Slashdot Mirror
Deathblow To a Voting Machine
SiggyRadiation writes "According to their newsletter (my English translation here), the Dutch group that 'doesn't trust the voting computers' has won a round against the industry and the civil servants that seem hell-bent on reintroducing voting machines — NewVote, made by SDU — that the Dutch minister of the interior has suspended. Apparently SDU provided 5 slightly different samples of its machine to the Dutch version of the NSA (well... the very humble Dutch version anyway) for testing purposes. Of those five, four machines emitted radiation in such a way that the votes cast could be monitored. SDU's NewVote received its final deathblow when it became clear that the one machine that stayed within the radiation limits used a green-on-red color-scheme for its screen. And that would be a small problem for the 4% of all men that cannot distinguish between red and green."
"NewVote received its final deathblow when it became clear that the one machine that stayed within the radiation limits used a green-on-red color-scheme for its screen. And that would be a small problem for the 4% of all men that cannot distinguish between red and green."
Good heavens. As a a person with good old-fashioned red/green colourblindness I assure you that this statement is false.
There is no way that 4% of men can't distinguish between red and green. There's some difficulty in some circumstances, but a green on red colour screen on a voting machine would almost certainly be readable. They'll use high-contrast hues.
The vast majority of red-green colourblindness results from a cone deficiency. In some circumstances it's difficult to make out some differences, but if I see a red shirt, I know it's red and not green. Green lettering on the red shirt would likely be completely readable.
However, I seldom see purple. Usually it looks blue to me.
No, no, no! That's not how voter-verifiable paper trails work! If you let the voter keep the piece of paper, they can use it to show how they voted (to collect a payment for their vote, or avoid being beat up or fired). If the piece of paper can't be visually read by the voter for them to know what it says, it isn't "voter-verifiable" any longer and doesn't allow immediate detection of fraud. Nobody wants to let the voter keep a piece of paper. (Well, almost nobody. There are some proposals where the paper is only readable using separate equipment which the voter is only allowed to access when alone, but that's a corner case and has problems of its own).
Instead, VVPT systems have a traditional physical lockbox. Think of the paper as being something behind glass; the user looks at it, validates that it says what they want it to say, and then press "yes" or "no". Press yes? It's deposited in a lockbox which can be secured via traditional methods. Press no? It's marked as void, or shredded, or whatever. It's not the voter's responsibility or burden to track the paper; rather, it's kept in the voting system for use in audits and recounts. (Audits being a very important thing -- having the ability to audit means you can take a sample of the physical ballots, check whether the proportions match what the electronic counters said, and know whether you have a big enough problem to require a larger recount).
This is still an improvement over pure paper ballots because you have the usability and accessibility enhancements associated with electronic voting, but the enhanced auditability associated with a piece of paper which a voter has looked at and approved.