Iowa Voters

A top officer for the Iowa Secretary of State (SOS) has emailed me a defense of the voting machine certification that occurred January 30. I reported on the meeting here and more specifically here. During the certification process a volunteer voter named Penelope had used an Automark ballot marking computer. With its help she had filled out a ballot that she was unable to see due to her visual impairment.

Sandy Steinbach, who represented the SOS at the meeting, notes that “The ballot was produced correctly, even though the device required re-booting after the anomaly occurred,” as I indicated in my story.

She goes on:

“The situation has been explained to me as the AutoMARK receiving information that it could not process. Because of this, it shut itself down.”

Of course the only information the machine receives is the voter’s choice of which candidates to mark, and processing those votes correctly is the machine’s only role on election day. I don’t think Penelope did anything malicious. She was merely trying out equipment she had never encountered before. And the ballot was indeed marked. But the machine froze in the process.

Steinbach continues:

“AutoMARK has a special role in the election process. This device does not count ballots; it marks them. The ballot will be counted by an optical scan counting device.

It is not unusual for state certification testing to raise concerns that were not seen in the ITA [independent testing authority] process. We are looking at the devices in a somewhat different way from the ITA process. The fact that the anomaly was discovered and is being corrected is a good thing. That is why we do the tests. We will prepare instructions for our election officials, so they can be prepared if this rare event happens on election day.”

Let’s hope it is a rare event. But in the January 30 test only a few ballots were created on the Automark machine. This rare event occurred on the very first ballot. The vendor’s representative acknowledged something similar had happened in Illinois tests. So is it really rare? Voters may have to find out for themselves.

Rewriting software is not for beginners. But what if your ballot counter is misbehaving? Should you reprogram it? Even in the middle of a public certification meeting?

They did it Monday in the Iowa Secretary of State’s office. Election Systems and Software of Omaha was there to gain official approval of its new equipment. But after they had used the new ballot marking gadget to mark some ballots for a test election, the ballot scanner made a mistake in the tally.

The scanner reads and counts those paper ballots at the speed of light. It can be quite reliable if the ballots are properly marked and the scanner is carefully calibrated and correctly programmed. But the scanner must know the rules for counting and that can be tricky.

Now this ballot scanner was not really being tested Monday, according to the folks in the room. It was indeed on the agenda, but it had also been tested and approved last year before the newest gadget for handicapped voters was for sale. Today it was needed to read and count the ballots created by the new gadget because the examiners wanted to see that those new ballots were actually decipherable by the M100 counter.

So it wasn’t the new gadget that was tripping up the older scanner, it was the programming in the scanner. Apparently this particular test ballot had not been tried during the earlier approval process. So now, what to do?

Phone home. The ES & S men called Omaha to see what could be done. It was decided to “burn new media,” thus reprogramming the scanner on the spot. The chair of the board of examiners observed “We’ve never had to do this before. We’ve never had to reprogram.”

About an hour later the technician stuck a memory card into the M100 and tried the small stack of ballots again. Success.

When I told software tester John Washburn about this by email later, he wrote back:

WHAT!! …REPROGRAMMED WHILE WE WAITED…!!

Calmer now. Pulse returning to normal.

Calm blue oceans.

Breathe in, Breathe out.

This stark response prompted me to ask another question: Were they tampering with already certified equipment when they reprogrammed the machine? He said, “Yes.”

What have we learned from this? We know the ballot scanner worked on Monday. It did what it was told by the software it was using. We know that you can’t really tell what is in the software. We know it is easily changed if the vote count is not going to your liking.

But we can also wonder a few things. Do election officials know what constitutes tampering with their equipment? Who do they trust to mess with the programming? Do some Iowa counties now have in their possession brand new equipment that is programmed to improperly count some ballots?

Just wondering.

You would have to be blind not to see that the Automark voting machine needs more work. Actually, it was a blind woman named Penelope who first saw it in Des Moines Monday. Or at least she discovered it. The people looking over her shoulder really did see it.

Penelope had come to the meeting of the Iowa Board of Examiners of Voting Machines to test the device being examined. It marks ballots for people who need help. It helps them while keeping their votes private. It enabled Penelope to use headphones to hear the ballot read to her and provided her with a button to push to mark her choices in every race on the ballot.

No one knows what Penelope did to upset the machine. She did not know anything was wrong because the machine gave her the marked ballot at the end of her testing. But the machine had also locked up and put an error message on the screen. There was no audio error message, Penelope said.

I was sitting where I could not see the screen. But we were all told that an “argument out of range exception error” had occurred. No one knew what that meant. The machine’s advocates called headquarters to find out. They reported that this had occurred before in Illinois’s testing, and a repair to the software was already being written in Omaha.

Not to worry. Penelope was happy. She said she would cease voting absentee and actually go to the polls in the future. She said it was “about time” provisions were being made for blind voters.

The board of examiners decided the machine could stand to be rebooted if this happened on election day. They focused on the fact that no damage had been done to Penelope’s ballot. They certified the machine. Several Iowa counties are planning to use it.

But I contacted John Washburn, a software tester with a decade of experience. He said such an error “is usually indicative of bad code being passed to an interpreter of some sort.”

Bad code? Already detected in testing in Illinois? If this stuff gets past the extensive federal testing we always hear about, how extensive can it be?

We are indeed lucky to have this problem appear on the device that marks paper ballots. But this same company has also sold computer voting terminals (with NO PAPER TRAIL) to seven Iowa counties. Did they use any bad code in those machines? Was Penelope present the day those machines were tested? Penelope—-HELP US!

The room was too small and the table was too crowded. The ballot marking machine locked up once and had to be rebooted. The ballot scanner got tripped up by a test ballot and had to be reprogrammed. The voting machine managers from ES & S were unable to answer some questions about their equipment. They spent a fair amount of time on the phone to HQ getting things worked out.

That was the scene Monday in the office of the Iowa Secretary of State. Election Systems and Software of Omaha had come to town to get state certification of a new piece of voting equipment. It is intended to make it easier for blind voters and others to cast private ballots.

More ballots were cast in Pocahontas, Iowa in 2004 than were used to test the equipment Monday. There was NO testing of security, even though security concerns were raised.

The three official examiners were underpaid for their time and woefully underpaid (Iowa Code 52.6) for the responsibility they shouldered. They had already announced their plans to purchase the equipment they were about to review. They had an obvious interest in running trouble free elections, but not much curiosity about implications of the errors they uncovered. They could have benefitted from outside expertise (Iowa Code 52.5), but the Secretary of State had not provided them with any.

Nevertheless at the end of the six hour session, the equipment was approved. No surprise here. Just another step in the implementation of the Help America Vote Act, a disaster as bad as the 2000 Florida recount that it was supposed to address.

They say you should never watch the making of sausage or the crafting of legislation. Add the certifying of voting machines to the list. And the next time someone tells you that voting machines are “tested and tested and tested,” send that person to me.

The Iowa Board of Examiners for Voting Machines will meet Monday morning (Jan 30) in Des Moines.

This will be the first meeting in Iowa since Diebold ballot counting equipment was hacked in a test in Florida reported this week (finally!) in the Washington Post. Similar equipment has been approved by this Iowa board and many Iowa counties have purchased it.

Tomorrow morning the Secretary of State’s office will announce the agenda and I will repeat it here. The meeting is at 10 a. m. in the conference room on the first floor of the Lucas Building. Lucas is south east of the Capitol. Here is a map of parking lots. Look for the green ones.

I have been told by a former member of the board that public input is very helpful to them. Can you attend?

Update: Here is the agenda:

***************************************
Date: Monday, January 30, 2006
Time: 10:00 am
Place: Office of the Secretary of State
First Floor, Lucas State Office Building
321 East Twelfth Street, Des Moines 50319

Agenda: Election of Board Chairperson for 2006

Examination and testing of Election Systems & Software (ES&S) AutoMARK Voter Assist Terminal, v. 1.0; and AutoMARK Information Management System, v. 1.1.10 with Unity 2.5 M-100 precinct optical scan ballot scanner.

Discussion and action regarding application for approval for use of the above Election Systems & Software voting equipment.

New Business

***************************************

The Hursti hack in Florida last month woke up some state election departments. But apparently not in Iowa.

When it was reported that Mr. Hursti had stolen an election in full view of the election workers, many states realized that they owned voting machines just like the Diebold equipment that had been compromised in Florida. Some of them spoke right up, including Maryland’s Election Administrator Linda Lamone. She wrote to Diebold for an explanation, even asking for daily briefings from the company, saying ” . . .this matter is of great concern to election officials in the state . . .”

What was our top election official doing? Running for Governor, seeking publicity by praising local officials for purchasing very similar Diebold equipment for their own counties. He also crowed about how

Iowa continues to lead the nation on HAVA. Elections officials across the state have shown extraordinary dedication to what was an ambitious project, and it has paid off.

Not so fast, Mr. Secretary.

Has Chet Culver asked Diebold to explain the Hursti hack to him? Have they complied? Has he discussed with them the similarities between Florida’s machines and ours? Has he warned local officials who may not understand the problem as well as they should? Wouldn’t that be worthy of a press release?

Iowa’s legislature failed to enact a voter verified paper ballot bill last year because our county auditors opposed it. According to State Representative Jeff Elgin and several county auditors, the House State Government Committee got these messages:

* auditors wanted no more mandates
* paperless DRE machines were already working just fine in Spencer and elsewhere
* the bill “does not do what you think it does.”

The bill had started off strong. Senate Co-President Jack Kibbie told reporters on the opening day of the session that a spectacular failure of electronic voting in North Carolina had his attention. Even though it was 2 months after the election, a statewide NC race was undecided because a paperless voting machine had failed to record 4,000 votes.

Kibbie introduced a bill to require paper ballots be produced by DRE computers. It blasted through the Senate 48-0. One Senator admitted to me that the Senate had experienced its very own electronic voting equipment failure right there in the chamber. I’ll bet that got everyone’s attention. They could imagine how embarrassed those Carolinians must have been.

But the auditors struck back. They employed an old theme in Iowa politics–local control, sometimes expressed as “No Unfunded Mandates!” Kibbie’s bill did not contain new money for buying printers for every DRE already in use. Clay county’s auditor even hinted at a HAVA hearing in 2003 in Spencer that they were jealous that other counties would get new voting machines paid for by federal HAVA (Help America Vote Act) money, whereas Clay county had spent its own funds for its touchscreen DREs.

Marshall County election director Dawn Williams told legislators that voter verified paper was a “flawed” premise because the paper would be “produced from the very same software that records the vote electronically”. She said better technology would be coming in the future, but the current technology is “an incredibly expensive placebo.”

Most auditors ran for cover. Some would not answer emails about the subject. Some said they would get paper for their counties, but they would not back any mandate. After all, the chair of the Iowa auditors association was that touchscreen pioneer from Clay county.

Very near the end of the legislative session, after all the HAVA hearings across the state, Secretary of State Culver weighed in. He said the topic most often raised at the hearings was by voters wanting to see their vote on the paper. He called for the passage of Kibbie’s bill, but nothing happened.

The Holy Grail of paper ballot legislation is HR 550. It is a US House of Representatives bill, introduced by New Jersey’s Rush Holt, a physicist from Princeton. The bill has garnered 159 cosponsors in the House. But none of them are from Iowa.

Holt’s bill does more than require that voters be able to see their votes appear on paper. It also requires election officials to check on the initial machine totals by comparing some actual ballots to the results reported by the machine.

There is no good argument for opposing this bill, except that it may require new equipment and more work by election officials. But it will restore confidence in ballot counting. That confidence is now eroding. We want trustworthy elections, not cheap and easy ones.

There hasn’t been much point in hectoring our Congressional delegation about their lack of interest in this bill. This holy grail was in Bob Ney’s committee and he clearly was not planning to drink from it. With Ney stepping aside this week and a genuine scientist in charge of the committee, the bill has another chance.

That means we have work to do. It is time to contact our Congressmen and find out why they have not yet co-sponsored this ballot protection bill. Why has Steve King sponsored a weaker bill? Why are Leach and Boswell both silent?

Put your own insights in the comments–no registration required.