E-voting Patches Skew Election?
Whammy666 writes "Wired magazine has an interesting story of how the much-maligned Diebold E-voting machines were allegedly secretely patched before Georgia state's 2002 gubernatorial election. The patches were never certified by independent testing authorities or cleared with Georgia election officials. The election produced an upset which ended in a major upset that defied all polls. A Diebold contractor tells a worrysome tale of how close to a third of the machines were crashing or locking up and how his tests showed the machines producing errors up to 25%. There are no paper audit trails with these systems so it's nearly impossible to check for fraud or malfunction after an actual election."
Even my cyncial mind is having trouble grasping the immense absurdity of the problem with these machines.
;-)
No kidding...
I've worked in firmware (specifically, POS lotterty terminals not all that unlike the Diebold voting machines). And the level of trouble these things have caused simply astounds me. Really, it doesn't take that much effort to come up with a stable, secure, fully auditable terminal. These people control all aspects of the machines! Literally nothing unexpected can occur - No poorly-behaved third party software, no bizarre user requests (with only a handful of choices, linked to a big touchscreen button, what can they do wrong?), no external hacking attempts (on a private net physically separate from the internet)...
If in my former work, if we had made terminals that bad, we'd have people rioting in the streets (literally). Even the few very minor flaws that came to light received front-page headlines in their respective jurisdictions (and in one case, globally), for something FAR more minor than crashes, recording the wrong user selection, or outright invalid data (yeah, *sure* three dark-horses all won by exactly 18181 votes).
Even in worst-case scenarios, such as harware failure, opening the chassis, or a network outage, the machines should respond gracefully by offlining themselves, thus summoning a field tech. And no auditing capabilities? Gimme a frickin' break! They either lie outright (on behalf of whoever bought various elections?) on that point, or have such a broken implementation they'd rather look like idiots for omitting such a "feature" than admit how badly they screwed it up.
But then, I coded for lottery machines, a field where large sums of money change hands. These Diebold machines "only" tally votes, thus expressing the will of the people in choosing who they want to lead them (assuming "each vote counts" has ever held true). Far less important, quite obviously.