or
I Was a Roving Tech for ES&S.Joyce McCloy of NC Verified Voting sent me a link to a job listing about a month ago for people to act as "rovers" during the NC primary election. The rover's job (one per county)was to assist county election officials with any problems they encountered with voting machines (OpScan and DRE) on election day.
While technical experience was preferred, it was not necessary. Pay was $300 a day plus mileage, meals and hotel expenses for two days. Training would be provided the week prior to election day.
Disclosure: All funds I receive for my efforts will be donated to charity, and documentation posted to certify this. My intention was to get a close look at ES&S systems and procedures as a citizen activist and to provide competent technical assistance if it was needed, not to make money. I was reluctant to take this job on since I knew it would give ammunition in some quarters that "David is in the pay of a voting machine company" and I really don't need THAT headache.
However the opportunity was too good to pass up, so I am risking it anyway.
I would welcome suggestions on who to donate the money to. I am leaning toward a cancer Charity in memory of Andy.
The training session was divided into two classes, one for OpScan and one for DRE. When I arrived, I didn't know this and walked into the first room with lots of ES&S people. I got through most of the I-Votronic class, before they realized I was set to work in an OpScan county.
The DRE training was eye-opening. Working the machine is complicated, with each voter needing to have an electronic cartridge to initialize the machine before they can vote. The number of things that can go wrong, while not extensive, is certainly much more problematic than on an optical scan system (the sheet for trouble shooting DREs is about six pages, for OpScan two).
The trickiest part of the system is the VVPB printer. Since they are using thermal paper, you have put the paper roll in correctly. This was stressed repeatedly during training. The easiest way to be sure you where feeding the printing side correctly was to scratch the paper's surface with a finger nail. The treated side will leave a mark, the untreated side will not. We were also warned to keep an eye on the tape during operation to be sure that it was in fact printing (the paper was fed in correctly and had not jammed).
Jamming was a concern for obvious reasons, but distressingly the system does not appear to detect jams, so the DRE will keep functioning even if the tape has jammed or is not printing. This is, of course, unacceptable. After the election, we had reports of paper being fed incorrectly in eleven machines in Mecklenburg Co. (Charlotte). I did not attend the training session for that county, or the training session for poll workers, so I do not know how well the issue was stressed.
When the printer is working, it records each and every action on the system. Move into supervisor mode, this is recorded. Each and every button you push is recorded, which is how it should be.
The downside is that it burns through paper, meaning you have to change the roll about every 100 voters. And of course, changing the paper roll, while simple enough for those familiar with the process, can be a major hurdle when performed by the unskilled under the stress of an election day. Older poll workers with poor manual dexterity or eye sight are going to have problems.
All of this demonstrates why DREs, even ones with VVPB, are a bad choice for elections:
1) You have to have multiple machines per precinct. OpScan requires one machine per precinct.
2) Only as many people can vote at a time as you have machines. With OpScan, any number of people can vote at the same time, as long as they have space for the ballot and a little distance for privacy.
3) Deployment of multiple machines means trucks must be hired for delivery. OpScan will fit in the trunk of a car.
4) If a machine fails, voting cannot occur on that machine. If battery problems ensue, or power is lost for a long enough period of time, voting STOPS. With OpScan voting occurs whether the scanner is working or not. The ballots are placed in the ballot box and scanned later.
5) The procedures for starting up DREs is a bit involved. Rebooting a hung machine is time consuming, and there are electrical cords and cables to keep track of.
Votes are stored on paper, on a compact flash memory card, and inside the machine. The flash card has a seal over it which must be broken to remove it. Removing the card during the election will halt the machine, and require supervisor intervention, and action by the judges at the polling place.
I was told that the supervisor password for the units would be the same for the entire state. It was explained that this was being done to help with deploying a brand new system statewide for the first time, and that in the future, it would be up to the counties to set the passwords.
OpScan training was a breeze compared to DREs. The AutoMark unit was particularly impressive, as it is the unit for use by the disabled. It has an audio feature for the blind, (with standard headphones supplied, but I noted that it had the old RCA-style audio jack as well). It also had a standard connection of sip/puff paddles for use by those with mobility problems. You can turn the screen off so no one can see how you are voting. I did this and closed my eyes, in order to see how well I could cast a ballot.
Child's play.
The electronic voice is a bit creepy, but the system worked as advertised. You insert your ballot in the slot (orientation is irrelevant, the system can work with the ballot no matter what way you you insert it, as long as you insert it straight. Insertion can be done by a poll worker since the ballot has yet to be marked).
Once it is in, the system prompts you on how to use the controls, then reads each race off to you and names the candidates as you scroll through them.
When you are through making your choices, you are given and opportunity to review and make changes. When you cast you vote, the ballot is marked and ejected for the voter to take (or an assistant or poll worker can take it and run it through the scanner for tallying. A person with problems using their hands will still need help at this point).
BTW, ANYONE can use the AutoMark. Poll workers are instructed to ask no questions if someone requests to use the machine to cast their vote (the machine would also allow the illiterate to vote). Once the ballot has been created, but before it is submitted for tallying, you can still change your mind, spoil your ballot and get another one. Once scanned, it goes automatically into the locked ballot box and
alea jacta est.
You cannot overvote with the AutoMark (provided the ballot is programmed correctly). You can undervote, but it is a bit harder to do accidentally, since the machine reads off each race.
For the rest of us, voting is like the SATs. Take the ballot and mark it with a pen.
Problems with OpScan and the AutoMark were pretty much in the paper jam department. Correcting them was pretty simple.
The OpScan unit was pretty simple to use, but had one flaw which has caused ES&S trouble of late. It uses a battery-backed SRAM PCMCIA card to store its votes. Why ES&S chose this type of card mystifies me. Compact Flash (which requires no battery) is a much better choice. In fact, it is what they use on the AutoMark. I hope to get an answer from ES&S on this issue.
You can under/over vote, which is a big disadvantage. Supposedly, the OpScan's are being set to detect overvotes and reject the ballot. It could also be set to reject under votes but this is not being done by order of the SBoE. I am trying to learn why.
On to election day.
There is not really much to say about it. Monday I drove down to Taylorsville, NC (Alexander Co.) and met Linda Mundy, the county election director. She was very solicitous, very dedicated, and very serious about avoiding problems. She was concerned about the memory card problems, and had an entire set of spares on hand to cover each precinct.
Alexander Co. is small, with only 16 precincts. Ms. Mundy loves her OpScan units and said that DRE was never an option in her county. Her voters are older and would not have welcomed DRE.
Election day came and the memory cards behaved themselves. There were a few minor hitches, the most frequent being that the OpScan unit was not seated correctly on the ballots box (the OpScan unit sits on a box the size of a small filing cabinet. This is the ballot box and is locked and sealed). These problems were all solved on the phone without the need for an on-site visit.
The day went on with nary a problem. At the end of the day, the PC cards were brought in (along with poll tapes and poll books), the cards read, and the resulted transferred to a separate machine which transmitted the results to Raleigh. The results were also phoned in for comparison.
And here we have the weakness of the system. I saw no indication that the resulted transmitted were encrypted. The file was straight ASCII and went, as best I can tell, by an FTP protocol. Handling the cards is a "chain of custody" issue, and the folks in Alexander followed the rules as best I could tell.
NC law mandates random audit of a percentage of precincts in each county. This recount allows us to double check the machines for accuracy and as a guard against fraud. If the hand-eye count shows deviation from the digital count, the paper count is taken, the digital count discarded. Any unexplained deviation can trigger a larger hand recount. While OpScan systems can experience deviations in the count due to improperly marked ballots or scanner calibartion issues, DREs should match hand counts 100%. Deviation on a DRE means that either: 1) You miscounted the paper. 2) The software/hardware malfunctioned, or 3) Someone tampered with the results.
The NC SBoE did follow the law, but broke security by announcing which precincts would be counted to the election directors PRIOR to the election. This was stupid and must not be repeated. A letter has been sent to the state director asking for an explanation of this decision.
All in all, this was a pretty smooth deployment. It was also NOT indicative of what could happen in November, as the general election will involved a LOT more voters. Hand-counting DRE's VVPB will be an inconvenience in the primary, but a major pain in a general election. Look for lots of whining from counties with DREs.
I predict that counties with OpScan will have fewer problems than counties with DREs in November.