Today is THE day of the week for US citizens : Senate and House of Representatives reelections. I am no american citizen, but since I have quite a few friends who are, I tend to follow these stories closely.
How does that relate to computers? Well, for those who don’t know about it yet, as opposed to here, they have electronic voting machines.
I have been asked several times to do my duty as a french citizen, and count the ballots. Here, it looks like this : among the people who actually got to vote, they randomly ask if we would help. If you say yes, you have to be there at the closing of the voting center. Then, in front of you, the appointed man running the show compares the number of ballots with the number of signatures. If this number is correct, him and a few randomly chosen people validate the fact. The some other randomly chosen people make stacks of 100 ballots each.
Then, you are with 3 other people at a table, and a fair share of the stacks. For each stack, one person takes the ballot, one writes the result, and the two other validate. We are encouraged to change roles at each stack. If either one of us has a problem with a ballot, and wants to mark it down as null, we ask the appointed staff to give us some advice and guidance, because technically, you are supposed to leave out any ballot that has been marked in any fashion after the print. However, we, the people, have to be fair and discuss it among ourselves. And we note the result only when the 4 of us agree.
In the end (it takes roughly 1h or 1h and half to count), the staff recounts the number of ballots and checks that the same number appears in our stats. We all validate everything and sign. And we go back home, elated, and happy to have made sure that when you vote, it means something.
It may seem pretty archaic compared to a shiny computer that does that 1 000 000 times faster than a bunch of people. Besides, the computer leaves no room to interpretation : a ballot is valid, or it’s not.
It’s true that if the system is well conceived, a computer cannot be coerced, influenced, or misled. The ultimate referee. At worse, it can be shut down, and that’s it.
But as soon as you start writing programs and working with computers, you see that a perfect computer, like every other perfect thing, doesn’t exist. There are bugs. There are hardware failures, power outages. There are greasy finger that will ruin the touchscreen, kids with corkscrews who “just want to know what that would do”. There are so many things in fact that could be wrong that it’s a small miracle that computers work as well as they are on a day-to-day basis.
Rule #1 of computers : computers are as good as they have been built. Alone and without power, all they do is rust and rot.
So, even though our system is archaic, in the end, we are pretty sure of the results. There are ways to fraud, but they take place outside of this counting method. Usually, the Bad Guys find a way to vote several times, so it just goes past the verification process. With a human controlling, it might go “tick” if 27 John Smith came to vote 10 minutes apart… A computer won’t notice, since the votes are valid, somehow.
When I was in the US back in 2000, I followed the recount saga from within. Then, in 2004, I took a close look at the results. And I was outraged when I saw an “expert” on TV explaining that counting error were solved thanks to the electronic voting machines. He said that since less than 1% of the votes were badly accounted for, it was a much safer way to do democracy.
1%! Out of 300 000 000 people, that makes 3 millions of faulty votes! Do you remember how close the 2000 vote was? Certainly less than that. So, “democracy” is better served if we have several millions as an error margin. Gosh I wish my banker played the same tune! An unthinkable thing in our old-fashioned system…
Anyway… a friend of mine came with a huge grin on his face and said “the USA are the first mathematically exact democracy”. He started explaining how some people managed to reduce the error margin to 0%. No paper ballot. Your vote leaves no trace. I will most certainly be correctly included in the final count (within 99% margin). If your vote is not counted, or is counted for a guy you didn’t vote for, then how can you prove it? No proof? No error then.
But that won’t happen, the optimists say. There will be control people. And since the voting machine is a closed box, with secret innards, no one will be able to temper with them. However, a safe is only as secure as its key… I’ll let you read Bruce Scheier’s piece on the topic, he is a much better security expert than I am. As for the control people : Quis custodiet ipsos custodes?
Here, to validate a vote in a small voting center, there has to be at least 20 people, most of them selected at random, that will vouch for the results. Guess we watch each other… If we are all crooked, then it doesn’t matter that the result is crooked, because it’s a small drop in the whole. If enough of the results nationwide are crooked, then it means that there are more crooks in the country that the prisons can handle.
Electronic voting machines might not be perfect in theory, but it’s hard to convince people that theory becomes reality quite often…
From ABC News
Randy Wooten figured he’d get at least one vote in his bid for mayor of this town of 80 people even if it was just his own.
For the record, 36 votes and one error, that’s 3% error margin ;)