Will the Next Election be Hacked?

September 22, 2006 6:12 AM

[O]ne muggy day in mid-August [2002], [Diebold consultant Chris] Hood was surprised to see the president of Diebold's election unit, Bob Urosevich, arrive in Georgia from his headquarters in Texas. With the primaries looming, Urosevich was personally distributing a "patch," a little piece of software designed to correct glitches in the computer program. "We were told that it was intended to fix the clock in the system, which it didn't do," Hood says. "The curious thing is the very swift, covert way this was done. . . . It was an unauthorized patch, and they were trying to keep it secret from the state," Hood told me. "We were told not to talk to county personnel about it. I received instructions directly from Urosevich. It was very unusual that a president of the company would give an order like that and be involved at that level."
- Robert F. Kennedy, Jr., Will the Next Election be Hacked?
posted by Saucy Intruder (111 comments total)

Very suspicious but still no smoking gun. One wishes he would've made a copy of the patch and the source code of the machine and posted it somewhere for further examination. If he asserts that patch did something other than fix the clock then please describe for us what exactly the patch did or show us the code.
posted by StarForce5 at 6:29 AM on September 22, 2006

see also
posted by sergeant sandwich at 6:30 AM on September 22, 2006

very tangentially related:

Googling for ATM passwords

I just think the entire concept of electronic voting needs to be scrapped. There's no way to secure it.
posted by empath at 6:32 AM on September 22, 2006

Synchronicity at it's finest - I just finished reading that article.

My major in college ( a hundred years or so ago) was Marketing Research. It was concerned primarily with designing, testing, and critiquing reserach methodology. It is well known, and obvious, that bad research design leads to bad results. The single most effective way to overcome bad design is to increase the sample size; the more people questioned, the more accurate the results. This is why exit polls have historically been considered, far and away, the most accurate form of polling.

The huge swings between the exit polls in the last two elections and the actual election results is more than enough, to me, to indicate shenanigans (treason). That more public outrage has not resulted is a disgrace.

I'm a little old to be fitted for a foil hat, but....
posted by Benny Andajetz at 6:36 AM on September 22, 2006 [1 favorite]

I agree with StarForce5, this is not a smoking gun. This could easily be a case of "we told them we fixed the bug but it turns out we didn't and to save face we're going to fix it on the sly and no one will know we screwed up." The software industry is filled with examples like this.

Why doesn't one of these "whistleblowers" just snatch a machine, dump the code, return the machine, and then go through it line by line, or realese it so everyone can do that. Blowing smoke only creates a lack of faith in the system, which, by the way, will discourage democrats from voting if they begin to believe the election will be stolen anyway.
posted by Pastabagel at 6:37 AM on September 22, 2006

One wishes he would've made a copy of the patch and the source code of the machine and posted it somewhere for further examination.

The fact that no one could or did proves something too.
posted by StickyCarpet at 6:39 AM on September 22, 2006

We shouldn't be dithering about the need for a "smoking gun." In a democracy, the mechanisms of voting for political representation should be completely transpartent. End of story. When we're dealing with something as valuable as your vote, the burden of proof should be on Diebold to prove they're not crooked, not on its critics.
posted by jonp72 at 6:42 AM on September 22, 2006 [6 favorites]

There's no way to secure it.

Actually, cryptographers like David Chaum and Ron Rivest have devised some very impressive voting systems that satisfy most desiderata of voting systems: verifiable, anonymous (to prevent vote selling), paper receipts, all the things we'd want. However, none of the buying decisions are made by anybody with any security, and whichever companies have the political ties get the deals.
posted by sonofsamiam at 6:44 AM on September 22, 2006

Oh, and as much as my inner crypto nerd loves those schemes, they are likely a little too complicated for most people to follow, and the faith of the voter in the integrity of his vote, by whatever means, has got to be a major goal.
posted by sonofsamiam at 6:45 AM on September 22, 2006

Ok, I brought this up yesterday, Benny brings it up now, so let's rehash it.

Ohio Oct 28-31 poll of likely voters Kerry 50%, Bush 46%, MOE 3%.

Ohio - election night results Bush 50%, Kerry 49%, third parties less than 1%, difference of roughly 118,000 votes out of 5.5 million, or 2.1%.

Ohio - Exit poll data predicts Bush win - voter 18-64 are split 50-50, voters 65 and over Bush 58, kerry 42.

Nothing here is inconsistent. The "very accurate" exit polling very accurately predicted a Bush win.
posted by Pastabagel at 6:46 AM on September 22, 2006

StarForce5: Very suspicious but still no smoking gun. One wishes he would've made a copy of the patch and the source code of the machine and posted it somewhere for further examination.

Back-asswards. Our elections process must be, and is intended to be, completely transparent. There must be no opportunity for suspicion at all. An election isn't a trial; no "innocent until proven guilty" standard applies. If there are good reasons for suspicion, the system has already failed. The legitimacy of an election lies entirely on the faith the public places in its fairness; if that faith falters, for any reason, true or untrue, the election becomes a sham.

Even if you disagree, I think your desire for a "smoking gun" in this case is pretty unrealistic. It isn't made clear what Hood's level of technical expertise is, but the patch was distributed on a flash memory card. Copying such a thing isn't impossible, but it's not like you can nonchalantly walk it over to the nearest Xerox machine, either. How would you propose to obtain the souce code for a voting machine, given that you only had access to a flash memory card that only contained a binary-level patch for said voting machine, possibly in a non-standard file system and possibly encrypted?
posted by Western Infidels at 6:47 AM on September 22, 2006 [1 favorite]

There's no way to secure electronic voting. But there's also no way to secure paper ballots. You can have as many kindly old ladies count and recount them as you like; but if you control the person that they report that number to, the election is stolen.
posted by kfx at 6:47 AM on September 22, 2006

the good news is, if Hillary is nominated, hacking won't even be needed
posted by matteo at 6:48 AM on September 22, 2006 [1 favorite]

Wow, Pastabagel, CNN links. That's deep right there.
posted by prostyle at 6:51 AM on September 22, 2006

sonofsamiam: ...have devised some very impressive voting systems that satisfy most desiderata of voting systems: verifiable, anonymous (to prevent vote selling), paper receipts, all the things we'd want.

I haven't really understood the "paper recipts" part of the electronic voting debacle. A voter gets a piece of paper that indicates the machine "knows" which candidates the voter wanted to vote for and -- that's it. A paper recipt guarantees nothing about the way votes will actually be counted. And once you've got all that paper involved, why not just go right back to good old-fashioned paper ballots?

It seems to me that paper recipts generate a false sense of security.
posted by Western Infidels at 6:54 AM on September 22, 2006

Wow prostyle, last time I checked CNN was a major news outlet. Sorry, was RFK jr referring to the exit polls conducted by the Village Voice, or Pacifica radio?

If you are suggesting CNN is lying (because we all know CNN and Ted Turner are pro-Bush), you need to prove that and come up with the real numbers. So far, this is the only source I can find for 2004 ohio exit poll results. If you know of another source, please provide it.
posted by Pastabagel at 6:57 AM on September 22, 2006

wow, prostyle, contributing absolutely nothing to a thread while snarking about the quality of someone else's contributions. that's deep right there.

what the fuck is your problem?
posted by sergeant sandwich at 6:57 AM on September 22, 2006

You can have as many kindly old ladies count and recount them as you like; but if you control the person that they report that number to, the election is stolen.

This is the difference, and it is an important one: with electronic voting, the cost of stealing an election goes wayyyy down and the likeliehood of getting away with it (due to many, many fewer people being involved) goes way up.

It seems to me that paper recipts generate a false sense of security.

That is a good point.
posted by sonofsamiam at 7:00 AM on September 22, 2006

Pastabagel: Why doesn't one of these "whistle-blowers" just snatch a machine, dump the code, return the machine, and then go through it line by line, or release it so everyone can do that.

It's simply not that easy. The software in the machine will be executable binary code, not source code. And without some information about the hardware it runs on, determining what it does, comprehensively and precisely, may actually be entirely impossible.

You may protest that there have been other amazing hacks of embedded systems (like game consoles, cameras, cell phones, MP3 players, etc), but such things generally rely on additional information sources that would not be available in this case. Like having the hardware available for examination and modification, like having multiple example programs to glean clues from, etc.

I think you're asking the wrong question. Why hasn't Diebold instantly and thoroughly cleared the air by simply releasing the source code themselves? What have they got to lose from such a move?
posted by Western Infidels at 7:05 AM on September 22, 2006

They claim it is to protect their intellectual property against competitors, which is a laugh.
posted by sonofsamiam at 7:08 AM on September 22, 2006

Why hasn't Diebold instantly and thoroughly cleared the air by simply releasing the source code themselves? What have they got to lose from such a move?

As we all have come to realize, it's most noble to spend your time in online discussions by deriding the inquisitions of the powerless and disenfranchised. What kind of communist are you?
posted by prostyle at 7:08 AM on September 22, 2006

If it is that easy, why can't the Democrats just hack the vote too?
posted by spilon at 7:09 AM on September 22, 2006

I am amazed at how much f*cking with voting machines Americans are willing to accept. It's mind-boggling.

If Diebold were interested in the integrity of their product they would welcome challenges, as opposed to fighting them every step of the way -- using the courts (DMCA) when they can, and using bribes ("lobbying") when they can't.

It's even more unbelievable that you people not only choose to allow this situation to stand, but that you then even publicly defend instances where the integrity of the machines is even further compromised by secret "adjustments". I really can't imagine why a people so obsessed with "freedom" and "democratic ideals" is willing to be so lenient with these mechanisms.

I'm certainly glad that Diebold doesn't have the opportunity of pwning elections here.
posted by clevershark at 7:09 AM on September 22, 2006 [1 favorite]

Western Infidels- Let me clarfiy - an insider only has to steal the binaries, because I'm guessing there's an emulation environment for standard PC OS's so these guys can code for the platform without having to load it to the device to test it, or the binaries will run native. (Actually, don't these diebold machines run windows?)

Then, that person can demostrate how the machine "steals" or dumps votes in real-time?

In other words, a whistleblower, an insider, would have this info.
posted by Pastabagel at 7:10 AM on September 22, 2006

This can only mean one thing: we need a photo ID for all voters, it's the only solution!

Warning, likely dim bulb derail spotted dead ahead.

Look, Diebold is a GOP company and will do whatever it takes to help the GOP.
The CEO himself said so. Take his word for it.
They can make better equipment, just look at the bank teller machines.
They will not make machines that cannot be compromised as that is part of the design, allowing easy compromise. Duh.
posted by nofundy at 7:11 AM on September 22, 2006

Paper ballots are of course insecure; we have centuries of history with low-tech election hacking. The only real way to ensure a modicum of integrity is to involve the voter herself in quality assurance. She must be able to trace her vote from the computer screen, to the paper receipt, to a scan of her paper receipt on the Internet, which is then provably aggregated into a precinct-wide tally.
posted by Saucy Intruder at 7:13 AM on September 22, 2006

I just think the entire concept of electronic voting needs to be scrapped. There's no way to secure it.

Which is why the Libertarians must find a way to be in control of these machines! Appealing to the logic and reason of poorly educated voters will never work....
posted by three blind mice at 7:21 AM on September 22, 2006

the diebold machines i have worked with had a paper recipt--but it was retained in the machine, not given to votors. votors could view it to reassure that their votes were recorded correctly. i've been told in this thread that it's a newer style.

the spectre of election fraud has always been around ... and whatever new style of ballot they get won't make a difference. however, usually it's just for small races. in order to pull off something at the national level, the shennagins have to take place somewhere higher up in the tabulation system,
posted by lester's sock puppet at 7:22 AM on September 22, 2006

If it is that easy, why can't the Democrats just hack the vote too?

That completely misses the point. Democrats are corruptible. The point is that elections shouldn't be hackable. And there should be no room for doubt in our minds about it.
posted by eustacescrubb at 7:23 AM on September 22, 2006 [1 favorite]

Pastabagel: ...an insider only has to steal the binaries, because I'm guessing there's an emulation environment for standard PC OS's so these guys can code for the platform without having to load it to the device to test it...

An emulation system would be another complex piece of software to either blindly trust or reverse-engineer in excruciating detail. How do you know the emulation system is an accurate representation of the real machine, after all?

(Actually, don't these diebold machines run windows?)

I don't know, but doing so would multiply opportunities for hidden tampering. Diebold could write an honest vote tallying application and then mess with the counts through a modification to or a security hole in Windows itself. The effects would be entirely hidden from a whistle-blower who only had the code for the honest half.
posted by Western Infidels at 7:25 AM on September 22, 2006

I've seen RFJ jr. Pimping that ridiculous thimerisol = autism stuff, so he doesn't really have a lot of credibility with me.
posted by delmoi at 7:26 AM on September 22, 2006

Then, that person can demostrate how the machine "steals" or dumps votes in real-time?

Not if the patch were designed to run once under a particular condition and then remove all traces of itself. You'd need the source of the patch, which as I understand it, no one has.
posted by saulgoodman at 7:27 AM on September 22, 2006

What Clevershark said.
The fact that electronic voting even exists is amazing to me.

Personally, I was more fond of the white marble, black marble method.
posted by Hanover Phist at 7:29 AM on September 22, 2006

Hell, just some of the code in the patch could have been malicious, and the patch could otherwise have served some innocuous purpose.
posted by saulgoodman at 7:30 AM on September 22, 2006

whatever new style of ballot they get won't make a difference

Not quite so.

Electronic tampering is much easier to hide because extreme expertise is required to detect it.

Paper tampering can be uncovered by any average Joe/Joanne with simple instruction regarding what to watch for.
posted by CynicalKnight at 7:31 AM on September 22, 2006

I distinctly recall in 2002 specific cases where the exit polls were incredibly different from the vote tally. At the time, the rumor was that electronic vote rigging was being trialed in Georgia.

http://www.commondreams.org/views05/0310-32.htm

USA Today reported on Nov. 3, 2002, "In Georgia, an Atlanta Journal-Constitution poll shows Democratic Sen. Max Cleland with a 49%-to-44% lead over Republican Rep. Saxby Chambliss." Cox News Service, based in Atlanta, reported just after the election (Nov. 7) that, "Pollsters may have goofed" because "Republican Rep. Saxby Chambliss defeated incumbent Democratic Sen. Max Cleland by a margin of 53 to 46 percent. The Hotline, a political news service, recalled a series of polls Wednesday showing that Chambliss had been ahead in none of them."

"Similarly," the Zogby polling organization reported on Nov. 7, "no polls predicted the upset victory in Georgia of Republican Sonny Perdue over incumbent Democratic Gov. Roy Barnes. Perdue won by a margin of 52 to 45 percent. The most recent Mason Dixon Poll had shown Barnes ahead 48 to 39 percent last month with a margin of error of plus or minus 4 points."

posted by mulligan at 7:33 AM on September 22, 2006

Then, that person can demostrate how the machine "steals" or dumps votes in real-time?

In other words, a whistleblower, an insider, would have this info.

What are you talking about? We all have that info. We know that anyone with access to the machine, and the ability to pick a lock and a specific program can steal the election if they get access to the machine.

The program doesn't need to be written on the inside, anyone with access to a machine, a lot of patience, and who knows how to program could do it.

This has already been demonstrated conclusively.
posted by delmoi at 7:35 AM on September 22, 2006

I distinctly recall in 2002 specific cases where the exit polls were incredibly different from the vote tally. At the time, the rumor was that electronic vote rigging was being trialed in Georgia.

Um, since polls are conducted with a certain confidence interval, probably 95%, then you know that about 5% of them are going to be wrong by a margin outside of their margin of error. It's just going to happen. Expecting exit polls to be right 100% of the time is, well, stupid.
posted by delmoi at 7:39 AM on September 22, 2006

Ok, this is an issue I've spent a great deal of time studying, and I'd like to take a moment to champion the idea of paper ballots. But let me clarify one thing:

There are many pluses to electronic voting. Instant totals, maleable ballots (same day corrections of "misprints"), and handicap accesibility are the major ones. Lets not throw the baby out with the bathwater. See India's election system if you want to see how simple an electronic voting system can be.

The feature people really want in this situation when they refer to paper is a voter verifiable paper ballot. That way the user can verify that there is a printed out ballot that matches their electronic selections. This "receipt" would be behind glass to prevent tampering, and would end up in a traditional ballot box for recount later should the election be challenged. The user would not be allowed access to this receipt in order to prevent vote selling.

The other key requirement is random recounts. Once the above mechanisms are in place, random polling stations (or perhaps those that differ widely from exit polls) could be quickly audited. After all, what is the point in having an audit trail if there is no way to indicate an audit is needed.

This leaves us with a trust but verify at will environment. We can trust the electronic results and enjoy their benefits, all while knowing that an accurate audit can be performed at any time.
posted by butterstick at 7:41 AM on September 22, 2006

The only real way to ensure a modicum of integrity is to involve the voter herself in quality assurance.

Yep. Be offered a printed receipt (optional, if you're paranoid about secrecy) and an easy way to verify within minutes (and also days and weeks and years later) that your vote is still in the system and was counted.

Also, while you're still in the booth, have a separate screen showing a running summary of who you have selected so you can check your work while you are still voting. Make sure it says "You have selected: Hillary Clinton for president of the United States, [etc.]" before you lock things in.

And put a nice big, fast shredder at each polling station to take care of any unwanted paperwork you carry out with you. Make the shredder itself transparent so you can watch your paperwork go through the system and fall out into the trash, or into your hand if you want to intercept it and make sure it's really unreadable. A fire would be nice, too, but I guess there would be environmental concerns...
posted by pracowity at 7:45 AM on September 22, 2006

I'm amazed at Americans just accepting this, or just bitching about it in online forums. Direct action, protests, campaigning, that's surely what's called for?
posted by imperium at 7:46 AM on September 22, 2006

Um, since polls are conducted with a certain confidence interval, probably 95%, then you know that about 5% of them are going to be wrong by a margin outside of their margin of error. It's just going to happen. Expecting exit polls to be right 100% of the time is, well, stupid.

so it is perfectly normal when 48 - 39 becomes 52-45 ?
that is an astonishing margin of error
posted by mulligan at 7:47 AM on September 22, 2006

52-45 in favor of the guy who was expected to get 39 and lose
posted by mulligan at 7:47 AM on September 22, 2006

At the same time, people who want hand counting of ballots as a solution are out of their minds. All you have to do in that case to steal an election is get a bunch of people to pretend to be members of the other party, and then when they get paired up with each other they steal votes. It would require a pretty big conspiracy, but it could work.

The problem here is unaccountable forms of voting, like touch screen systems with buggy code and no paper trail. As long as you have a paper trail, you're fine in my book. That the voting machine companies would fight this is amazing, but their motivation is most likely cost.
posted by delmoi at 7:48 AM on September 22, 2006

Er, delmoi, Western Infidels above pointed out why the paper trail is a useless defence. In summary, again, a printout from one of these election-theft-droids doesn't have to correlate with the actual vote registered. You can almost hear the Dieboldites snigger as they prepare to accept that "safeguard". Just one more memory card away from some stolen mid-terms...
posted by imperium at 7:53 AM on September 22, 2006

butterstick and pracowity:

the diebold machine i worked with had these features, other then a tally sheet for you to take home. one aspect of elections that makes fraud easier is the fact that the vote itself is anonymous--who you voted for is not something we can track. i'm pretty sure that the electronic version can't track that either--but i'd have to review that proceedure again to be sure.

the shredder is a bad idea. judges of election return spoiled ballots and whatnot in sealed envelopes.
posted by lester's sock puppet at 7:54 AM on September 22, 2006

That the voting machine companies would fight this is amazing, but their motivation is most likely cost.

You don't think having the power to change election outcomes to make them more amenable to their business interests would be enough motivation? Or are you factoring that into cost?

Now can we please talk about the new information in this article?
posted by saulgoodman at 7:57 AM on September 22, 2006

spilon: I'm about 90% sure that Democrats ARE hacking the vote -- I'm looking at you Maryland.
posted by empath at 7:58 AM on September 22, 2006

And, I say that as a died-in-the-wool liberal Democrat. I just know that local machine politics in my neck of the woods are dirty, dirty, dirty. That has nothing to do with the national party. The recent fiasco with Al Wynn's primary theft from Donna Edwards is a case in point.
posted by empath at 8:00 AM on September 22, 2006

so it is perfectly normal when 48 - 39 becomes 52-45 ?

First of all that was not an exit poll that was "The most recent Mason Dixon Poll". Most recent meaning before the election, and Mason Dixon doesn't do exit polling, as far as I know.

pre-election polls and the general election differ all the time, because pollsters use 'turnout models' to try to predict who will go to the polls. Turnout models improve the accuracy of the polls in general, but if something unusual happens in an election the model won't work (look at the CT primary as an example. Lamont was ahead by like 10% and only one by 3%. Pollsters were expecting only hardcore democrats to come out like a normal primary, and instead about 43% of democrats showed up at the polls).

Exit polling doesn't have that problem, but you didn't point out an exit poll, you pointed out a pre-election poll which are wrong all the time.

And to answer your question, yes it is normal for polls to be wrong outside their margin or error at least 5% of the time. If you look at 100 polls you're probably going to find about 5, which is way off. Exit polls are not given to everyone, they are a random sampling just like any other poll and so some of them are going to be wrong.

Think about it like a game of poker. You have an ace high flush and there are pairs on the table. You think you're going to win, but every once in a while you're opponent is going to have a full house. It's not likely, but it will happen every once in a while. (That happened to me a while back and I still remember it. Bleh :P)
posted by delmoi at 8:01 AM on September 22, 2006

mulligan - that Atlanta Journal and Constitution poll referenced in your quote is a regular crappy, tiny sample size poll taken 4 days before the election. It's not an exit poll. We already know that different polls taken days before an election can have4 wildly different results.

And John Zogby is a salesman selling polling. To say no poll predicted what happened should make you draw the inference that the polls are inaccurate, not that reality is inaccurate.
posted by Pastabagel at 8:01 AM on September 22, 2006

what delmoi said
posted by Pastabagel at 8:02 AM on September 22, 2006

Er, delmoi, Western Infidels above pointed out why the paper trail is a useless defence. In summary, again, a printout from one of these election-theft-droids doesn't have to correlate with the actual vote registered. You can almost hear the Dieboldites snigger as they prepare to accept that

I'm talking about a voter verified paper trail, of each vote. Not just a count. I'm not retarded.

I think optical scan ballots are the best things, as they can be counted by machines and recounted (if need be) by people as well.
posted by delmoi at 8:04 AM on September 22, 2006

How about a law that requires stating a margin of error for each vote counting machine, and then a requirement that every election with a margin of victory falling within the margin of error must be decided instead by coin toss (or pistols at high noon, whatever)?
posted by Pastabagel at 8:08 AM on September 22, 2006

So let me get this right, delmoi, you said that:
The problem here is unaccountable forms of voting, like touch screen systems with buggy code and no paper trail. As long as you have a paper trail, you're fine in my book.
If you mean you don't support any touch-screen system, not just those with no paper trail, then I apologise for misunderstanding you. If you think a touch-screen system, in particular a Diebold one, can be made secure, please let me know how.
posted by imperium at 8:08 AM on September 22, 2006

Western Infidels: The Diebold machines use a standard PCMCIA memory card that fits in any laptop. This is th card that had the patch code on it. You don't have to be a genius to stick that card in a laptop and copy whatever is on it then post it on a blog somewhere.
posted by StarForce5 at 8:19 AM on September 22, 2006

If you mean you don't support any touch-screen system, not just those with no paper trail, then I apologies for misunderstanding you. If you think a touch-screen system, in particular a Diebold one, can be made secure, please let me know how.

There is a kind of touch screen system where after the user votes, a paper ballot comes out of the machine, indicated how they voted. The voter can read the printed ballot and approve of it. It seems like this method would be just as secure as any other paper system.
posted by delmoi at 8:22 AM on September 22, 2006

Delmoi made the point in a previous thread that it doesn't matter if the system is bulletproof if it cannot be seen to be bulletproof. And this is neither. Would you rather have a computer decide who gets the parachute, or would you rather draw straws? How you vote is your business, but I can visualize Harper jumping on the Diebold bandwagon. Just wait.
posted by weapons-grade pandemonium at 8:25 AM on September 22, 2006

Right Del. I mean you get a receipt from your ATM, right? States how much you took out and how much you had left (in my case those two figures are generally $20 and $1.57, respectively).

Why the fuck can't you get a receipt from the electro-ballot box? Especially considering that Diebold makes ATMs...
posted by Mister_A at 8:26 AM on September 22, 2006

The problem with this: If the election is rigged, what are they going to do? Invite everybody to bring back their receipts so they can count them again? They can't even give away all the lottery prizes. People have lost or forgotten to check winning tickets for millions of dollars.
posted by weapons-grade pandemonium at 8:32 AM on September 22, 2006

Well you could drop one receipt into a lockbox and keep one for your ownsome. Then if there's any notion of irregularity, crack open the lockbox.

The point in letting the voter have a receipt is to make sure that his/her intentions are captured by the ballot. Thus, if you goof up and vote for the wrong guy, or the machine is malfunctioning, you can get a do-over.
posted by Mister_A at 8:35 AM on September 22, 2006

ANd I don't want to hear any crap about how hard this is. It is not hard. GOing to the moon is hard. Making Cameron Diaz look pretty is hard. THis shit is easy.
posted by Mister_A at 8:36 AM on September 22, 2006

it is normal for polls to be wrong outside their margin or error at least 5% of the time.

Yes it is - but not in exit polls (until recently). Google Dr. Steven Freeman, University of Pennsylvania and read, read, read.

According to Dr. Freeman (considered theexpert in this area) says, historically, incumbents track +/- 1/2 a point and challengers break +/- 4 points. Exit polls in the last three German national elections, combined, have been off only .026%.

Exit polls are not typical polls. Their sample size is much greater, and the respondents are all actual voters. Like it or not, the large discrepancies of the recent past cannot be so blithely dismissed.

I am not saying with certainty that illegality has taken place, but applying the data we have and Occam's Razor, I believe that is the most likely explanation.

P.S.- This is, most certainly, not a partisan issue. Everyone should want and demand a transparent and verifiable voting procedure. Whether they work or not, voting machines do not provide this.
posted by Benny Andajetz at 8:37 AM on September 22, 2006

OK, that sort of paper trail is really not going to help. Let's assume for the sake of argument that all the whistle-blowers are right and Diebold regularly fixes elections. If we went your way then their machine could show a voter a picture of Kerry's cheesy grin, give them a piece of paper that says "You voted for Kerry! Congratulations!", and still record a Bush vote. That's the problem, no? Plus what w-g p said above.
posted by imperium at 8:38 AM on September 22, 2006

So you make it a routine to count the ballots from the lockbox at the conclusion of the election. It's not that hard, people. The electronic tallies can be used as non-official projections. It's not that hard.

Or just mandate a count of the paper ballots whenever exit polls show the race is w/in 5% margin... it's not that hard. This is the easiest thing we've ever been tasked with as a nation. If your legislator is not pushing for election reform, throw him the fuck out.

There is NO TECHNOLOGICAL CHALLENGE to running a free and fair election. The challenge lies in overcoming the entrenched corruption in govt. I'm not even talking about outright vote stealing, just the chumminess between "our" "elected" "leaders" and the various "captains of industry" who are bleeding us dry. Take the money out of politics, and all this shit disappears.
posted by Mister_A at 8:45 AM on September 22, 2006

card in a laptop and copy whatever is on it then post it on a blog somewhere.

Except that what's on the card will be compiled code, which is pretty much indecipherable. You need the source code to tell what the code is doing.
posted by saulgoodman at 8:47 AM on September 22, 2006

Correct me if I'm wrong, but wouldn't the natural laws of statistics mean that most polls outside the margin of error are only slightly outside, much like a bell curve? I can only see four explanations for a 48-39 turning into a 45-52, or 49-44 into 46-53.

1- The pollsters did a horrible job. Considering that this was not an isolated incident, it's highly unlikely.
2- Statistical normal distribution is not applicable to voting. Possible, but unlikely. It certainly seems to work most of the time.
3- All of the undecideds went for the underdog, on top of the margin of error working in their favor. Not exactly realistic.
4- Vote fraud, which has been known to exist in democracy, including American democracy, and which would explain the handful of unreasonable outliers. Not to mention this.

"In the letter, O'Dell said he was "committed to helping Ohio deliver its electoral votes to the president.""

If you talk about doing it, and it happens, don't be surprised when you're a suspect.

Both parties do it. We shouldn't tolerate it from either. Demand transparency and accountability from everyone.
posted by Saydur at 8:47 AM on September 22, 2006

Democrats should simply hack-back against the neocon’s banana republic machines. It’s been widely proposed, after all, that hackers do magic and elect Hugo Chavez as our next president.

If that doesn’t work, maybe the generals (who have sworn to uphold our former constitution) will take a tip from the coup in Thailand.

Yet, I hope we aren’t so totally screwed that a military coup becomes necessary. But then on the bright side, I hear the whores in Bangkok’s the red-light district never missed a beat through that little regime change.
posted by BillyElmore at 8:52 AM on September 22, 2006

Demand transparency and accountability from everyone.

Yes indeed Saydur, and make this an election issue. Write your elected reps. from county commissioner on up and tell them that you expect them to fix this problem or else. We need a paper trail for every vote cast in every election in this country. It's not that hard. Politicians who won't work toward this goal, a goal that represents the very essence of this country, do not deserve to stay in office.

I know, I know, what if they fix the election, right? You can only make "little" fixes without drawing attention. If every poll, incl. exit polls, has been 60-40 against the incumbent, and then all of a sudden he/she wins 50-49, people smell the rat, and then you're in the realm of criminal justice.
posted by Mister_A at 8:54 AM on September 22, 2006

I'm reading the Steve Freeman article again, but right away I notice two things: 1) RFK appears to have stolen the opening line for his rolling stone article, and 2) he's relies on the same CNN data for his paper that I cited above. Freeman, however, claims to use uncorrected raw data collected from the sites, suggeting in his paper that the corrections change the results. But if you look at pages 4-6 of the paper more closely you see that he isn't certain whether the data he has is partially corrected or not, or whether he has all of it:

These data were reportedly not intended for public release,10 and were available to
late evening election night viewers only because a computer glitch prevented NEP from making
updates sometime around 8:30 p.m. that evening.11 They were collected by Jonathon Simon, a
former political survey research analyst, and are corroborated by saved screen shots (see Figure
1.1). I happened to have sixteen CNN exit poll pages stored in my computer memory from viewing
the evening before, and in each case, his figures are identical to mine. The numbers are also
roughly consistent with those released elsewhere (Appendix B shows Slate numbers at 7:28 EST).

Furthermore, you notice looking at the exit poll data that there isn't a raw number - there is no singlke statistic that shows how many voted for bush vs kerry - its cross cut into various demographic groups. But what Freeman does is statistically bogus:

To derive the “predicted values” used in Tables 1.1 and 1.5, I combine the male and female
vote, weighted for their percentage of the electorate. Ohio exit poll data (Figure 1.1) indicate that
51% of men and 53% of women voted for Kerry. Since the electorate is 47% male/53% female,
Kerry’s overall share of the exit poll was calculated as (51% x 47%) + (53% x 53%) or 52.1%.

But the percentage of M and F's in the electorate is not the right statistic to use. He should use the percentage of males and females that actually voted, not the percentage that were eligible to vote (polling all vs. likely voters, for example).

Like I said, I'm still working through it, but with each page, his case gets weaker and weaker. IT amounts to predicting the weather, the prediction being wrong, and then the following day questioning your memory of what the actual whether was....
posted by Pastabagel at 8:59 AM on September 22, 2006

...people smell the rat, and then you're in the realm of criminal justice.
Unlike the realm we're in now with previous frauds, somehow? Although I agree your proposals to pressure candidates are a good part of the solution.
posted by imperium at 9:00 AM on September 22, 2006

You can only make "little" fixes without drawing attention.

That's the beauty of hackable machines- they are just the end of a long chain of "small fixes". First, you gerrymander as much competition away as possible. Then, you legally disenfranchise as many as possible (see picture ID brouhaha). Then you apply the secret sauce to a relatively few machines in tightly contested races.

Everyone's aware that 90% of the questionable races were in Florida, Georgia and Ohio - and they all broke Republican??!!
posted by Benny Andajetz at 9:03 AM on September 22, 2006

Exit polls are very reliable as many here have stated.

The exit polls in 2000 Florida had Al Gore as the winner and the good hair teleprompter stenographers on TV just couldn't figure out why the polls were so wrong!
posted by nofundy at 9:13 AM on September 22, 2006

Ok, throwing my 2 cents, grew up with another electoral system into the ring/outsider to the U.S. (I'm Canadian) perspective...

Ok, so I'm thinking paper trails are out - could someone explain why paper ballots are out? I'm not looking for snarks - yes, they take time to count. Big deal - so you put off your election result for a few days. I think people would be willing to wait a few days for such an important task as ensuring your democracy has transparent and accountable elections (which century are we living in again?) The benefit of paper is that people can physically count the ballots where you've marked your choice; granted, we're a 10th of the size, but really - can someone think of a viable objection to not using paper? Seriously? Bueller? Anyone?

Isn't part of the proper and wise use of technology actually knowning when not to use it? Or is this the good ol' technological fallacy in American popular thought thing - that process has to be computerized to be legitimate? I suppose it's too much to ask for the U.S. system to switch to paper ballots now, though, right?


posted by rmm at 9:15 AM on September 22, 2006 [1 favorite]

Smoking gun or no smoking gun, Democrats having to hack the system (wtf?!) or not hack the system, seriously, someone help me with this. Sorry to get all emotional too early in the morning, but this just boggles my mind.
posted by rmm at 9:17 AM on September 22, 2006

I'm with rmm. We in the UK use paper ballots for all purposes, and it works. It doesn't take any longer to count for nearly 300m people as it does for 60m people. You just need five times the staff. If the US government can spend $80+bn "spreading democracy" in Iraq, surely they can afford a coupla days overtime in Ohio?
posted by imperium at 9:24 AM on September 22, 2006

If the US government can spend $80+bn "spreading democracy" in Iraq, surely they can afford a coupla days overtime in Ohio?

That would be true if the US Government believed democracy was as important as the War on Terror.
posted by NationalKato at 9:27 AM on September 22, 2006

I agree with the sentiment that this is really not that difficult.

There are two things computers do really well...they count, and they print. This situation is fucking ridiculous.
posted by edverb at 9:33 AM on September 22, 2006

I agree with StarForce5, this is not a smoking gun. .
posted by Pastabagel at 6:37 AM PST

And if there WAS a 'smoking gun', what would you do?
posted by rough ashlar at 9:42 AM on September 22, 2006

The reason why a paper version (on top of the electronic count) works is this: Obviously, if there is some suspicion all the paper version ballots can be recounted. However, there is the further requirement that various random samples of the paper have to be counted (every election) which should (statistically) match the electronic count. An amazingly small percentage of the paper has to be counted (like .1 % or something) in order to detect fraud on the part of the electronic count. This of course assumes people *read* the paper version before dropping it in the box...
posted by R343L at 9:48 AM on September 22, 2006

This thread brought to you courtesy of Al Gore.

Thanks Al.
posted by tadellin at 9:51 AM on September 22, 2006

agree with StarForce5, this is not a smoking gun. .
posted by Pastabagel at 6:37 AM PST

And if there WAS a 'smoking gun', what would you do?

Just what do you people consider a smoking gun anyway? Even a literal smoking gun wouldn't meet the standard of evidence held forth by some of you--I mean, geez, just because someone's holding a smoking gun, that's not necessarily proof they fired it, or that it's the same gun used in the murder--for all we know, the gun was used in self-defense, etc., etc...
posted by saulgoodman at 9:52 AM on September 22, 2006

Yes Saul, sometimes if it looks like a duck, etc. It's high time we got the duck out of here!
posted by Mister_A at 9:59 AM on September 22, 2006

Very suspicious but still no smoking gun

The problem isn't the lack of a "smoking gun" in hand — the problem is that someone can walk up to the voting machine, reprogram it without any record of changes made, and walk away.
posted by Blazecock Pileon at 10:05 AM on September 22, 2006

Well, a smoking gun would be a witness/ document indicating that person x told person y to do Z to a machine before an election, Z screws with the votes, and the machine was used in an election.

Though I suppose having Z present on a machine used at a polling place would be enough too.

Now that you have this, you (the police, legislature, etc) can investigate this, bring people in, question them, grand juries, etc.
posted by Pastabagel at 10:12 AM on September 22, 2006

Then, one muggy day in mid-August, Hood (Y) was surprised to see the president of Diebold's election unit, Bob Urosevich, (X) arrive in Georgia from his headquarters in Texas. With the primaries looming, Urosevich was personally distributing a "patch," (Z) a little piece of software designed to correct glitches in the computer program. "We were told that it was intended to fix the clock in the system, which it didn't do," Hood says. "The curious thing is the very swift, covert way this was done." (Z)

"It was an unauthorized patch,(Z) and they were trying to keep it secret from the state," Hood (Y) told me. "We were told not to talk to county personnel about it. I received instructions directly from Urosevich. (X) It was very unusual that a president of the company would give an order like that and be involved at that level."
posted by saulgoodman at 10:27 AM on September 22, 2006

Well, a smoking gun would be a witness/ document indicating that person x told person y to do Z to a machine before an election, Z screws with the votes, and the machine was used in an election.

Meet Clint Curtis. Link goes to sworn affadavits, testimony under oath, video & transcripts.
posted by edverb at 10:29 AM on September 22, 2006

Give each voter a computer generated unique number for the election (not like a SS#, just for this poll).

Have the voting machines hooked up to a giant verifiable database on the Internet, where each vote is tied into that unique number.

Go on the Internet after the election, and just double check to make sure that they have your vote the same on your printout and in the database.

If the two don't match, then there has been tampering.

It's not that difficult.
posted by MythMaker at 10:31 AM on September 22, 2006 [1 favorite]

It's not that difficult.

Well, there are subtleties that your scheme does not cover. However, the gist of the plan is along the right lines and the fact that what is required is so plain, and yet seemingly purposefully not implemented is in itself enough to raise eyebrows.

Everyone knows what a good voting system looks like and what has been purchased for us is the opposite. You could hardly make the voting process more vulnerable and less transparent. There is not one defender of this system outside of the vendors.
posted by sonofsamiam at 10:40 AM on September 22, 2006

MythMaker: Perfect, except for one thing. How to prevent vote buying/selling in your proposed system?
posted by saulgoodman at 10:40 AM on September 22, 2006

tiny sample size poll
---
Exit polls are not typical polls. Their sample size is much greater

To clarify, size of the sample doesn't really matter nearly as much as people think. 1500-2500 respondents in a sample is quite sufficient for populations of 100000, one million or a hundred million.

Exit polls do not have 'much greater' sample sizes, nor would it really help if they did.
posted by spaltavian at 10:43 AM on September 22, 2006

My second sentence in the last post should read: "1500-2500 respondents in a sample is quite sufficient for target populations of 100000, one million or a hundred million."
posted by spaltavian at 10:45 AM on September 22, 2006

How to prevent vote buying/selling in your proposed system?
posted by saulgoodman at 10:40 AM PST


Given the lack of concern over the buying of votes in Congress, why not allow the citizens to get in on that sweet vote buying action?
posted by rough ashlar at 10:50 AM on September 22, 2006

Ohio - Exit poll data predicts Bush win - voter 18-64 are split 50-50, voters 65 and over Bush 58, kerry 42.

That exit poll was modified after the fact to reflect the "actual" outcome.

In reality, (t)here was a whopping 11.7% difference between exit polls and reported vote margins. [PDF]
posted by If I Had An Anus at 10:52 AM on September 22, 2006

According to Dr. Freeman (considered theexpert in this area)

Huh? He's a business school prof who happened to look into it. I'm not aware of him having any particular reputation, good or bad, as a pollster or as an analyst of voting behavior.

Exit polls are not typical polls. Their sample size is much greater, and the respondents are all actual voters. Like it or not, the large discrepancies of the recent past cannot be so blithely dismissed.

I still suspect that some -- but by no means all -- of the difference we've seen is just that exit polls are polls on election day, and the rise of early voting and no-excuse-needed absentee voting has meant that lots more people are voting before election day. If those people skew Republican, you'll get election outcomes that are more Republican than the exit polls.
posted by ROU_Xenophobe at 10:57 AM on September 22, 2006

To clarify, size of the sample doesn't really matter nearly as much as people think. 1500-2500 respondents in a sample is quite sufficient for populations of 100000, one million or a hundred million.

In statistical sampling, larger is always more accurate - see The Law of Large Numbers.

In exit polling, the respondents are also all actual voters - a point you conveniently neglected. Also, for meaningful geographical voting info, samples need to be much higher than you imply.

My point stands: Exit polling has, historically, been very accurate - until recently.
posted by Benny Andajetz at 10:57 AM on September 22, 2006

The "receipts" have to stay at the polling center for all kinds of reasons, but the two main ones are prevention of vote selling, and hard copy for manual recounts.

Any other "solution" is no solution at all.

I'm a big fan of the computer records the vote, prints hard copy, voter reviews hard copy, hard copy is locked in ballot box for recounts if necessary plan.

This is not hard, it is achievable with current technology, and should meet any reasonable inspection by neutral parties.

So, you get the benefit of immediate tallies and up-to-the-last-minute ballot corrections of computers, with the security and tangibility (is that a word?) of paper ballots, but yet no hanging chads or hand drawn X's that cover two boxes.

Where is the failing of this system?
posted by Ynoxas at 11:03 AM on September 22, 2006

Ok, so I'm thinking paper trails are out - could someone explain why paper ballots are out?

Rather than being out, paper ballots are the most common form of voting used in the US. Granted, they're optically scanned unless a recount is needed.

Why don't we hand-count all of them?

Because our ballots aren't like your national ballots. We don't have a single, solitary office with a few candidates up for grabs.

Our ballots have a mishmash of federal, state, and local offices, often with an additional section of state and local referenda and initiatives. It's not uncommon for there to be 50--100 separate elections or ballot questions on a given ballot.

Hand counting all of those means either using 50--100 times the people of a Canada/UK style of election, which means lots of people handling the ballot any one of which might have some bad intent, or taking 50--100 times as long. And it introduces significant new possibilities for error -- in a UK or Canadian parliamentary election, you don't have to worry about someone looking at the vote for one office and entering it as a vote for another, but that's a nontrivial possibility with big, messy ballots.

These are the same reasons why some Canadian jurisdictions use optical scanning for local elections.
posted by ROU_Xenophobe at 11:03 AM on September 22, 2006 [1 favorite]

I'm all for the simple solution: do away with electronic voting machines altogether and go back to what really was working all along: paper ballots counted by people, not 'central tabulators'. This whole obsession with making electronic voting machines work is nonsense. The owners of the companies have very clear ideological axes to grind and have been caught multiple times grinding them. Anyone who thinks those machines have any legitimacy in providing a fair election deserves the government that has been rigged for (or against) them.
posted by mk1gti at 11:04 AM on September 22, 2006

Beyond the shenanigans with the 2002 Georgia election that mulligan has discussed above, there was also the Voter News Service debacle on election night, when VNS announced about 5:30 p.m. on election day that they were not satisfied with their exit polling results, and shut down reporting for the whole state. Essentially, that left no statewide independent news organization based exit polling going on in Georgia on election night, and that was an important unofficial "audit" control that called the results into question for many of us.

2 statewide races in that election, that ultimately went Republican (the governors race and the Chambliss-Cleland Senate race) were clearly impacted by 100 memory cards in heavily Democratic precints in Atlanta that "went missing," (story of interest is about 2/3 down the linked page, from the Nov. 19, 2003 post) as well as various problems with machines around Savannah polling places, where voting was delayed throughout the morning, and many voters left, without being given any other means of voting. And many of us remarked immediately to one another about the absurd reversals of reputable pre-election polls taken not only by the AJC, but by other news organizations around the state.

Not only was there no independent exit polling on the evening of that election, but the Diebold system in use was bought by a Secretary of State (Cathy Cox) who quickly became a "recommender" quoted on the Diebold Web site, and who rammed through the $54 million purchase of the Diebold system with what many considered insufficient testing, or staff training. When I asked poll workers at my precinct in Buford, GA for a demo of the machines in 2001, at the last official paper election conducted, not one of the poll workers there could answer any question I posed about security, or vote total recording.

Georgia is renowned as a state with a long history of questionable election practices, including dead men voting twice or three times, and it's a state with a high tolerance for heavy handed manipulation of the electoral process. The Diebold system just makes sure that there is now no independently auditable vote result trail to bring up in post election lawsuits.

"Will the Next Election be Hacked?"

Sure. And with the practice and experience of 2002 and 2004, it'll be hacked better and cleaner than ever.
posted by paulsc at 12:31 PM on September 22, 2006

Here in Santa Barbara, we use scantrons. You fill them out with a felt pen, they're counted electronically, and dropped into a lockbox. Anyone can look at them and read them to do a manual recount.

Electronic counting with a verifiable paper trail, and it's easy and quick. I like the idea of manditory recounts when the exit polls disagree with the electronic count, as well as randomly-chosen recounts. None of this is hard to understand or hard to implement. The only reason not to do it is precisely because the powers that be want insecure elections.
posted by Humanzee at 12:57 PM on September 22, 2006

In statistical sampling, larger is always more accurate - see The Law of Large Numbers

The increase in accuracy after a few thousands is absolutely trivial, while the cost continues to grow. This is why you'll never see polls with many more respondents than this. Large isn't usefully more accurate than past this point.

In exit polling, the respondents are also all actual voters - a point you conveniently neglected.

I didn't neglect it; I didn't address it because I don't disagree.

Also, for meaningful geographical voting info, samples need to be much higher than you imply.

Only because you are then dealing with multiple target populations, each needing at least the minimum needed proportion of respondents to total population. Exit polling is usually more accurate because of this, which has a side effect causes the total sample group to be large. This size increase in of itself has no substantial effect on the poll's accuracy.
posted by spaltavian at 2:17 PM on September 22, 2006

The increase in accuracy after a few thousands is absolutely trivial, while the cost continues to grow. This is why you'll never see polls with many more respondents than this. Large isn't usefully more accurate than past this point.

That's not really accurate.

You'll see samples of more than ~1500 when you want to make inferences about subpopulations. If you want a random sample where you want to say something about the population as a whole but also to say something about black respondents with a reasonable margin of error, you might need 5000-10000 respondents.

This is why things like the Current Population Survey takes about 50,000 responses every month.
posted by ROU_Xenophobe at 3:33 PM on September 22, 2006

Well, the last two elections were hacked in one way or another, so I don't see why this one will be any different.

I think I'm gonna try to hack it so that the Republocrats don't get enough votes to get on the ballot and the Greens and Libertarians do. That should be interesting.
posted by nyxxxx at 3:34 PM on September 22, 2006

Why don't we hand-count all of them?

Because our ballots aren't like your national ballots. We don't have a single, solitary office with a few candidates up for grabs....Hand counting all of those means either using 50--100 times the people of a Canada/UK style of election, which means lots of people handling the ballot any one of which might have some bad intent, or taking 50--100 times as long.

So, em, what did you guys use pre-computer age? Or did real democracy arrive in 1970?

I don't recall a long delayed announcement of Nixon's first victory, or Kennedys (granted, both fiddled results in certain locales: they didn't have the power to pervert an entire election.)

GOP + Diebold = Integrity? I don't think so.
posted by dash_slot- at 3:46 PM on September 22, 2006

So, em, what did you guys use pre-computer age?

Mechanical voting machines, mostly. They became common not long after the US adopted the Australian ballot in the late 1800s.
posted by ROU_Xenophobe at 4:57 PM on September 22, 2006

One thing is undoubtedly true: the rest of the world looks at the USA in astonishment.
posted by five fresh fish at 6:46 PM on September 22, 2006

Will the Next Election be Hacked?

The last 2 where, why should the next one be different?
posted by signal at 8:20 PM on September 22, 2006

If I understand correctly, for some posters none of the individual pieces of evidence reaches the level of indicating a problem that sorely needs to be fixed. Fair enough. I am a scientist, and I do randomized trials of individuals. If I look at the results for each individual, there is just not enough evidence there for me to draw a conclusion. It's only when I aggregate all of the individual pieces of evidence that I see the pattern and can draw a conclusion. At what point do these items individually suspicious, but not rising to the level of evidence by themselves aggregate to a problem? I'm not seeing any criteria by which you would judge that a problem exists.

Now, either you have a vested interest in never achieving that level (e.g., maybe you like the outcomes of the hinky election results) or you have an extremely high level of proof needed before you demand a thorough investigation of the mechanism at the heart of our democracy. Which is it, guys? I'm curious to know.
posted by Mental Wimp at 4:17 PM on September 23, 2006

Consider also that some people are so invested in their "party" that they will never admit that it has failed/been hijacked; many people are so invested in their country that they will never admit there are deep flaws in its political structure/mechanism; many people are so invested in the myth of American Greatness that they will never admit otherwise.

If I recall correctly, the Italians never solved their Mussolini problem. Were they not an Axis (ie. evil) power during the length of the war?

The USA will be the same.
posted by five fresh fish at 5:14 PM on September 23, 2006

The October Surprise
posted by homunculus at 1:02 PM on September 24, 2006