Slashdot Mirror
Evoting in India, Maryland
Anonymous Coward writes "EVMs are back in the news again. The BBC is reporting on the use of over a million Electronic Voting Machines (EVM) in India for Parliamentary elections in April. With a billion people and an electorate of 668 million, it is by far the largest democratic election exercise in the world. A picture of an EVM is provided." And Kierthos writes "An article on Yahoo! News mentions that Maryland's voting terminals will be wrapped in tamper proof tape, which 'just protects that malicious code physically', according to computer scientist Avi Rubin. Also mentioned are California's ongoing system of e-voting, as well as a point on whether Diebold should be banned in California after using uncertified software in last October's election."
After everything we've heard about Diebold in the past few months - thier ties to Bush, uncertified software, etc - does anyone really trust them to accurately count and record the results of the votes?
Maybe the states that are still using Diebold machines know something I don't, but I really don't see why you'd want to take such a risk with something as important as voting.
Twenties Retirement
I've followed the developments in Maryland closely, and what's been noticeably absent from every report I've seen on the subject has been any discussion of what the consequences would be if the tamper-proof tape shows tampering.
More to the point: can anyone disenfranchise a whole bunch of voters by just damaging the tape, deliberately or accidentally, while voting?
With the elections being as close as they have been, shutting down machines in a few heavily [democratic|republican] districts could easily change the results of the election. You may not know exactly what each individual machine has recorded, but it's easy enough to find areas that you can expect to reliably vote for one of the major parties. If you were to get the results from machines in Denver thrown out, for example, who do you think would benefit?
Twenties Retirement
This whole argument (aside from the Diebold fiasco(s) ) stems from the Florida Election of 2000 fiasco.
Florida used punch cards. Punch out the perforated block, bingo you've voted.
The fiaso occurred because, what constiuted a "vote" was being subjectviely defined... by whatever party happened to be reading the ballot. Some puches were partially knocked out. Did that constitute a vote? If so, if there was one punch out for one candidate and a partial punch for another, did that invalidate the vote or did it count for the whole punch or the partial one?
On top of that, while they were handling the ballots during the recount, some of the punch outs were coming off!
And don't think you're safe with your pencil and paper! Oh no! It's politics. Any side will find anyway to hem and haw about interpretations of rules and ballots.
That's what partially kicked off this whole EVoting craze in the US. To try to prevent such a thing from occurring again.
We need to think carefully about this tech but we also need to embrace it. We already let automation run our reactors, manager all of our money, keep us from running into each other at intersections, etc.
1) A custom OS WILL have bugs. All new software does. Do you want to vote on buggy software?
2) Part of the outcry (at least here) against e-voting is exactly that - that nobody can see the source, which means we have no way of knowing if it's correct, if it has backdoors, etc.
3) Nothing is 100%, expecially when people are involved.
Twenties Retirement
Who's to say the source doesn't contains election-rigging code, and locking it up so no one can review it. Even if it was "open source", someone at each poll location would have to review the source and compile it there in front of a few officials just to make sure no one loads a malicious binary. That may not even prevent a code snippet:
while(1)
{
voteRepublican();
}
from being sneaked in to the source undetected.
$cat
And the real problem with the 2000 election is one that is not often addressed directly, but is ceratainly relavant to the current topic.
All voting is a statistical process. No system is perfect, there will always be errors. Thus the system has a margin of error.
The 2000 vote was the problem it was because the vote was inside the margin of error, thus no amount of fiddling, recounting, whatever, could possibly resolve the issue. Statistically speaking, the vote was a dead heat and the only reason it had to be decided by the dead heat in Florida was because it was a dead heat pretty much everywhere else as well.
In terms of the "problem" this is indicative of the choices of candidates being a coin toss to most of the populace, which is, essentially, how we resolved it. By using technology to reduce the margin of error we can avoid the political brouhaha of coin toss elections by allowing one candidate to "win" by 20 votes or some such, but it does nothing to cure the political problems that lead to such dead heat elections in the first place.
Do you want Frog ala Peche, or Peche ala Frog?
Not to mention the problem inherent in such elections where a goodly portion of the voting populace look at the opposing candidates, flip their coin, look at it, then just say "Fuck it, it doesn't even matter," and stay home on election day.
Give us statistically descernable candidates and we just might have election results statistically significant.
Of course, to the candidates themselves such an idea is anathema.
KFG
Maybe I'll even vote next time.
Oh very nice attitude. Exactly what we need to ensure good governance. Educated people sitting at home on election day. After all, you can't be bothered to spend 20 minutes going to the polling booth once every 5 years, can you?
Don't bother complaining about the government again. You don't have that right.
Before diving in to what kind of design we should be using, I think some time needs to be spend deciding the design parameters. The solution should probably:
The Nevada Gaming Board has been cited as a good example of the kind of extremely paranoid testing and auditing that needs to go into a system this important. However, for a voting system we've added some new and challenging criteria- anonymity, ease of use, and fairness. None of these individually are difficult, but when combined with the testabilty and auditability become particularly challenging. How do you ensure that individual votes are getting properly registered while still maintaining the anonymity of the votes?
Personally, I don't see how all of these criteria can be met in a "remote" (e.g. web) voting system. However, I think these problems are all solvable with our current technology, if we are careful. In fact, I think that if a system were designed carefully, we could even come up with a system where we can, if necessary, confirm (validate) a region or even nation's voting results by storing individual voting results on voter-owned smart cards.
Assume we set up a system where every voter is issued a voting smart card that they retain possession over. When you go in to vote you stick your smart card in the voting machine. You then vote, and it records your choices on the card. Audits could then take place after an election by having randomly selected voters come in and stick their smart cards into a seperate vote validation system that retallies the results and allows voters to confirm their vote selection. Using statistics, you can set a threshold for when the error level is too suspiciously high, and require revotes in the regions with anomalous results. By using different vendors to provide the voting machines, smart cards, and vote auditing system, you can greatly increase your assurance that no entity has tampered with the voting results. Apart from the influences of the media... and politicians... and education system... and religions.....
On second thought, forget the whole thing. :)