So I'm driving to dinner tonight, and I am listening to NPR on the way. A Ohio NPR story about how Diebold voting machines have been investigated and found to be insecure catches my attention, so I turn it up. Ohio has been looking at giving Diebold a multi-million dollar contract to replace the voting machines we have here (paper and chads) with electronic voting machines. They introduce one of the authors of the study, a Dan Wallach from Rice University. Dan starts to speak about how they tested...

Wait a minute.

Dan WALLACH?

"Dapper" Dan Wallach?

From Texas? Rice University?

As in OUR dwallach?

Yup, that's him. Right on his homepage there is a link to the analysis. Kudos, Dan, you spoke very well and gave a good argument against Diebold's rebuttal. I would say that there was a slight slant towards caution regarding the voting machines and their insecurity, but they did mention that it was not a Johns Hopkins "official" study. I thought it was way cool that I heard someone from this BBS on the radio.

That's really cool. Congratulations Dan!

Cool!

Yup, that's me. A couple weeks ago, the source code to Diebold's voting terminal mysteriously appeared on a web site in New Zealand. We put together a team of researchers to study it, and ultimately cranked out that report. At this point, we're in something of a P.R. battle with the vendor and some of their customers who can't seem to wrap their heads around just exactly how easy it is to compromise an election using one of these things. More recently, a Diebold spokesman confirmed that the code we read is exactly the code that ran in November 2002 in Georgia, among other places.

So, yeah, this sort of thing can be a lot of fun (saving democracy and all), but it's also stressful. It's not often that I get woken up by a phone call at 7am (central) from an L.A. talk radio station that wants do a live interview at 7:40am (central == 5:40am pacific). It's been quite a ride. Of course, test driving zippy cars can be an excellent stress reliever...

Dan, I applaud your efforts to improve our voting process. Maybe you can now use your notoriety to suggest a more mathematically accurate voting algorithm that doesn't implicitly force a two-party system.

I'm going to go out of my way to see if I can't get home before Sunday, as per our offline conversation, as I want to shake your hand.

Maybe you can now use your notoriety to suggest a more mathematically accurate voting algorithm that doesn't implicitly force a two-party system.

"It makes no difference which one of us you vote for. Either way, your planet is doomed... DOOMED!"

"I think I'll vote for a third party candidate!"

"Go ahead, throooow your vote away"

*grumble grumble*

*punch*

Don't blame me, I voted for Kodos.

Edit: Yeah, it was Kodos, not Kang. I'm slippin'.

My current favorite voting system is approval voting. It's simpler than ranking systems and has some nice mathematical stability properties. I actually like the idea of computerized voting terminals (as long as they have paper audit trails) because they can be easily reprogrammed to support these sorts of voting systems.

Once you've got technology that can support fancier voting systems, then it's a, umm, simple matter of politics to get it adopted.

I'm going to go out of my way to see if I can't get home before Sunday

Context for everybody: I'm going to be in D.C. this week for a conference and staying this weekend at my fiancée's parents' place near Fairfax, VA. This puts me within striking distance of meeting up with Bitt in NC and playing with his whizzy car...

Yeah, I've got no problem with computerized terminals (as long as they are still at a voting place and not virtual terminals over a network), but there's no excuse for making a system that's worse than what now exists. Computerized terminals could even go so far as to show the voter's choices back to him to verify -- something that would have at least helped those idiots in Florida who couldn't figure out how to vote.

Yeah Bitt, those computerised terminals are AWESOME and yes you can go back and forth through the pages. Once you vote this way you'll wonder why we used the "chad" system for so long....although I have no idea as to who makes the version for my home state.

I've never seen a chad system. Befoer I could vote, I remember my folks voting in a booth with the levers. I thought they were way cool. Shortly before I turned 18, they got rid of those and brought in paper ballots. The candidates' names are set next to a broken arrow and to vote for them, you finish the arrow. It's then run through a scanner of some nature.

I think that the killer app for a computerized voting machine would be to allow the voters to later go back and confirm that their vote was registered properly. When I put it in the machine, I always feel like my ballot is going into a black hole.

I remember the pull-lever machines too. They were cool. But my county (thank goodness) went straight from those to the electronic ones where you push a button next to each candidate and a red light lights up next to their name. Then when you're done you press a big, green "VOTE" at the bottom of the machine. Plus, the names are all lined up with a large amount of space around each one of them, so as to avoid any kind of "butterfly-like" confusion

We still have lever machines. I mocked the butterfly ballots and was told the old lever machines were obsolete. Odd, never seems to be any confusion, and they work... (well, good luck finding parts, but...)

I'm all for simple. In Canada, we get a card with the candidate names down one half, with big ol' circles (dime sized) beside them. All you had to do was take the pencil they gave you, and put an X in the circle. Then people counted them.

What's wrong with the pencil'n'paper system?
Instant feedback, irrefutable and correctable. We have compulsory voting here and the instructions are always clear on the forms. if you deviate from the form, it's invalid. The senate foting papers are quite large but it's good to feel democracy in action as you hold a novelty cheque sized voting card.

I think that the killer app for a computerized voting machine would be to allow the voters to later go back and confirm that their vote was registered properly. When I put it in the machine, I always feel like my ballot is going into a black hole.

You get this when you vote your shares for a public corporation. But, there's no anonymity. For public elections, you want voter anonymity for a lot of reasons. If you could ever tie a voter back to their vote, you could use that to intimidate them (if they vote against your preferred candidate) or bribe them (as a reward for supporting your preferred candidate). As such, once you leave the polling place, there should be absolutely no way to prove to anybody, yourself included, that your vote was tallied correctly.

On the other hand, you're rightly concerned about the "black box" effect of wondering what happens when you cast the ballot. That's why we support having paper in the system. Paper records are much less malleable than electronic records. If you saw the physical paper that was printed by your voting terminal, you'd have somewhat more assurance that your vote would be around to be recounted, among other things.

But, there's no anonymity.

I was thinking of something a little more cryptographically secure. My crypto is a little rusty, but I seem to remember several algrorithms that allow you to verify information without revealing identity.

And screw paper. The vote should be stamped into a stainless plate, like a dogtag. That'd be cost effective.

I was thinking of something a little more cryptographically secure.

In that case, you would vote, and when you were done, you'd get a paper receipt with a bunch of seemingly random digits (or a barcode version of the same). Later on, you could type those numbers into your computer, press the "verify" button, and get a message saying that, yup, your votes are in there, but we still can't confirm who you voted for.

Maybe you can do better than that, but not by much. Compare that to the confusion among voters trying to make sense out of these receipts and I don't think it's worth the bother.

I think you ought to be able to confidentially get back what the votes were. But I could be wrong.

But from a usability point of view, just install readers somewhere (public libraries seem a good place) that have barcode readers. Scan your receipt and see who you voted for.

I think you ought to be able to confidentially get back what the votes were. But I could be wrong.

As Dan said, the problem with this is that once you have physical proof of who you voted for, you can be paid/threatened for your vote. Here at berkeley our on campus elections provides some kind of crytographic key to retreive what your vote was counted as, so it is possible. We don't worry about the bribery aspect because the voting takes place online and the stakes aren't that high. I thought the source code for this was open, but I can't seem to find it.

Recipt printers cost nothing, it really doesn't make sense that these voting terminals can't keep a paper trail. The problem with a paper trail is that it's going to expose what everyone in elections knows but won't admit -- that a margin of error exists whenver large numbers of ballots are counted.

Anyone want to place bets on when the first election has to be reheld because of a glitch in the computer system?

Matthew

Anyone want to place bets on when the first election has to be reheld because of a glitch in the computer system?

How about where it will take place. Washington maybe? Perhaps near Redmond?