16 hours at a polling place

The serial numbers and integrity of each security seal must be verified by the inspector and one of their clerks before opening and after closing the polls.  The inspector reverifies these seals every hour.  Vote By Mail ballots must be surrendered before an election day ballot may be issued.  The public listing of who in the precinct has voted must be updated hourly.

Election security protocols were on display all day for all to see, a reminder to everyone of the importance of free and fair elections.  I was one of the thousands of volunteer clerks and inspectors who staffed the hundreds of precincts in San Francisco on Tuesday, November 5, 2019.  It was an incredibly rewarding experience and this account will not do it justice.

My experience began when I rather uncharacteristically answered a cold email from the San Francisco Department of Elections in September by responding to their intake questionnaire, which was hosted on Survey Monkey.  (Resourceful people will find infrastructure wherever they can.)  The second half of the questionnaire was actually a quiz that assured a basic level of English reading comprehension and tested the prospective clerk’s ability to sort names and numbers and to perform basic arithmetic.  Then came the phone interview which was as concerned with my willingness and ability to work a 16-hour day as with my demeanor and motivations.

I was able to take on this 16-hour day thanks to Slack’s VTO — volunteer time off — benefit.  “Off” is not generally how I’d describe a long day filled with responsibilities but, in truth, a poll worker’s day strings together many long periods of boredom in which there was genuinely nothing productive we could be doing.  It was actually a nice change of pace.  I had time to get to know the four complete strangers who were with me from 6:00am until 10:00pm.  By lunch, the others were making fun of the two of us employed in the technology industry for checking in on work on our phones.  Our polling place was in a residential garage, open to the brisk fall air, and most of us found a few moments to stand outside in the sun to warm up.  All the while we inquired about each other’s lives and what motivated us to volunteer at the polls.  Four out of five of us lived in the neighborhood.  The fifth was a high school student less than a mile away.

Small talk ended abruptly whenever a voter arrived.  The first clerk found the voter’s name and address in the roster and presented their options in case of any complication.  The second clerk prepared their ballot.  A third stood by to operate an accessible ballot marking device.  A fourth stood by to supervise the voter in operating the ballot tabulation device.  When, instead of one voter, a whole crowd appeared at once, we fought our adrenaline-fueled instinct to hurry.  “Slow is smooth.  Smooth is fast.”  All five of us were first-time poll workers.  We speculated that turnout for this election was pretty low but chose to make lemonade by reframing it as good practice for March and November 2020 which, we expected, will be far busier.

The first two clerks set the tone for the others through a series of unspoken but universally understood cues.  Only voters who would eventually insert their ballots into the tabulation device signed the roster, which afforded an opportunity to verify the expected number of voters and ballots later.  The blue privacy folders, while of course offered voters privacy, also signaled to the final clerk that that voter was authorized to insert their ballots into the tabulation machine.  A provisional ballot envelope signaled clerks to direct that ballot into the red box along with Vote By Mail ballots dropped off on election day.

There was no time to learn this protocol on the job.  The polls had to open at 7:00am to avoid disenfranchising a time-constrained, early-rising voter.  (Indeed, there was one voter standing outside our polling place at 6:57am.  He was out the door again, sticker on shirt, just a few minutes after 7:00am.)  Unable to learn on the job, the mandatory poll worker training class was critical to our ability to operate a free and fair election.  The training classes were held in courtrooms at City Hall.  Attendance was tracked via sign-in at the door and feedback forms collected at the end.  The training followed the 100-or-so-page written manual and stopped to survey the room on several true/false questions immediately after the material was presented.  The manual also included “job cards” which served as the on-the-job reference materials that we used on election day.  The last half of the training was devoted to following the job cards to assemble, configure, and operate all of the equipment we’d encounter at our polling place.

Even the combination of this training, the manual, and the job cards didn’t make the process fool-proof.  I found it easy to become overconfident in my memory and intuition when in reality I dearly needed the job cards in order to complete my tasks successfully, especially when opening and closing the polling place.  I think two exceptionally well-trained people — people who’d had many hours of drilling on all the jobs — could operate a polling place successfully.  But polling places only operate twice a year, at most, and the skills imparted by such drilling are likely to atrophy.  Add to that the fact that equipment and procedures change potentially every election and this system of training, documentation, and human redundancy makes perfect sense.

The quality of the documentation, in particular, was very high.  It was notably better than any architectural documentation or runbook I’ve ever read (or written) working in the technology industry.  In this context, inconsistency is toxic.  Inconsistency, even if it doesn’t result in ballots being lost or miscounted, erodes the voting public’s confidence in their polling place and in free and fair elections, in general.  The job cards spell out the precise protocol to be followed by each clerk to process each voter.  These things would make Atul Gawande proud.

One of the more productive ways my fellow poll workers and I spent our periods without voters was to discuss the phrasing we used when offering non-English ballots (to every voter, because you never know) and provisional ballots.  The job cards suggested language for clerks to use at the beginning of each job but elided further conversational details.  This was great!  It focused our attention on the most important aspects of our interactions with voters; it made sure these details stuck.  Less important details were left to be discovered — the job card got the conversation off to a good start trusted the clerk and voter to communicate effectively from there.

There were a lot of other details left to the poll workers and we got a few of them wrong, at first.  We posted at least five signs outside the polling place directing voters to the entrance, which was an open garage on a wide mostly-residential street.  However, the port-a-potty placed outside for us obscured the view of the open garage door when standing next to the telephone pole where one sign was mounted.  One voter complained that he’d been standing in front of the garage next door, which had a different address, for several minutes, worried that his polling place had closed.  After hearing this, we moved several of the signs around to ensure all of them pointed very literally to the open garage door and, specifically, that folks looking at the sign would have come around the port-a-potty far enough to see the open garage door.  Once inside, voters had to navigate the rather serious grade in the center of the garage that led to a drain in the floor.  At first, a rug was taped down with yellow tape to alert voters.  By the end of the day, a chair with a “watch your step” sign occupied the steepest part of the drainage, saving voters with limited mobility from falling into the hole.

Given the degree to which electronic devices, search engines, and autocomplete feature in all our daily lives, I was surprised that nowhere in any of the training or documentation was there a plea to trust in the paper information retrieval systems that lie at the heart of a polling place.  Alphabetized lists, printed and bound, with a range like “Be - Ca” printed at the bottom of each page were all a human being (who’d passed the poll worker intake quiz) needed to efficiently find a name in the voter roster.  Each row of the alphabetized roster included an integer that identified a row in a numerically sorted index of voters, too, which was used to publicly document voter turnout in near-real time.  Ascending numbers on ballot receipts facilitate counting excess ballots after the polls were closed.

There’s a whole extra category of things they don’t tell you in training (presumably because they don’t want to humble-brag):  San Francisco elections are world-class.  There are at least four different ways a voter may cast a ballot, including weeks ahead of time, in person, at City Hall.  There are field officers from the Department of Elections deployed throughout the city on election day who area incredibly responsive to the tiniest need at any polling place.  There is a paper audit trail for every action taken throughout the system.  Voter turnout is auditable in near-real-time on election day, precinct by precinct.  Voters can look up the date their specific ballot was counted on the Department of Elections’ website after the election ends.  New for this year, images of every electronically tabulated paper ballot will be available for download, offering independent auditors the opportunity to audit the count in its entirety.  (These ballots do not contain any information that identifies the voter who cast them.)

One of the many ways folks in San Francisco can vote, and one that’s a very important part of expanding the franchise, is Vote By Mail.  Around 70% of San Francisco voters are registered to receive their ballot by mail.  They have the option of mailing it in, dropping it at any polling place in the city, surrendering it and casting a ballot at their designated polling place, or surrendering it and casting a provisional ballot at any polling place.  (Not that it matters in terms of your vote being counted but dropping your Vote By Mail ballot at any polling place, even your own, is the second-least-efficient method of voting because the envelope must be opened and tabulated by Department of Elections personnel after the polls close.  Provisional ballots are the only less efficient method because they additionally involve cross-referencing the voter’s name and address with the roster of voters to ensure no one votes twice.)

A rather large number of voters were surprised that they were registered to Vote By Mail.  Some recalled being out of town for last year’s election but didn’t expect their choice to stick.  Others seemed genuinely flummoxed and even a bit upset.  With only one exception (to be revisited shortly), all of these folks successfully cast ballots.  Many opted out of Vote By Mail for the future.

Why do so many Vote By Mail voters drop off their ballots on election day?  Stickers.  Vote By Mail drop-offs accounted for almost half the ballots cast at my polling place and I can’t explain this any other way.

We also processed a lot of voters that greeted us with some variation of, “I used to vote at the fire station.”  Unfortunately, thanks to the incredible density of precincts in San Francisco, many of these folks who made the seemingly rational assumption that the polling place they walked right past on their way to the fire station must be theirs turned out to be wrong.  Polling places in San Francisco end up in fire stations, schools, churches, and even private garages.  The incredible density of precincts helps prevent long lines and avoid long, out-of-the-way commutes in order to maximize voter turnout by making voting convenient.  One could argue they went too far:  More than a few folks were angry that their designated polling place was two blocks away when they lived across the street from this polling place.  There is, of course, no reason to expect the polling place to be at the geographic center of every precinct.  After we explained the options available, one would-be voter actually stormed off without even voting; this was the low point of the day.

In this situation, we were trained to offer voters a few options.  Firstly, we used a paper map with all precincts drawn to identify the voter’s precinct by number and then a numerically sorted index of precincts to identify their polling place.  (This is another example of a paper information retrieval system that works exceptionally well.)  Their first option was to go there to cast their vote.  If that’s too far or too inconvenient, they have the option to cast a provisional ballot at our polling place.  By the end of the day we’d honed the language we use to describe these ballots to minimize their negative connotation.

There were also an alarming number of voters who claimed their Vote By Mail ballot never came in the mail.  To us, though, a Vote By Mail ballot that never arrived is the same as a Vote By Mail ballot left at home.  We walked those voters through their provisional ballots, dealt with their disappointment and frustration, and got them their sticker.

The fact of the matter is that a provisional ballot is simply a ballot that is counted by hand after the polls close and after cross-referencing the voter’s name and address with the roster of voters.  It’s the same ballot.  Their vote counts.  And yet, we had folks run home and rummage through piles of paper to find their Vote By Mail ballots to avoid casting their ballot provisionally.  Many chose to walk the extra blocks to their designated polling place to avoid casting their ballot provisionally.  I know what you’re thinking but, no, provisional voters get stickers; it isn’t that.  There’s an opportunity for the Department of Elections to tackle the stigma associated with casting provisional ballots and increase voter turnout.  I’m not sure how to tackle this — most obvious solutions involve computers and I’m not excited about bringing those into the election process.

All told, my polling place processed 262 votes including election day ballots, Vote By Mail ballot drops, and provisional ballots. That represents more than ¼ of the precinct’s registered voters and doesn’t include any of the approximately 700 Vote By Mail ballots that could have been mailed or dropped at another polling place.  I know my work had absolutely nothing to do with voter turnout but, nonetheless, I feel proud to have played a part in those votes being cast and counted.

I volunteered to work at a polling place because free and fair elections are important to me, because election security is interesting, and because my employer made the time and space for me to do this.  I got to know several of my neighbors by sitting and standing with them all day in a garage just down the street from all of our homes.  I feel more a part of my community.  When my wife came to drop her Vote By Mail ballot, she brought our kids and a box of home-made cookies to lift our already high spirits.  The sugar carried us through the end of our very long day.  We all walked the same direction towards our homes together, a newly acquainted group of neighbors happy to ensure there are free and fair elections and stickers for all.