the third election in a row in which the exit poll disparities showed the same suspicious pattern seemingly indicative of Republican vote tampering. At what point will the time come for Americans to call a spade a spade -- just as they did in Georgia?