Some weeks ago Lott wrote this article, where he dismissed concerns about fraudulent electronic voting as "conspiracy theories". As far as I can tell Lott has no expert knowledge about computers, and rather than do any research into the electronic voting machines, he has just invented his own version of the way these machines work. For example, he wrote:
After the election, most electronic voting machines transfer the election results to a compact disk or some other "read only" format. These CDs are then taken to a central location where they are read into a computer.
However, as Avi Rubin, who is a computer science professor at John Hopkins, states about Diebold machines:
All of the tallies are kept on PCMCIA cards.
PCMCIA cards are not "read only"---stick them into a laptop and you can change any of the information stored on one.
In this thread at The High Road, pro-gun activist Jim March explained how he emailed Lott, writing:
Dr. Lott,
You may remember me. I'm the gun rights activist out in California who has been tracking racism and misconduct in concealed weapons permits; I have an autographed copy of "More Guns, Less Crime" and I am otherwise a huge fan of yours.
Which is why I was saddened by the factual errors in "Voting machine conspiracy theories"
March went on to correct the numerous errors in Lott's article, in particular:
Problem: one, all three of the biggest and most suspect vendors (Diebold, ES&S and Sequoia) do NOT use read-only CDs. They use PCMCIA read/write memory cards. Security problems with the data on these memory cards has been noted before. One of the Diebold internal memos that has been leaked discussed that very issue, in a report of a conversation with a California SecState staffer name of Lou Dieder:
March finished his email with:
Calling myself, Bev Harris, Professor Rebecca Mercuri, Dr. David Dill, Professor Doug Jones and others "conspiracy theorists" doesn't address the underlying disaster here, and I am deeply disappointed in your column.
Professor Lott, you did important scholarship on the CCW issue in '97 with Prof. Mustard. That, you researched. If state governments always acted sanely, the CCW laws in California, New York and elsewhere would have been reformed shortly afterwards. You of all people know how screwed up things can get in the public sector.
Re-think your position here. You're being used as a stooge of some very corrupt and unethical people. You're better than that.
Pay attention to us geeks on this one, sir.
I will await your reply before making this document public.
Lott's reply:
I appreciate your comments, and I appreciate your interest in my work on guns. I don't have time to respond in depth, but I will assure you that I am not a "stooge" for anyone, nor did anyone ask me to write this piece. I have done a lot of work on voting machines, having been for example the statistical expert for the minority report from the USCCR on the 2000 election and having been the statistical expert for the state of Ohio in evaluating the different voting machines. I also worked with USA Today as one of their statistical experts. I do appreciate your information and I will look at it, though I should tell you that I have talked to computer programming experts as well as those who designed these machines (as well as their competitors). Attached is also a paper that I have done that is very critical of electronic voting machines with respect to another issue. Again, no one asked me to write this.
March replied to Lott, giving details of the problems with Diebold and providing links where all of his information could be confirmed. They exchanged a couple more emails. So what did Lott do next? He publishes another op-ed in the Washington Times repeating his false claims including:
Most electronic voting machines transfer the election results to a compact disk or some other "read only" format. These CDs are then taken to a central location where they are read into a computer.
It would have only taken Lott a few minutes in the first place to find out how electronic voting machines transfer their results. But he didn't. After March corrected him, it was even easier for him to find out for sure that he was wrong. But he didn't. This is the same behaviour that got Lott into trouble over the survey. If he had done his research properly in the first place he would never have made the 98% claim. And if he had been willing to admit to making a mistake he would never have told the story about conducting a survey in 1997 that has caused him so much trouble.
The latest installement in the story is here. John Fund in the Wall Street Journalquotes Lott's saying that concerns raised by March and others sound
"a lot like an effort to anger some people into voting while providing the basis for lots of election litigation if the results are close."
Fund then asserts that March has lost credibility because he stands to profit if a lawsuit he is help to bring against Diebold is successful. I never thought that the Wall Street Journal would come down against the profit motive.
Lott has misunderstood the problem entirely. There are two things to worry about: 1. changing the votes after they have been cast (fixed, partially, by using write-once media, but much better addressed using paper receipts); 2. changing the software in the machines to misrecord votes (fixed only using human-readable records outside the machine, which means paper). He describes an audit procedure which would be pretty easy to circumvent (for instance, by modifying other parts of the software in the chain) or which could be circumvented by dishonest election officials.
Basically he has no idea what he's talking about.
I have done a lot of work on voting machines, having been for example the statistical expert for the minority report from the USCCR on the 2000 election
By the way, Tim, if you find yourself with a long coffee break any time, there is some absolute pure gold in the reports Lott wrote as a "statistical expert". IIRC (Atrios spotted it), he was trying to prove that black voters had not had their votes disallowed with any greater frequency than white voters, and did so with the following regression on county level data:
votes disallowed in county = x1 + x2 (%BLACK) + x3 (%WHITE) + x4 (%HISPANIC)
unsurprisingly, this collinear regression reported that none of its coefficients were significant ... I seem to remember there were a few other gems.
If the USCCR, state of Ohio, and USA Today hired someone as an expert who was actually that clueless about both the technology and basic statistics, there's only one logical conclusion we can reach: those organizations have problems distinguishing real experts from frauds.
dsquared: The next version was even worse. Seriously. He reversed himself and found that Black votes were more likely to get disallowed, but that was because the Republican Black votes were 50 times as likely to be disallowed. Insane.