I heard a radio interview on NPR over the weekend with some computer security experts talking about the possibility of massive fraud involving the Diebold computerized voting machines used widely in today's elections in the US.

There was also an independent Princeton study discussed here.

One of the radio commentators described how a potential scam might work:

Voters (including me, this morning) are given a card, similar to a bank card, that's inserted into the machine to record their votes electronically. The voter then hands the card back to the poll worker, to be used again by the next voter.

Before the polls open and the voting begins, the cards are initially "cleared" of votes and it is verified that each card holds zero votes. In reality, however, the software does not verify that each candidate has zero votes -- only that the total votes on the card is zero.

Now the scam: Let's say candidate A and candidate B are running. It's possible, using malicious computer code, to pre-load a card so that candiate A has, say, 100 votes while candidate B has -100 votes (yes, I mean a negative number of votes). Remember, the computer only verifies an initial total of zero. Then let's say that 500 people vote using that card whose votes are evenly split -- casting 250 votes for A and 250 votes for B. At the end of the day the totals will be 350 for A and 150 for B. Not only is it rigged in favor of candidate A, but now intense scrutiny of the card by a security expert, after the fact, would show only the new totals -- the bogus "-100" for B having been overwritten by a "+150" -- and all evidence of tampering has vanished!

Incredibly, the expert on the radio claimed that there is currently no safeguard against this. Moreover, the Princeton study showed that malicious code can be propagated from card to card like a virus.

How scary is that?! Has anyone else heard of this? Can Diebold, which also makes ATMs, be so security-deficient that this scenario is plausible? I know it strains credulity, and I don't want to propagate a hysterical urban legend-in-the-making. Reality checks are most welcome.

I'll respond to my own query in the spirit of clearing Karrow's plate of Unanswered Questions.

I rarely tune into to radio or tv broadcasts, I just happen upon them. So the NPR radio piece was apparently in conjunction with Hacking Democracy, a documentary film airing on HBO just prior to the election. It's described in detail here. (Say what you will about Wikipedia, but if nothing else it's up-to-date!)

Also, I was unaware at the time of a parallel AP thread called Diebold Magic.

I haven't seen the entire documentary yet, but I caught about the last 20 minutes of it while surfing my tv dish a few days after my posting. The hack I described is just one of many ways of defrauding the Diebold voting process. Here's the way the Wiki. article describes it:

quote:

This method demonstrated, contrary to a previous Diebold statement, that a person attempting to rig the votes of a precinct would need access to only the card, not the voting machine or tabulation software. This method, when cross-checked between the voting machines and tabulation software, appears legitimate, and further produces a proper zero-vote print out to verify the card is correct before voting begins.

Near the end of the film, a woman weeps after realizing how easily corruptible the voting system is. Between sobs, she says something like (my paraphrasing): Many American soldiers have died defending our right to vote democratically, and this makes it all seem in vain.

More info can be found by Googling the title of the documentary, or at the external links provided by the Wiki. article.

An interesting article to show how easy it was.

Corrupt voting machines.

Thanks to both of you for the articles. I guess there's no evidence of the Diebold memory-card tampering scheme actually occurring in an election, but it still worries me.

Sounds like there may well have been hacking involved in the 2004 Florida (where else!) election, though with an optical scan system.

Supposedly the Diebold systems can make an optional paper record, but precincts can and do decline to pay the extra cost involved. That should be mandatory -- not optional.

I used to think that those who opposed the shift to computertized voting were just being paranoid or technophobic, but I've completely reversed my thinking on this.

Given the enormous economic and political motivation of politicians and other criminals to influence election results, the only sensible approach would be the universal use of computerized voting machines, all of which produce paper records. This would (1) provide immediate but tentative election results, followed by (2) old-fashioned paper vote counting. Probable winners and losers could be declared soon after the election, while some weeks later the results could be certified via paper vote counting. This would also (3) uncover wide discrepancies between the two methods as evidence of machine hacking, while even pointing to the perpetrators.

Diebold should be made to face serious competition from other manufacturers, with governments providing huge market incentives for anti-hacking security measures. Many elections are decided by less than 1% of the total vote, making zero-tolerance of fraud a mandatory goal. Given 21st-Century technology we shouldn't have to continue worrying about the mechanism of fair elections.