Tuesday, November 14, 2006

Schroedinger's Vote

Electronic voting is not a technical problem, it's a philosphical problem about how we need to think about our democracy. It's half alive, and half dead. This echoes the famous thought experiment called "Schroedinger's Cat." Erwin Schroedinger, a physics dude, demonstrated how, in order for particle physics to be practical at a macro or engineering level, it needed to be thought of probablistically.

I'll paraphrase: "Imagine you've got a cat, and you put it in a steel box. If a radioactive particle decays in one hour, the chances of which are 50/50, it will cause poison gas to be released into the box, killing the cat. Because the particle's chances of decay are exactly 50/50, the cat is half alive and half dead." This is a very flawed thought experiment, first because Erwin was obviously a sick man, and next because with my cat I'd be very sure of his status at all times, since he would be growling and roaring while trying to claw the box to pieces from the inside out, quite possibly succeeding. (Given a heavy enough dose of ketamine I wouldn't know for certain if Lord Cat were living or not. Which would, quite honestly, be a relief if traveling. This is a cat who, when the ketamine began to wear off while in a real box, frightened a fellow airline flight passenger straight out of his seat.)

My point is, our votes go into steel boxes, and like the hypothetical cat they are half alive and half dead; and like radioactive particles, democracies have half-lives, and this one is still decaying. I'll look at both sides of the vote in some downwater posts. This one was triggered by scientist Al C. sending along this story:
Catbeller writes "The AP is reporting that Randy Wooten, mayoral candidate for Waldenburg Arkansas (a town of eighty people) discovered that the electronic voting system hadn't registered the one vote he knew had been cast for him ... because he cast it himself. The Machine gave him zero votes. That would be an error rate of 3%, counting the actual votes cast — 18 and 18 for a total of 36."

From the article: "Poinsett County Election Commissioner Junaway Payne said the issue had been discussed but no action taken yet. 'It's our understanding from talking with the secretary of state's office that a court order would have to be obtained in order to open the machine and check the totals,' Payne said. 'The votes were cast on an electronic voting machine, but paper ballots were available.'"
During a lively debate amongst code jockeys at the slashdot link over vote machine engineering, someone correctly observed: "Vote error has nothing to do with statistics. Does an NCR cash register make mistakes?" A voting machine should be like a cash register. A good democracy would put out paper ballots in triplicate, this isn't hard to do, and vote counts would be re-checked as a matter of course. Yet what we're stuck with, as Still Life Living says, is "An Electrical College." As long as we vote electronically, changing an election is as easy as initiating a pre-written sub-routine or simply writing toxic code.

No comments: