Recently, I had the opportunity to test three of the latest voting machines equipped with accessibility features. The Michigan Bureau of Elections held a Mock Election which allowed testers and poll workers to experience each of the machines and tabulate votes for both disabled and non-disabled voters. From my testing of the three machines, it was apparent that the usability of each system ranged from pleasant to downright frustrating. Below are some initial impressions of each machine, presented in the order they were tested.
Dominion ImageCast Democracy Suite
The voting system from Dominion included an accessible keypad, touchscreen, and a printer for paper ballots. Initial set-up of the machine required the use of a digital programming card which included information to load and verify the ballot. While I was able to insert the card, several set-up steps needed to be performed by a poll worker. Among these were choice of language, and the screen privacy guard option, which allows a voter to turn off the visual screen output. It's worth noting that this is the only machine which does not allow the voter to change this setting after initial set-up.
Once speech was finally available, I was presented with initial instructions read by Google's Android text-to-speech voice and an options menu which allowed me to change volume, speech rate, and visual display options. Unfortunately, the maximum volume was not loud enough for a noisy room, and the fastest speech rate was less than what is available on Android and too slow for an advanced speech user.
The keypad features buttons in various shapes which can be readily identified. Left and right arrows are on the left side while up and down arrows are on the right. There is a large X in the center which is used for selection. Dedicated buttons to adjust the volume and speech rate are found near the top. All buttons have braille labels near them, though the layout of the keys often made the placement of the braille labels confusing. The design choice to place the two sets of arrows far away from each other is perplexing at best.
The machine was plagued by user interface issues, often requiring the voter to press several key presses to accomplish a simple task. For example, when reviewing a ballot, if the user wanted to change a vote from NO to Yes, no less than 9 key presses were required to accomplish this task. In addition, the function of the right and down arrows are duplicated, as well as the up and left arrows. I was told this was done because of the needs of low vision users, but it made the navigation of the ballot needlessly time-consuming and complicated. Often, help and tutorial messages were spoken before important content, such as when speaking the name of an entered write-in candidate.
Another issue arose when speaking the names of the candidates and ballot proposal language. This information was spoken using the Cepstral text-to-speech engine, with the recordings in a much lower quality and volume than the rest of the speech feedback. Using the same text-to-speech voice throughout the system would be ideal. Care also needs to be taken when speaking the titles of ballot proposals and other items. The word millage, a common election term, was mispronounced.
Help information was given throughout the process, and presented in the manner of screen reader hints. Speech could be easily interrupted if the user chose to not listen to the help information.
While I was able to complete and print my ballot, I'm hard-pressed to recommend this system in its current form. That being said, many of the issues identified are software-based and could be fixed using a firmware update.
Hart intetcivic Verity Touch Writer
Hart intetcivic calls their Verity system "The Future of Elections". To be completely blunt, if this is the case, I'm worried for the state of accessible voting equipment.
Set-up involved the poll worker entering in a code to load the appropriate ballot using the touch-screen. This process did not include speech feedback and was not accessible. Once the ballot was loaded, prerecorded instructions in a male voice were spoken through the headset.
The accessible keypad includes two buttons (Select and Help), and a dial called the Move Wheel which can be turned using the thumb. The dial emulates arrow keys and allows the user to go through menus while the Select button locks in the current choice. The use of only three controls was an intentional design choice, but it quickly became limiting when attempting to efficiently navigate the screen.
The initial screen included a menu to adjust audio settings including volume and speech rate. To adjust the volume, one must select the raise or lower options and then press select for the new volume level to take effect. This is the only machine of the three tested which did not include dedicated volume and speed controls, which presents a hassle if one wants to make adjustments during the voting process. Only three speech rates were available, with the fastest option still quite slow for advanced users. In addition, since human speech is used throughout the process, the faster speech level resulted in choppiness and audio artifacts which made it more difficult to understand the recorded prompts.
I did not complete my ballot with this machine because of one major reason...HORRENDOUS LAG. Users of electronic devices may often become frustrated when it takes a quarter second or more to hear audio feedback after pressing a button. When using the Move dial on the Verity, it often took 3 or 4 seconds for any feedback to be given after the dial was turned. In addition, after pressing the Help button, it was often difficult or impossible to interrupt the instructional message and return to the previous screen.
After spending about 10 minutes with the machine and still working on my first ballot selection of 23 contests, my frustration level reached a point where I had completely lost interest in completing my ballot.
With my faith in modern voting technology quickly running out, I moved to the last of the machines, The ExpressVote from Election Systems & Software. ES&S purchased the assets of the former AutoMARK system, and the design of this model takes many cues from the previous version, which is a good thing.
I walked up to the machine and inserted my paper ballot into the reader, which immediately caused speech feedback to begin. No intervention was necessary from the election workers.
The keypad includes a rocker button for Volume labeled VOL in braille and another for voice speed labeled TPO for tempo. To the left of this is a five-way navigation pad with a select button in the center. A button to turn on and off screen input can be found near the top. Beeps are heard when buttons are pressed, and speech feedback is given within a quarter second. A more modern male voice is used on this model, as opposed to Eloquence speech on the AutoMARK, but it was clear and easily understood.
For those familiar with the AutoMARK, the voting process was nearly identical. Up and down arrows are used to move through ballot choices, and right and left arrows move between contests. For new users, contextual help information is given as hints. Warnings are given if a ballot question is skipped without the appropriate number of votes or if a user attempts to vote for too many candidates in a contest. Overall, I completed my 23-question ballot in about 5 minutes.
Of the three systems tested , the ExpressVote is the only one I am comfortable recommending in its current form. Set-up was achieved independently vy the voter, prompts were spoken efficiently, and a ballot could be completed using the fewest number of key presses.
It's worth noting that none of the machines tested offered support for refreshable braille displays, meaning that deaf-blind individuals are still excluded from the accessible voting process. In addition, the machines often produced ballots which looked differently than those completed by voters not using these machines, meaning it would be easy for one to discern electronically-completed ballots from others.
In conclusion, it became quite apparent that not all election equipment is created equal, and many systems fall short of providing a true voting experience. As states look to modernize and replace their voting equipment, it becomes important for disabled voters to become a part of the process and provide their feedback to help ensure that accessible and usable equipment is purchased. I'd like to thank the State of Michigan's Bureau of Elections for the opportunity to participate in this process, and hope they use the feedback provided to make smart decisions about the future of voting equipment.Category: Articles
I got to test the ExpressVote here in Maryland recently. While I didn't have any others to compare it against, from your description it does sound like the optimal solution of the three. I had no trouble using it myself.
You must be logged in to post comments.
J.J. Meddaugh is an experienced technology writer and computer enthusiast. He is a graduate of Western Michigan University with a major in telecommunications management and a minor in business. When not writing for Blind Bargains, he enjoys travel, playing the keyboard, and meeting new people.