More Diebold problems.

November 16, 2003 8:08 AM

Anthony Argyriou uncovers what seems to be a serious problem either with California voting machines or the vote tallying system: The Secretary of State's summary of votes on the Davis recall shows three counties--Alameda, Kern, and Plumas--that apparently had zero voters who didn't vote on the recall. Not one. All three counties used Diebold machines. Other counties ranged from 0.5% to 10.3% of voters not voting on the recall. More from Rick Hasen, a top election law scholar. [Via Volokh.]
posted by monju_bosatsu (41 comments total)

Only someone with a pathological hatred of the Bush Adminstration would find this surprising or suspect.
posted by rushmc at 8:18 AM on November 16, 2003

Looks like the Diebold systems might be inserting "Default Value: Yes".

Maybe the best thing to come out of the California Recall will be bringing the Voting Machine scandal up to critical mass BEFORE November 2004. Wouldn't that be ironic?
posted by wendell at 8:44 AM on November 16, 2003

I don't think that anyone is positing this as some Republican plot -- look at Alameda, one of the questionable counties, it voted 70% against the recall -- but this result is really strange. Just look at it statistically: there's <1% chance (making up figure here, but it's low) that these results could have come about from chance.
posted by brool at 8:49 AM on November 16, 2003

What's it going to take to get this story into the mainstream? The net has been buzzing for weeks about how awful electronic voting is. I've seen a few stories in print, but where's the outrage?
posted by Nelson at 8:50 AM on November 16, 2003

Everyone who has a pathological hatred for The Bush Administration honk three time - at your earliest convenience.
posted by donfactor at 8:50 AM on November 16, 2003

Of course, if the machines' programming had inserted "Default: Yes", then maybe Alameda County would've been 72% against the recall... But only the state-wide totals counted for anything. Not close enough to affect the final outcome, but potentially very telling.

I'll just honk twice.
posted by wendell at 8:57 AM on November 16, 2003

rushmc: and why exactly one should hate King George to find this suspect ? I don't think it's surprising it's a well known fact that computer aren't 100% reliable (nothing is) and they can obscure enormous errors in a few line of codes, unlike other devices.
posted by elpapacito at 9:06 AM on November 16, 2003

Could a voter using a Diebold touchscreen even complete the process without voting on the recall? Given the results, it seems to me that the user interface simply did not allow the case of a user just voting on the other aspects of the ballot and not voting on the recall.

What I find telling is the large discrepancy in results associated with the different voting systems used. That means that the results can be tweaked quite a bit, probably in the 5% range, simply by choosing different voting systems for counties that are predominantly Republican / Democrat. Another way to tweak the results is to simply claim technical malfunction and delay the vote in some counties, so that predominantly poor voters exceed their allotted time away from the workplace (for reasons unknown to me, you Americans seem to have votes on regular workdays).

And let's be clear about one thing: Electronic voting will benefit anyone trying to cheat, not just the Republicans.
posted by Eloquence at 9:07 AM on November 16, 2003

Come now, in a recient election in Iraq, they had 100% of the people vote in favor.

Why SHOULD there be any suprise in a 100% pro-vote.
posted by rough ashlar at 9:10 AM on November 16, 2003

And California's Secretary of State (a Democrat) is getting on the case. He has ordered an audit of ALL California counties' voting systems. On the other hand, he has already certified the recall election.
both stories from LATimes, registration may be required; I'm hopelessly registered myself.
posted by wendell at 9:22 AM on November 16, 2003

Why SHOULD there be any suprise in a 100% pro-vote.

I'd like to see a 100% Pro Vote. These Amateur Voters are screwing things up badly.
posted by wendell at 9:23 AM on November 16, 2003

So... much... hatred... can't... move... arm... to... HONK!

I live in Alameda Country, and I gotta say, voting on a Diebold system while knowing they're running MS Access on WinCE made me more than a little unsure at to whether or not my vote would count.

Granted, Alameda is a super liberal country (whee!), and the results aren't too out of line, but odd results need explanations. I'll ask around and see if anyone remembers not voting on the recall issue. (Cuz hey - even one would indicate something fishy).

And what if it's not pathological hatred, per say, but rather pathological insouciance? The pathological pathology of pathos... moo hoo ha ha...
posted by Coda at 9:59 AM on November 16, 2003

rushmc, donfactor - one doesn't have to be a Bush hater to recognize that there's something truly odd in these numbers. It defies all experience that a half million people could vote in Alameda County and NOT ONE OF THEM forgot to vote for the recall. It also seems odd that you believe this is something Republicans only could take advantage of.

I hope you're not representative of those who share your political views - because if you are, you've sacrificed common sense, logic and the essential vigilance that liberty requires for partisan bickering.
posted by pyramid termite at 10:32 AM on November 16, 2003

Rehetorical question #29620: Can you be a Diebold-hater without being a Bush-hater?
posted by wendell at 10:35 AM on November 16, 2003

Thanks for pointing this out. My mom works for the Sacramento Bee (though she's a hated advertorialist) and I've sent this link to her in the hopes she can get someone in the newsroom pissed about it.
posted by aaronetc at 10:41 AM on November 16, 2003

This really needs more media attention. Good to know that the LA Times is reporting on this, but I'm amazed nothing there has been little mention in any other major media outlets (NY Times, Washington Post, TV Media)
posted by destro at 10:46 AM on November 16, 2003

I was just seeing what it felt like to jerk the old knee, but it doesn't really suit me. (I guess you should have used quotation marks.)
posted by rushmc at 10:48 AM on November 16, 2003

er, I guess *I* should have used quotation marks, rather... (looks over shoulder to see who changed that...)
posted by rushmc at 10:49 AM on November 16, 2003

Wendell: Yes, you can be suspicious of Diebold without being suspicious of Bush, even if many are suspicious of both. What I find fascinating about the whole thing is not that the machines can go wrong-- heck, the old lever machines could screw up. But, election officials could pop open the lever machines on the morning of the election and muck about to make certain that everything working properly. This can't be done with these electronic voting machines: there's no paper trail and the innards of the machine (now code) are considered proprietary. Partisanship aside, that's disconcerting, because electoral integrity requires (at the very minimum) 2 things: anonymity and transparency. And without a paper trail, and without access to how the machines work, the Diebold machines aren't equipped for transparency.
posted by .kobayashi. at 10:51 AM on November 16, 2003

I say dump all the frikkin' electronic voting machines and give me a paper ballot and a indelible marker - simple and straightforward.
posted by wsg at 11:07 AM on November 16, 2003

maybe someone can set up an online thing where voters can go to register how they voted, and if enough people do it, it can serve as a check if there are major discrepancies in Nov.04?
posted by amberglow at 11:08 AM on November 16, 2003

If the claim is that every person in Alameda county voted either yes or no on the recall, then all that needs to be done is publicize the claim. If you find one reliable person that claims not to have voted on the recall, then Diebold in trouble.
posted by rdr at 12:27 PM on November 16, 2003

whoops, I meant every one who voted in the election, not every person who in Alameda...
posted by rdr at 12:28 PM on November 16, 2003

Most likely, the machines were set to (erroenously) mark lack of vote on recall as an 'undervote', and refused to let the vote be cast. However, its not an undervote, voters were free to not vote on that question and vote for a replacement anyway.

This was the proposed explanation when this story came out the day after the election..
posted by benh57 at 12:56 PM on November 16, 2003

And let's be clear about one thing: Electronic voting will benefit anyone trying to cheat, not just the Republicans.

Eloquence: No, no, no. Electronic voting will benefit anyone trying to cheat if they control the electronic voting process. In simple terms: those with power can arrange to keep it if they're willing to cheat.

Isn't it lucky that Diebold is "a model of integrity and reporting and clarity and disclosure and consistency"?
posted by iffley at 1:00 PM on November 16, 2003

I looked up the number of registered voters per county (as of 2000), and it seems that the situation here is that everyone who went to the polling place in those counties voted on the recall, not that everyone who is a registered voter voted on the recall. I had originally interpreted the FPP as saying that those counties had a 100% turnout, which I found unbelievable. But the actual situation is a lot more plausible.

If you look at the county-by-county results [pdf], you'll see that there's a fairly wide variation in the undervote. In Sierra County, 10.3% (168 people) went to the polls but didn't vote on that question; in Alpine, only 4 people (0.7%) skipped it; the statewide average was 4.6% undervote.

If I remembered my stats classes better, I could compute the probability that the three 0%-undervote counties are legitimate. But just from eyeballing the data, they look plausible. I'd still like there to be some investigation, though, considering that both of the counties using the newest, least-accountable Diebold DRE equipment are in the 0%-undervote set. (The third county uses an older Diebold optical-scan system which does leave a paper trail.)
posted by hattifattener at 1:13 PM on November 16, 2003

Just off the back of an envelope: Kern has 8800 voters, the chance of voting at best = 99.5%, so [1 - (0.995 ^ 8800)] = about as close to 0 as you like. Not a totally legitimate analysis but enough to raise suspicions.
posted by brool at 2:17 PM on November 16, 2003

Just 0.995 ^ 8800. Sheesh.
posted by brool at 2:19 PM on November 16, 2003

hatti, no, it's not statistically plausible that you would get 0 in absolute numbers in Alameda, where there are 422269 votes total, which means that even a 0.1% undervote (much lower than in any other county except for the three 0/Diebold ones) would result in over 400 votes in the "not cast" column. That there is not a single vote in the "not cast" column for the Diebold counties indicates with virtual certainty that there is a difference in voting equipment that accounts for this. This does not mean the equipment has been tampered with, as has been explained above, but probably that it did not allow certain types of votes. It is unlikely this had an actual impact.

On a different note, Diebold touchscreens appear to eliminate the principle of secret voting, since it's trivial to get the content of both CRT and LCD screens using so-called tempest attacks (listening in on electromagnetic radiation), which requires expensive shielding to prevent (unlikely they would pay for that). In theory at least, you can now beat up people if they vote for an option you don't approve of.

Here's a conspiracy theory for you: Up until now, it was possible to rig every electrion through corporate owned media. The California recall proved that you can do virtually everything if the media play along (labeling Arnie a brute is not investigative reporting, exposing his Enron, Buffett, Rothschild etc. connections is).

The Internet is risking to make an impact in the 5% area, which could be enough to prevent a desired result from being achieved. Now, to compensate for that effect, you get Diebold voting machines that allow any predetermined result to be implemented. Pretty soon American elections will be run like professional wrestling, with millions of audience members knowing that they are faked, but enjoying the show.
posted by Eloquence at 3:50 PM on November 16, 2003

...and millions of others furious that the arena doors are locked...
posted by Mars Saxman at 6:16 PM on November 16, 2003

More than anything else, there is both the technical problems and the *market* problem: who else but Diebold makes electronic voting machines? (Shades of Halliburton.)
After the incredible hoo-ha of the Presidential election, lots of States and their semi-autonomous counties panicked with their (given) terrible systems of tabulating ballots and decided to go electronic.
So who do you call? Diebold!

And a short time later lots of experts start announcing that Diebold's software is nichts gud. Diebold's response: silence the critics with the DMCA. This is a train wreck in progress and waiting to happen!

And it's only going to get worse. Everything Diebold touches will cause a fight, as will the paper ballot alternative. Is there a single county in the US that isn't caught out with this?
posted by kablam at 6:41 PM on November 16, 2003

Diebold's CEO (thanks, iffley) says "I never imagined that people could say that just because you've got a political favorite that you might commit this treasonous felony atrocity to try to change the outcome of an election," he said. "I wouldn't and couldn't."

O'Dell was quick to point out that he has done nothing illegal. He also said he has no daily involvement with Diebold's election-systems division, which is based in Texas - and run by a registered Democrat.

He said the elections business is responsible for just $100 million of Diebold's $2.1 billion operation.

Despite his regrets, O'Dell said he will not stop supporting Bush's campaign. He said he went to Bush's fund-raising event at the invitation of the campaign.

"They had an event for Pioneers and Rangers, and I am one - and proud of it," O'Dell said. " . . . [But] if I have kind of overmixed that with my corporate entity, I feel badly about that."
 
posted by troutfishing at 7:19 PM on November 16, 2003

And it's only going to get worse. Everything Diebold touches will cause a fight, as will the paper ballot alternative. Is there a single county in the US that isn't caught out with this?

I'm envisioning lawsuits in every district nationwide. 2005 should be fun...line up your lawyers now to beat the rush.
posted by rushmc at 8:00 PM on November 16, 2003

I'm a resident of Alameda County, Berkeley to be specific, and I find the zero nonvoting rate on the recall completely ridiculous. My local polling site is in a nursing home and I was there when the Diebold touch-screen voting machines were first implemented in either 2001 or 2002. I had to wait in line a long time to cast my vote, because many voters were senior citizens who had never encountered a touch screen before in their entire lives. When you have dozens of elderly voters completely inexperienced with touch screens and who already have infirm joints in their hands, I find it absolutely impossible to believe that there wasn't at least one who failed to vote on the recall (not considering that some people might have opposed voting on the recall on principle). Since Alameda County contains both Berkeley and Oakland, two cities which lean heavily liberal and Democratic, I can't help but think something fishy is going on here.
posted by jonp72 at 11:00 PM on November 16, 2003

I write computer software for a living and have serious concerns about electronic voting machines and think they shouldn't be used, but think it's totally feasible that the voting software could have been set to require that the voter vote on the recall if they're going to vote at all. i.e. not allow submission of a ballet unless yes or no had been selected on the recall question. I'd honestly be more surprised if it wasn't set that way.

Remember, it wasn't that every eligible voter in those counties voted, but that every voter who showed up at the polls and voted for other questions also voted on the recall questions.
posted by wrffr at 11:01 PM on November 16, 2003

not allow submission of a ballet unless yes or no had been selected on the recall question

I was under the impression that leaving a particular question blank is completely legit; and that any voting system should allow a question to be left blank.
posted by kickingtheground at 11:05 PM on November 16, 2003

Actually, it looks like human error, not machine error/design.

http://electionlawblog.org/archives/000255.html and http://blog.danceslut.net/archives/000275.html
posted by wrffr at 11:14 PM on November 16, 2003

er, by "human error", i mean error getting the alameda county certified results to the secretary of state, not voter error.
posted by wrffr at 11:15 PM on November 16, 2003

Upon further reflection, Eloquence, you're right. I was assuming some sort of symmetrical distribution (a normal distribution, say), but in this case that's obviously not appropriate, if for no other reason than that a negative abstention rate is impossible.

A commenter to Argyriou's original post (the one linked from the FPP) says that he's an Alameda county voter and that according to his memory the machine either disallowed or discouraged him from skipping the recall question. Hm. This doesn't match with the links that wrffr just posted; maybe he's misremembering.

kablam: The other big purveyor of (non-auditable) touchscreen machines is Sequoia.
posted by hattifattener at 11:36 PM on November 16, 2003

There is at least some mainstream attention being pai to the issue (though no specific mention of the undercount in this Editorial) by the SJ Mercury News.
posted by billsaysthis at 11:09 AM on November 17, 2003

Anyone care to speculate on the American fascination with voting machines? Canada has equivelent population with way less density as California and we seem to get by just fine with low tech, auditable, pen and paper.

It's hard to imagine a system that is less confidence inspiring than Diebold machines with out that system being openly corupt.
posted by Mitheral at 1:25 PM on November 17, 2003