EVT/WOTE: Design a complete voting system, then ask vendors to satisfy needs.

Editor’s Note: August 8th and 9th, we attended the EVT/WOTE (Electronic Voting Technology / Workshop On Trustworthy Elections) in San Francisco. Over time, we are highlighting several papers and talks from the conference.

Monday’s keynote by Dana Debeauvoir, County Clerk, Travis County, Texas, was a call to design a new voting system to meet the needs of voters, election officials, and integrity advocates, then provide the specifications to vendors to compete to satisfy those needs. Debeauvoir pointed out that HAVA worked the opposite way: Officials with little notice, and little computer expertise were forced to comply with HAVA on short notice. They had to choose from what was made available by vendors.

Debeauvoir presented a possible draft design for a complete system, to start the conversation.
Keynote talk: <.pdf> <video>     Full diagram with notes: <diagram>

This is just one draft on one possible design

Debeauvoir emphasized that this is just one start at a design. It provides thoughtful ideas that highlight the benefits and challenges in designing, building, and implementing a comprehensive design. It would take a lot of cool heads, research, contemplation, and redesign to get the job done well.

Perhaps most intriguing and controversial is the idea of a ballot marking device for all in-person voters. There are several advantages: The elimination of incomplete in-person votes; write-in accuracy for in-person votes; perhaps easily integrated to serve those with disabilities; savings in paper. There are potential disadvantages: Requires a backup plan (paper ballots available?) in case of power failures, software errors, equipment failures etc. – a backup plan may or may not offset savings. Would the value of more accurate votes be worth the purchase, maintenance, programming, testing, and operational costs?

A good design would likely be a system with components that integrate, yet are not all required, or can vary from jurisdiction to jurisdiction. The Debeauvoir draft would work fine without the ballot marking device, ballot marking devices only in voting centers, or ballot marking devices optional in polling places intended mainly for use by persons with disabilities.

The value of careful design and planning with everyone at the table:

The way to waste time and money, while creating more problems than those solved is to go forward without a comprehensive review, without a comprehensive design, grabbing at solutions without looking at the big picture. This can continue through time, retrofitting one solution after another onto an inadequate system solving one problem while ignoring other problems and creating unintended consequences.

Sometimes buying add-on or standalone systems works and sometimes it does not. It usually depends on the level of integration and implementation costs, for example:

  • Relatively easy: It would seem relatively easy for Connecticut to use independent scanners to assist in post-election audits. Such a systems do not require integration with our current optical scanners. We would need a moderate number of scanners, easily updated software, and off the shelf equipment.
  • Extremely costly and difficult: If we were to purchase ballot marking devices as described in Debeauvoir draft design, without acquiring new scanners, it would be very complex, far from optimum, and severely limit options by requiring ballots that would work with our existing optical scanners. Considering the expense we would be stuck with the system for a long time, or huge sunk costs for a short term gain.
  • In the middle: It we were to purchase electronic check-off (poll books) devices, they should communicate effectively with a new or redesigned central voter registration system. They would not need to integrate with our existing scanners. We would need several per polling place, yet considerably less than the number of ballot marking devices.

The solution is taking time and including everyone at the table in some way: Election officials, voters, technologists, integrity advocates, and vendors. Taking time to consider what could work and what might not, taking in the perspective of the variety of needs across Connecticut and the experience of jurisdictions across the country.

At some point in the future we will need to replace our optical scanners. Assuming that is in the five to ten year timeframe, this would be a good time to participate in understanding and contributing to an integrated design. The alternative is scrambling later, risking expensive, less than optimum solutions.

A pleasantly surprising answer

In the Q&A after the talk, I suggested the audit portion of the design needed transparency, so that the public could have confidence in the result. The the whole system should take transparency into account.

I was pleasantly surprised by Debeauvoir’s’ answer. She agreed that transparency is important. It was overlooked in the diagram as they assume and provide for transparency in their operations.


Leave a Reply

You must be logged in to post a comment.