Several others have described their days as poll workers. Here’s my story of working at my own polling place in North Oakland, Alameda County. I arrived at my polling place at 6:00 A.M. and did not get home until 10:00 P.M.
The quick summary for those who tire of Brian’s long prose:
- The training of poll workers is inadequate.
- The voting machines face numerous security and technical problems.
- About 15% of my voters refused to vote on a machine without a paper trail.
- Most of those voters were also extremely angry that their only alternative to the machines was a “provisional paper ballot”. There were numerous heated arguments about the word ‘provisional’. People do not want a provisional paper ballot that may or may not be counted and that will not be counted right away. They want a “true paper ballot” that always counts and that is counted on election day.
- Absentee voters (and perhaps election officials) do not understand the rules for absentee voting.
- Being a poll worker is extremely stressful and exhausting and you should fall down and worship the poll workers at each and every election you vote in from now on.
- My view now is that the best election system is the simplest election system. In every single aspect of the election the paramount question should be: is there a simpler way to do this?
Prior to the Election
But the story actually begins a week prior to the election when I went to my three-hour training for first-time precinct inspectors and judges. (The names for poll workers in CA are odd. The poll worker in charge of a polling location is called the ‘inspector’. That was my job. The two other types of poll workers, whose assignments seem to me identical, are called ‘judge’ and ‘clerk’. There is also a program for high-school students to work as poll workers. They do everything the other poll workers do and are called ‘student’.)
The moral of the training story was that it was inadequate and revealed numerous security concerns about the machines. (I independently reached the same conclusion as Ed Felten, that it would be best not to publish such security problems until after they could no longer affect the election.) Here’s what I learned at training.
The Diebold Accuvote TS machines began being delivered to polling places in Alameda County two weeks prior to the election. The machines are cable-locked to a cart and the entire collection of machines is shrink-wrapped. This cable-lock is like a bike or computer lock with a rotating dial of numbers acting as the lock. The code for the cable-lock was extremely easy to guess, was the same for every polling place in Alameda County, and had been used in prior elections. This means that if one wanted to tamper with a machine and had access to a shrink-wrap machine in order to cover your tracks, you would have, in some cases,up to two weeks in which to tamper with the machines. These machines were delivered to schools, churches, and senior centers all over the county, and in my case, the cart of machines was simply right inside the main entrance, not in a locked closet or secure inner room.
Each machine was also secured by a single blue plastic cable-tie. (They sort of looked like this, but different.) I don’t recall learning that this would be the case in training, and had they not been there, or had there been a generic plastic tie, I doubt that the average poll inspector would have been concerned. If someone relocked the larger cable-ties and shrink-wrapped the cart up, I suspect those machines could go into service without any alarm.
The person leading the training also said a disturbing thing. She said that we should expect problems with the machines and that if one stopped responding or had a problem that usually the best way to fix the problem, as with most computer problems, was simply to turn the power to the machine off and then turn it back on. These machines run a version of Windows CE. Having had the misfortune to have used Windows machines in the past, I distinctly recall that about the worst thing to do to troubleshoot them was to simply power-cycle them. I’ve lost plenty of data that way before. Her confidence that there would be such problems was also not encouraging.
The scariest part of the training though was how much information there was to convey about both the pencil and paper procedural matters of running a polling place and learning the myriad of facts about the machines, combined with how little time was alloted to learning all of this information. I am a person who generally catches on pretty quickly when being taught new material, and I left that training truly baffled as to the exact extent of my duties on election day. One gentleman expressed this sentiment at the end of our training session by asking if there would be a “refresher” class before the election (which was only a week away). We were told we could come to an advanced/condensed version of the same class we had just had the next day. What people needed though, was not a faster run-through of the same material, but a slower one (or several slower ones!)
We were instructed to set up the machines the night before so that their internal batteries could charge overnight. They are supposed to arrive mostly charged, but in case of power outages on election day, they want the batteries at 100%. Only the student poll worker assigned to my polling place was able to assist me with set up on Monday. We spent about an hour arranging the room, unpacking the machines, and plugging them in. I turned on each machine to ensure that they would turn on and to be sure that they were charging. All five machines seemed fine, so I powered them off (they still charge while powered off) and used the blue plastic cable ties to close the machines. Again, the physical security of the machines overnight is only as safe as the polling place’s security generally. My polling place was fairly secure, with a code required to open the external doors, but I doubt every location is as secure. This provides another 12-hour window where someone could physically tamper with the machines and potentially go undetected.
As the inspector, I was also instructed to pick up various supplies several days before the election. Included in these supplies are the keys to the voting machines, the cards that are inserted into the machines, and the voter card encoders that programs the cards. Consequently, a corrupt inspector would also have a several-day window of early-access to the machines that might be undetectable.
When I arrived at 6:00 A.M. and began setting up the machines, one of the five machines had a malfunctioning screen. It would not come on, and remained grayish-black as if the monitor connection was simply out. There was an obvious little green power light indicating that the rest of the machine was receiving power, but you can’t read choices off of a blank screen, so that machine was closed back up and I called the problem in to the County.
The polls opened at 7:00 A.M. and we were by far busiest from 7-9 A.M. It was not until after that busy period that a replacement machine was delivered. When it arrived and we printed “the zero report” that supposedly indicates that no votes have been cast on that machine, it printed out a report on every single race or proposition in the County, regardless of whether they were races or measures that voters at my polling place were eligible to vote on. Also, while all the other machines indicated the name and number of my polling place on their screens, the new machine read “Precinct Coordinator” instead.
I was never told in training that the cards that are inserted into the machines are polling-place specific and control which ballot a voter receives. Consequently, I believed that if I allowed voters to use the new replacement machine they would be asked to vote on dozens of measures they were not entitled to vote on. Besides being enormously time-consuming and confusing to voters, I worried this might invalidate their votes. That was not a risk I was willing to take, and so I called the County to describe the situation. The person I spoke with apparently did not understand my situation (or wanted to be cautious too) as they did not tell me that voters could use such a machine and receive the correct ballot (as we did later in the day). A former inspector for this location came in during this machine’s down time and she too did not know that we could use that machine in its current state. Instead the precinct coordinator came around a little later and explained that the cards controlled the ballot and so the new machine would not require voters to vote on dozens of out-of-area matters. We then began using that machine without incident.
These cards that are inserted into the machines are also programmed by a Voter Card Encoder (VCE). My polling place received two VCEs, one of which was in an envelope labeled “Emergency” and was not supposed to be used so long as the other one was available. Our original VCE stopped working a little over half-way through the day. We had been told that these VCEs operated on batteries and that their batteries could be drained by leaving a card in them unnecessarily. Consequently we were very careful not to leave a card in the VCE and while other poll workers operated the VCE most of the day, I watched them use it many times and I never saw us leave a card inside it. Nonetheless it stopped working. We used the Emergency VCE the rest of the polling day, and luckily it lasted through closing. Had it failed too, I am not sure how long it would have taken my precinct coordinator to arrive with another VCE. In the meantime our only recourse would have been to provide voters with “provisional paper ballots”.
Voters Want A Paper Trail
I don’t remember the exact numbers, but I believe about 314 people voted on the machines at my polling place and about 78 voted on provisional paper ballots. Most of those provisional paper ballots were voters who did not want to vote on machines without a paper trail and who requested a paper ballot instead. It was a common request all day long. Many voters would ask if the machines produced a receipt or paper trail and were very disappointed when told that they did not. Some would go ahead and vote anyway on the machines with an air of exasperation. This especially happened when voters were told that their only alternative to the machines was a “provisional paper ballot.”
There is nothing that pisses off a voter more than telling them their vote is “provisional.” That word was the subject of several heated discussions in my polling place. I found myself in the unlikely position of defending the absurd system that Alameda County adopted. The provisional ballot has a removable stub that the voter keeps (I was not told this in training and only figured it out along the way.) and the provisional envelope also has a removable stub that the voter keeps. The envelope stub contains a phone number that the voter can call to confirm that their vote has been counted, 28 days after the election.
That number, 28, also really pisses off voters. They do not want to be told that their vote MAY be counted 28 days from now. They want their vote counted and they want it counted today. I would try to explain that it would be counted so long as they did not also vote absentee or try to vote in other polling places, etc. but this never made them feel any better. Voters want a true paper ballot that is counted as soon as the polls close. (This sounds like a reasonable request to me, but it is not an option that was provided.)
Absentee Voting Issues
I also had two cases where an absentee voter from a county other than Alameda stopped by to drop off their absentee ballot. While we took numerous Alameda County absentee ballots all day, I was not sure we could accept another county’s absentee ballots, so I called Alameda County’s Registrar when the first such voter arrived. I was told that unless that absentee voter was registered in Alameda County, we could not accept another county’s absentee ballot and that if we did accept it, then it would not be counted or forwarded to the other county for counting. Instead, I was told that the voter’s only recourse was to immediately drive to their home county and deposit the absentee ballot at the first polling place they found in their own county. The voters from Yolo and Monterrey counties were not very excited about this idea and I doubt that either did this. Instead, they were simply disenfranchised for not understanding the rules surrounding absentee ballots. After my first such conversation with Alameda County Registrar’s office about this, someone from the Registrar’s office called me back to ask what I had been told to do. I described the advice I had been given (as I did above) and they confirmed that this was correct. I have since spoken to a friend who was registered absentee for San Diego County and he deposited his absentee ballot yesterday in an Alameda County polling place after a long series of phone calls with various agencies. They told him the exact opposite of what I was told and that he could deposit it and have it counted for all non-local State and Federal matters. So now, I have no idea what the correct advice was.
Overall, this was one of the most stressful and exhausting 16-hour days of my life. While entitled to two 1-hour breaks, I took no breaks at all, both because we were so busy and so that my other three poll workers could take their breaks. The importance of the enterprise weighs on you and the desire to move people through quickly and accurately preys on your mind. I cannot imagine that I would agree to do this again, despite how important I think it was that someone who has turned a computer on before was there to manage things.
I also learned that voting is a human enterprise subject to human frailties. I do not believe that the number of signatures we collected equaled exactly the number of computer votes cast (it was off by less than 5), nor did the total number of provisional ballots we received exactly match the number we had filled out, spoiled, or blank at the end of the evening. (This was also off by less than 5). It irritates me that I was not able to watch every single thing that each of my poll workers did and so I was not able to prevent every single mistake. (Do not get me wrong. They were hard-working, well-intentioned people who deserve our thanks.) We also ended up with about three provisional ballots that were not properly filled out on the front with a voter’s printed name and signature. I know that I told every voter to whom I gave a provisional ballot that they had to fill this out and that every provisional ballot I received back I double-checked for this information before placing it in the ballot box. I doubt that these blank ones will be counted and that makes me mad. Perhaps if the voter saved the stub and calls in to check on their vote’s status, they can be matched up and counted. But I doubt that the registrar receives many such calls.
This leads me to believe that the best voting system is the simplest system. It is so difficult to avoid errors and mistakes, to train volunteer poll workers, to get voters to follow instructions, and so on, that every single aspect of the voting process should be examined with the paramount goal of simplification at every single stage. I don’t know what the result of a careful study of voting with such a simplification-goal would be. I suspect that it would not involve any sort of touchscreen machine, because the steps necessary to ensure the security and verifiability of such machines are anything but simple.
The best argument for touchscreen voting machines is their ability to better serve voters with disabilities and to allow those with disabilities to vote privately when they might otherwise require assistance. However, I did not have a single disabled voter request to use the headphones or keypad on our machine with these features. Fewer than five voters requested the magnified ballot, and the one blind voter we had wanted to receive assistance from his son rather than use the headphones. (I make no claims about how representative my location might or might not be.) So, it seems to me that a simple paper ballot with ovals filled in with pen or pencil might be the best and simplest system available. Braille and magnified paper ballots are possible. And even for those with vision problems who do not know Braille, a textured paper ballot could be devised that would allow such voters to vote unassisted by anything but a pre-recorded audio tape. A paper ballot could also be designed with larger boxes for those with fine motor skills disabilities.
However, I used to use optical-scan machines when teaching in Southern California. The machines I used had a rate of error that is unacceptable for an election. (Perhaps better machines exist.) I think election work might need to be like jury duty and when the polls close, an army of willing and unwilling vote-counters should descend on the ballots for a manual count. I don’t care if I don’t know the results until some time the next day. The inauguration isn’t until January, anyway.