Poll Inspector Training

I headed down to the Annex building for my poll inspector training yesterday. Overall, it was a good experience.

There's roughly 4,800 poll worker positions in King County. According to our trainers, as of 1 or 2 weeks ago, they were 1,000 people short.

Everyone, step up and be a poll worker. Call 206-296-VOTE or email pollworker@metrokc.gov. If you care about democracy, want to learn more about our elections, or enjoy sitting around gabbing with your neighbors, then this is the job for you.

Below, I relate the experience, with extra emphasis on what I learned and observed about the new Diebold AccuVote TSx electronic voting touchscreen machines.

The Class

There about 20 people in my class. Another 20 in the poll worker class next door.

I was easily the youngest person in class. One guy's tee-shirt said "I'm a Senior Citizen - Gimme My Discount!" He looked to be about the median age.

The trainers (Sylvia, Ann, Bill) were pretty good. They were very professional in their conduct, patient, very helpful, and tried very hard to make sure we understood what they were explaining. They even did a good job keeping the class on schedule. Pretty hard with people like me interrupting with questions.

I was less excited by the training materials and strategy. I'm a visual learner. Audio doesn't do much for me. Perhaps a better way to teach the procedures would be to show before explaining and having us do it ourselves. That'd cover everyone's learning style. I know it's a lot of work, but it would have been better if the materials we used for practice were closer to the real thing we'll use on election day. The differences were distractions.

For extra credit, I'd like to see role playing during training. Because we learn best by teaching others. The trainers do a good job covering many scenarios and how to handle them. It'd be great if we could pretend it's election day and we all actually go through all the motions.

A fun exercise was proofreading the example Poll Book and Inspector's List. Some of the errors were non-obvious.

I was pretty impressed by the questions from the class. There's just so many details. And the Elections people are always improving things. Small stuff that wouldn't be obvious until you do it. Like including a small flashlight with the large black ballot bin the optical scanner uses. It helps you look around inside to make sure you removed all the ballots. Also, using matching colors for transmittal slips and delivery bags. Red for red, blue for blue. That's pretty clever.

Hands On

I finally got to use a Diebold AccuVote TSx touchscreen voting machine.

The Printer

As I observed before, the printer for the voter verified paper audit trail is crap. Clearly an afterthought and done on the cheap. While showing us how to install the device, the trainer warned us that the big lid doesn't lock in the up position, so be mindful that it doesn't crash down on your hand. Like a guillotine. Geesh.

I've read accounts of people "hearing" their paper ballot being printed. The AccuVote TSx has a noiseless thermal print head. Just like at the grocery store. The machine itself makes printing noises by playing an audio file. That just doesn't sit right with me. It's a lie.

We were also warned to not wrap the paper around the take-up reel too tightly, because it could break the machine. If true, that's just sad. Imagine the cash register breaking if the clerk wound the paper tape too tight. I suspect poor design and cheap parts. That belief is reinforced by how it looks. We're paying a BMW premium price for cheap Yugo cars.

Audio Ballots for the Blind

I tried out audio features of the Diebold AccuVote TSx. Seeing how the justification for these machines is to allow the disabled to vote in "private", I wanted to see what I'm getting for my money. It was very illuminating.

I can't imagine a blind person approving of the AccuVote TSx. Maybe I'm wrong.

We all know voice mail hell, right? That's how the AccuVote works. It doesn't have to be a bad thing. But Diebold didn't do very well.

You use the detachable 12-key pad to navigate the audio menus. I'll call it the telephone user interface, or TUI. You use '1' and '2' to change the volume, "asterick" and '#' to change the playback speed. There's probably a reason '5' is used as the first ballot choice. Though '6' is for moving to the next ballot. I guess the organization of the options made no sense to me.

I was really surprised the audio doesn't have automatic pitch adjustment. When you make the playback faster, it's like listening to the chipmunks.

Also surprising, the speed control is not bounded. I imagine the only speeds that would ever be useful are normal, fast, and faster. Why anyone would slow the voice down, like the intro to Prince's song "1999", is beyond me. You can easily, therefore meaning accidently, speed up or slow down audio enough to make the system unusable. That's a user interface design flaw.

My favorite feature, by far, is the buzz sound when you do something wrong. The AccuVote TSx plays a "bzzz" when you type at the wrong time. It's actually a bit louder than the voice audio, so it'll probably hurt your ears when wearing headphones (I tried, it hurt). Modern TUIs allow type ahead, interruption, high level navigation, feedback that you've selected an incorrect key, etc. The AccuVote TSx just buzzes. Very clever.

In my opinion, as an expert user interface designer that has never made a product for the blind, the audio feature of the AccuVote TSx is deeply flawed.

By comparison, the Vote-PAD uses a model of audio tape player favored by blind users. Big buttons with embossed symbols. Intuitive controls. Familiar technology. Reasonable price. I believe that in this case, simple is better and the Vote-PAD did it right.


There's a bunch of settings for how your ballot appears on screen. The best settings are "High Contrast" and "Large Text". That's black on white with large text. The smaller texts is some areas are too small, even for me. The use of color throughout the product is just awful.

With the smaller text, multiple ballots can be shown, kind of simulating the paper ballot. Usability wise, I'd prefer a single race per screen, kind of like a wizard approach.

Typographically, meaning document and forms design as well as the rendering of the text, buttons, and other visuals, there's a lot of room for improvement. Some elements get clipped, like running into each other or off screen. The layout is inconsistent, which seems to be a Diebold trademark. I don't know how much is controlled by our people at the Elections office vs provided by Diebold. Either way, the skills of a talented graphic designer are desperately needed. It's embarrassingly bad.


We've all heard about the problems voting using touchscreens. The common excuse is "miscalibration". Yea, sure. I'm not sure that can explain vote switching.

As many know, Snohomish County had problems with a different product from Sequoia Voting Systems called the Edge Advantage (?). I've asked a lot of people for their take on those problems. One novel explanation (theory) was that it was raining election day. When a wet sleeves accidently brushed the touchscreen, a vote was registered. Well. Okay. I've heard stranger things.

So another inspector and I intentionally tried to misvote. To test this idea, I unbuttoned my sleeve to let it hang down. Dragged it back and forth. Raised it up and down. Touched with both a finger and my sleeve. The other inspector tried a bunch of stuff too. Even placing both hands and forearms on the screen. I couldn't misvote with the sleeve. The other inspector had to purposefully try to misvote.

I didn't think to first get my sleeve wet. I don't know what the touchscreen technology is. Probably induction. So that might work.

We weren't taught to calibrate the touchscreens. There's a separate poll worker who will tend the machine, a position called the AVU Judge. Maybe they do the calibration, if at all.

Anyway, I currently think the miscalibration excuse is tired and probably wrong.


Well, one thing is certain: electronic voting makes our elections a lot more complicated. King County isn't wrong there. There's a few sources of complexity.

The first comes from having 5 different types of ballots: poll site, absentee, challenge, provisional, and now virtual. For each ballot type, you have to handle the same kinds of errors and mistakes, including spoiled, canceled, and abandoned ballots. As well as undervotes and overvotes.

Because we vote in our own precincts, each with its own mix of races, we all need to be issued our ballots. For virtual ballots, you're issued a voting card (a smartcard, like a credit card with a chip inside) when you sign in. The card is encoded with which ballot to use. You walk to the machine, insert card in slot, and vote. Note that there's a separate computer device used for issuing voting cards.

Other sources of complexity come from using Poll Books, handling provisional ballots [1], and custody of the ballot. With electronic voting, you trade the paper ballot, which is how God intended us to vote, for a virtual ballot on a voting card, poll tapes, and memory cards.

See? Electronic voting, which is unconstitutional, expensive, and unreliable, makes everything a lot more complicated. All traits it shares with forced mail voting.

Further, what is obvious with a paper ballots becomes invisible with voting cards and virtual ballots. For instance, it's real clear when someone spoils their paper ballot and asks for another one. You tear the ballot and write a large "SB" across it. Very obvious. Using the voting card, you're trusting the computers to get things right. Which has always proven to be a bad idea.

[1] Which our touchscreen machines don't do, unless you're a blind absentee voter who decided to come into the poll to vote. Yea, the rules get that nutty. We students were groaning.


One bummer is the location. The Annex is near the Museum of Flight and Boeing field. Bus service is pretty bad. I left a bit late (can't help socializing). Leaving, I spotted a lady from my class on the sidewalk. She had that look, so I stopped to ask if she needed help. She'd been outside waiting for the bus for about 45 minutes and had to get downtown. So I offered her a ride. During the trip, she explained that prior training was at North Seattle Community College, which was very accessible for her.

It might be that the Inspector training is only offered at the Annex. I'm dubious of spending more money on new facilities, like our King County Executive has been pushing. But the Annex really doesn't work out very well for bus riders.

< A key opportunity for voting integrity | Maria on the Palouse >


Are you a poll worker this year?
Out of town, sorry.
Can't get out of work, sorry.
Other reasonable excuse, sorry.
Soon, I've signed up to defend democracy and will be trained shortly.

Votes: 4
Results | Other Polls
Display: Sort:
I took the Inspector training this morning, and thought I'd post my observations/recollections. You beat me to it, so here are my comments, observations, etc. Differences could be due to a different class, slightly different teaching of the course, and of course a different person taking the course and recounting their experiences.


Earlier this year when I was working as the 36th Dems' Pollworker Coordinator we were referring to these machines as Disabled Access Voting Equipment. Now they are referring to them as Accessible Voting Units.

The AVU equipment

I did not spend any time actually playing with the machine's user interface. I did try to feed the voting unit my Kinkos electronic payment card in all four orientations; it rejected it each time. This card is not an exact simulacrum of the voting card, but it is fairly close. The instructors seemed relieved it didn't break the machine.

There are three parts to the equipment:

  1. The handheld card activating unit.
  2. The card.
  3. The voting unit.

Inspectors have spare cards in case one gets damaged or walks away, but only one is supposed to be "in play" and remains in the custody of the AVU judge when not being utilized by a voter.

Inspectors take it at "Interface Value"

Let me be very clear that inspectors know absolutely nothing about the functioning of the software, what the software or hardware components might be, or what is actually encoded on the card.

We are taught to deal with it at interface value, and if something goes wrong we call the help desk.

I bring this up just in case someone somewhere makes the argument that inspectors are smart people who know about elections stuff, and so if there was something flawed in terms of the functional design we would spot it. Not so!

To put it the other way, inspectors are not giving tacit approval or endorsement of this equipment because they don't speak out about it: they simply have no basis to make any informed criticism.

The operational process

The following is the operational process for the AVU track:

  1. Voter goes to their precinct judge.
  2. Shows identification.
  3. Verifies their address.
  4. Signs the poll book.
  5. Is asked by the precinct judge whether they would prefer a paper or electronic ballot; at this point, presuming that the voter requests an electronic ballot, the process continues:
  6. The precinct judge prints AVU in the poll book.
  7. Fills out a (carboned duplicate) AVU encoding slip.
  8. Instructs the voter to take their copy of the encoding slip to the AVU judge who then activates the card according to the information on the slip.
  9. Voter uses the activated card to access the voting unit.

When encoding the card, a "precinct portion" is apparently encoded. This is not the same as a ballot code, which is utilized to differentiate ballot styles. No satisfactory explanation for this was given.

(Note to self: slip includes a "precinct portion" code which in the sample materials does not correlate to the ballot code which is used to differentiate in the case where a precinct has multiple ballot styles for instance when only part of a precinct is in a particular water, fire or LID. Need to examine some actual pollbooks on election day and ascertain just how many of these portion codes there are.)

If subjected to adequate scrutiny, presumably the number of ballots cast, the number of records on the paper tape, and the number of issued AVU encoding slips would have to agree.

Not-quite-provisional absentee ballots

Many absentee voters deliver their absentee ballots, properly sealed in the two nested envelopes, to the polls. That has not changed.

The situation can arise that an absentee voter has lost their ballot, or one of the two envelopes, or simply wants to vote at the polls; in these circumstances we have them vote a provisional ballot. That has not changed.

Here's the interesting wrinkle: Unlike poll voters, we don't ask absentee voters requesting to vote at the polls if they would like to vote with the AVU: the AVU is not capable of handling provisional ballots. However, if an absentee voter comes in and specifically requests to be allowed to utilize the AVU to cast their vote, there is a special procedure we follow which involves calling the help desk to ascertain that their ballot has not been received by mail and presumably invalidate the ballot should it later arrive by mail. So: it is possible for someone who is an absentee voter to cast a non-provisional ballot, if they're willing to use the AVU; but we don't tell them about it. In fairness, the rationale given for this is that disabled voters who have been utilizing absentee ballots may prefer to come to the polls and utilize the AVU instead.

If you really want to screw up closing the polls have five inspectors do it

They broke us into groups of five inspectors and had us walk through a mock polling place closing. They did not throw too many curves at us, but the materials were not complete and some steps had to be omitted as a practicality.

I don't know about the other groups, but ours made a complete hash of it. Contributing factors including having five people performing a process which only one person would normally be in charge of, some judges who had done it before and therefore "had a system" and some who had not, some errors of our own making (and others assuming the ones who made the errors knew what they were doing).

In spite of the fact that it was a complete fiasco, working through the screwups was a good way improve our understanding of the process.

My takeaway lesson: Make sure to follow the procedures. If people get in a hurry, then everything needs to stop while we get it back on track and as counterintuitive as that might seem we'll get out of there a lot faster.

by m3047 on Sat Aug 19, 2006 at 04:53:21 PM PST

* 1 5.00 2 *

This is fabulous, getting these detailed reports from the 2 of you.  Thanks.

I'm fixated for the moment on the idea that the machines make no noise in printing -- but  instead there's an audiofile of printing noise generated to make the voter feel better.

That seems so creepy to me.  Is that really what's going on?

by noemie maxwell on Sat Aug 19, 2006 at 07:05:52 PM PST

* 2 5.00 2 *

Display: Sort:




Seattle Port Commission










Environmental Issues






Challenging the Corporate Media Blockade

Watch Live or Archived Shows:
Seattle SCAN
South End PSA




Make a new account


Recommended Diaries

Washblog RSS Feeds

Political Contacts

Local Media

Coastal/Grays Harbor
Aberdeen Daily World
Chinook Observer
Montesano Vidette
Pacific County Press
Willapa Harbor Herald
KXRO 1320 AM

Olympic Peninsula
Peninsula Daily News
Bremerton Sun
Bremerton Chronicle
Gig Harbor Gateway
Port Orchard Independent
Port Townsend Leader
North Kitsap Herald
Squim Gazette
Central Kitsap Reporter
Business Examiner
KONP 1450 AM

Sound and Islands
Anacortes American
Bainbridge Review
Voice Of Bainbridge
San Juan Journal
The Islands' Sounder
Whidbey NewsTimes
South Whidbey Record
Stanwood/Camano News
Vashon Beachcomber
Voice Of Vashon
KLKI 1340 AM

North Puget Sound
Bellingham Herald
The Northern Light
Everett Herald
Skagit Valley Herald
Lynden Tribune
The Enterprise
Snohomish County Tribune
Snohomish County Business Journal
The Monroe Monitor
The Edmonds Beacon
KELA 1470 AM
KRKO 1380 AM

Central Puget Sound
King County Journal
Issaquah Press
Mukilteo Beacon
Voice of the Valley
Federal Way Mirror
Bothell/Kenmore Reporter
Kirkland courier
Mercer Island Reporter
Woodinville Weekly

Greater Seattle
Seattle PI
Seattle Times
UW Daily
The Stranger
Seattle Weekly
Capitol Hill Times
Madison Park Times
Seattle Journal of Commerce
NW Asian Weekly
West Seattle Herald
North Seattle Herald-Outlook
South Seattle Star
Magnolia News
Beacon Hill News
KOMO AM 1000
KEXP 90.3 FM
KUOW 94.9 FM
KVI 570 AM

South Puget Sound
The Columbian
Longview Daily News
Nisqually Valley News
Lewis County News
The Reflector
Eatonville Dispatch
Tacoma News Tribune
Tacoma Weekly
Puyallup Herald
Enumclaw Courier-Herald
The Olympian
KAOS 89.3 FM
KOWA FM 106.5
UPN 11

Ellensburg Daily Record
Levenworth Echo
Cle Elum Tribune
Snoqualmie Valley Record
Methow Valley News
Lake Chelan Mirror
Omak chronicle
The Newport Miner

The Spokesman-Review
KREM 2 TV Spokane
KXLY News 4 Spokane
KHQ 6 Spokane
KSPS Spokane
Othello Outlook
Cheney Free Press
Camas PostRecord
The South County sun
White Salmon Enterprise
Palouse Boomerang
Columbia Basin Herald
Grand Coulee Star
Walla Walla Union-Bulletin
Yakima Herald-Republic
KIMA 29 Yakima
KAPP TV 35 Yakima
KYVE Yakima
Wenatchee World
Tri-City Herald
TVEW TV 42 Tri-cities
KTNW Richland
KEPR 19 Pasco
Daily Sun News
Prosser Record-Bulletin
KTCR 1340 AM
KWSU Pullman
Moscow-Pullman Daily News



Democracy for Washington tool to email legislators by committee
WA House
WA Senate


Photo courtesy of photographer/thankyoult.org




Inspired by Rob McKenna's Fake Attorney General Letterhead
GIF of Letter




Medicine Takeback Program
Return unwanted and expired medications for free and safe disposal.