We worried quite a lot about the integrity of the vote in the midterm elections. Steven D did most of the writing on the topic, and it might seem like we were being alarmist since the elections went off relatively smoothly. That’s not the case in Florida’s 13-District. The Orlando Sentinel has taken a look at the 17,846 ballots that registered no vote in the House race. Before I get into what they found, I just want to make a few points.
The district includes the city of Sarasota, but also parts of four other counties. The undervotes are six-times higher in Sarasota than they are in the other counties. The Republican, Vern Buchanan, was certified the winner by state officials on Monday. He beat Christine Jennings, the Democrat, by 369 votes. Jennings actually carried Sarasota County by a 52-47 margin.
So, just looking at the high level of undervotes in Sarasota leads me to believe that she would have won the election if there had not been a very high and unexplained number of undervotes in that county.
Let’s look at what the Sentinel discovered:
The Sentinel reviewed records of 17,846 touch-screen ballots that included no vote in the tightly contested 13th District congressional race to determine whom voters selected in other major races.
The analysis of the so-called “undervotes” examined the races for U.S. Senate, governor, attorney general, chief financial officer and agriculture commissioner.
The results showed that the undervoted ballots skewed Democratic in all of those races, even in the three races in which the county as a whole went Republican.
In the governor’s race, for example, Republican Charlie Crist won handily in Sarasota, easily beating Democrat Jim Davis. But on the undervoted ballots, Davis finished ahead by almost 7 percentage points.
In the agriculture commissioner’s race, Republican Charles Bronson beat Copeland by a double-digit margin among all voters. But on the undervoted ballots, Copeland won by about 3 percentage points…
…About 15 percent of ballots cast on Sarasota’s touch-screen machines registered no choice in the bitterly fought congressional race. That percentage was about six times greater than the undervote in the rest of the House district, which spreads into four other counties.
Since Election Day, dozens — if not hundreds — of voters have reported problems at the polls. Some say their vote for Jennings never registered after they touched her name. Others say they never saw the congressional race on the machine’s screen.
It’s hard to estimate how many votes Jennings lost because I don’t have a breakdown of how many undervotes were in each county. To get an idea I will use the assumption that all of the undervotes were in Sarasota (they weren’t) and use Jenning’s 52-47 margin there. Fifty-two percent of 17,846 is 9,280. Forty-seven percent of 17,846 is 8,218. The difference of 1062 votes is almost triple the certified margin of victory (in the other direction).
This number, although based on a faulty assumption, is high enough to suggest that the wrong candidate has been certified as the winner. The Sentinel’s study, however, suggests that the problem specifically targeted Democratic voters. It’s not clear how this was done. But their study suggests that the missing votes skew to Jennings by a significantly higher margin than 52-47.
Here is how things stand right now:
A circuit judge in Leon County District Court, William L. Gary, ruled that the state must move quickly to test Sarasota County’s paperless e-voting machines to help determine what, if anything went wrong in the undervote for Florida’s Congressional District 13
At the same time he said Plaintiff, Democratic Candidate Christine Jennings must include Election Systems & Software, the company that manufactures the paperless iVotronic machines, so they can have input on potentially propietary information. Jennings’ lawyers have also indicated they would be seeking to investigate the elections software source code, which Gary’s order allows.
Monday, Jennings filed suit against the county and state canvassing boards and included her Republican opponent Vern Buchanan, who was declared the winner in the race by 369 votes. At question is roughly 17,000 votes (other sources say more than 18,000) that were not cast for District 13, while votes for other political offices were counted. Faulty programming is suspected. Jennings news releases and statements from her lawyers Kendall Coffey and Mark Herron also give varying figures.
The state had already had scheduled a test of the equipment for next Tuesday. Both sides have now been invited, via the judge’s order, to look in on the testing, if they choose.
The Palm Beach Post editorializes:
To help determine whether there was anything wrong with touch-screen machines that produced an abnormally high number of no-votes in the five-county District 13 race, Gov. Bush tapped Alec Yasinsac, a Florida State University professor. A week before the Supreme Court decided the 2000 race, Mr. Yasinsac proudly proclaimed, “I’ll never be a passive political participant again” while wearing a “Bush Won” button. More recently, he supported Republican Tom Gallagher for governor.
Gov. Bush defended his choice by noting that Leon County Elections Supervisor Ion Sancho once hired Mr. Yasinsac to review computer code, The Miami Herald reported. A spokeswoman for Secretary of State Sue Cobb, however, told the Herald that Mr. Yasinsac had been hired because “he was ‘based locally,’ has strong credentials and approached the office to be a vendor.”
Mr. Yasinsac is supposed to assure the accuracy of touch-screen machines that some voters question because no paper records exist to document results. Voters expect impartial work, but an acquaintance of Mr. Yasinsac’s who works for the liberal People for the American Way called him “a strong advocate for electronic voting machines and a vociferous opponent of requiring a voter verifiable paper trail.”
Gov. Bush and the Cabinet certified Republican Vern Buchanan the District 13 winner Monday after a recount showed him 369 votes ahead of Democrat Christine Jennings. She promptly sued for a revote – remember that from 2000? She’s upset because more than 18,000 ballots in Sarasota County, which she won by 53 percent to 47 percent, showed no vote. Other races did not experience such high undervote counts. Fittingly, her attorney is Kendall Coffey, who represented Al Gore in 2000.
The courts didn’t grant a revote then, despite a confusing Palm Beach County ballot. This time, a revote would not have national implications, but it will be more challenging to prove machine error, not human error. Still, if the state is going to allow touch screens, the state and the counties that use them are obligated to make sure that the machines function properly. A critical, impartial review of the touch-screen machines is in the public interest, but the choice of Mr. Yasinsac makes it less likely that the public will get a credible one.
You can read more about this race here. It appears likely that the House Administration Committee will have hearings to decide whether to accept the certification of Buchanan as the winner when they convene the new Congress in January. The chairwoman of that committee will be Juanita Millender-McDonald (CA-37). It will make great theater and possible sweet justice to have a black woman chairing the hearings on a stolen election in Katherine Harris’s vacated House seat.
I wonder if the program of the voting machines targetted straight Democratic tickets. That a disproportionate number of such tickets registered an undervote is indeed very suspicious, which may explain why the company responsible for the program is not open to a program audit.
I’m sure that, with enough pressure, they’ll provide a version of the program to audit.
Any squawking about proprietary code should be easily countered through NDAs (non-disclosure agreements). Companies use these all the time; anytime I go interview someplace I have to sign an NDA just in case the company wants to tell me something the public at large (or more to the point, their competitors) aren’t supposed to know.
It would be interesting to put the raw data from the race up on the ‘Net (without any identifying voter information, if there is any) to see if the collective brainpower of a whole pile of smart people can crack the secret. I’ll bet it wouldn’t take long for someone to figure out that you had a 70% chance of not registering a vote in the race if you voted for more than four Democrats, or something like that.
Fifteen percent of the votes did not register a choice in Sarasota. That was six times higher than in the other counties, so we know it was not due to chance.
Second, the ballots that were affected were disproportionately voting for Democrats.
Voters complained of not having the option to vote for the house race and also of having their vote not register. That’s where it gets confusing.
A good hacker would not be so blatant. There’s no need to do anything visible.
It would be really dumb to write a program that did something like not offer the Congressional option for every sixth person that votes Dem for Governor and Senator. Yet, that appears to be what the defect was.
exactly, for the statewide candidates appear first, followed by the Congressional race. The program must have been written to trigger a skip or a nonregistration if certain constraints involving the statewide candidates were met. Although it may seem obvious to us, to a regular voter or to a candidate who is not tenacious this could easily be ignored.
That’s why I’d like to see the data. If you could show a chain of events like
( IF GOV_VOTE == "DEM" THEN( IF SENATE_VOTE == "DEM" THEN
(SKIP_HOUSE_VOTE)
)
)
that shows up in 16% of the ballots with democratic Governor and Senate votes and that the undervotes for House are what you’d expect everywhere else, that would be pretty strong evidence of malicious intent.
Actually, BooMan, if I were doing this it wouldn’t be “every sixth time.” If you could get a list of votes in sequence something like that would stand out like a sore thumb. I’d do (if (random(100) < 16)) to make it happen one time in every six on average.
You’d also want to make sure that the House option wouldn’t show up if someone goes back and changes their vote from Democrat to Republican in one of the “trigger” races; but now I’m getting way far afield.
It just seems hopelessly amateurish.
Why not just have the machine act normally and just count wrong? Why undercount it?
And why pick on just one race so it leaves an obvious fingerprint?
I know you can write a program that erases all traces of itself. But this seems like such a ham-handed hack that there might be traces in the code still there.
As for machine malfunction, it seems to be precluded by the uniformity of complaints coming from the Dems and the statistics on who was affected.
Take it from a professional programmer. Not all hacks are well thought through, and many have consequences the programmer didn’t intend. I’ve written more than my share of both and been victim to many others. So I can believe this was done by someone who just thought it was a good idea at the time and didn’t think it all the way through, and only had a few machines to work with before the election.
The investigation I propose would only serve to show what happened. It wouldn’t show why it happened, or why whoever did it, did it. It’s possible it could point to a genuine glitch in the code as it came from the factory, although the specificity in race and location makes me think that’s not the case.
Yes, the statistics show a simple audit may not suffice. If a deception was planned, I think a routine audit would have been part of the planning process as well.
im down here just south of sarasota right now and the letters to the editors in all the local papers are calling for a revote, and to get rid of the woman in charge of the election whose name escapes me at the moment.
my parents voted in charlotte county one county south of sarasota county i believe….they said they didnt have any trouble voting…and they finally came back to their senses and voted for the democrats.
ps its freaking cold down here.
yes, it is a very chilly environment for Democrats in FL-13.
needs to be brought up to the international standards expected of African countries. Then we wouldnt have to worry about such things. This is obvious. Howveer, is there a will to have an honest electoral system?