Rage against the machine
The web is abuzz these days about potential irregularities and electoral fraud (and here, and here, and here, here, and here).
Let me summarize what I gathered in the past 2 days:
- 30 to 56% (I have found both numbers) of the voters in the US voted through an e-voting machine. The law apparently varies from county to county (another aberration when considering that the outcome is national), so some counties provide paper copies, but as I understand most don't.
- These e-voting machines are made by private parties (ES&S and Diebold). They are black boxes, use proprietary software, so are under total control of...the people? No, a few individuals in a couple of big corporations. Even if none of the fraud claim is true, there's already a big problem right there. One of the founding principle of the democratic process is to have transparent elections. That's a major failure of democracy to have surrendered this transparency to a bunch of private individuals.
- One of the most troubling factor is the unusual wide gap between the exit polls and the actual counts. Apparently the exit polls are usually very close from the actual results, at the point that there are used as an indicator of electoral fraud. Several articles report that there is a surprising correlation between a disparity in the exit poll vs actual counting results and the use of no-paper-trail e-voting machine. Look at the following figure (click on it to enlarge). This is a subset of the discrepancy between exit polls and machine tally for states using paper ballots and states using e-voting:
Now, I'm a scientist, and although this sure looks fishy, these are only a few examples, 9 to be precise, out of 50 states. I have not found an extensive list nor a statistical analysis of this. If you pick them up carefully out of the 50 states, may be you can pull up something that looks like a tendency while it is nothing else than 3 sigma events. I'd dearly like to see this analysis done seriously on all states, or do it myself if I had all the data. - People with responsibilities in the political system are asking questions.
I don't know yet what to believe. Basically, I have a hard time to understand how such electoral fraud could be covered up. I would naively assume that, even though the control is relinquished to these few e-voting machine companies, you still need to involve a lot of people to pull up something this big. How could this stay secret? That's my basic question and primary reluctance to believe what sounds like nothing else that a conspiracy theory, but it's still very troubling. Any comment appreciated (if anyone ever read this!).
Update 1: Here is an interesting story from another angle, but that also underlines the worries we can have about the whole thing.
Update 2: Am I dreaming?? For computer savy people, you can have a look at this paper. It's a very thorough analysis of what kind of hacks one version of the Diebold machines are prone to. I can't believe it. They are prone to "man-in-the-middle" attacks, they actually run a Windows CE OS (!), and could eventually be tampered with by Voters, Poll workers, Internet providers, OS developper and, of course, Voting device developper. I don't have a formal software engineer education, but I write a lot of software (C, interpreted languages), and I know enough to know that (1) no software can be deemed bugless, (2) increasing the complexity of the software increases the chance for bugs exponentially and (3) using high levels tools (in that case, C++ and using an OS to run the software) also increases the chance of bugs or tampering. I can understand why most computer scientists are worried about the whole e-voting issue.
