Corvus Consulting

Today is a big deal in the United States. In the midst of this historic election, there’s a lot of anxious uncertainty around the reliability of the voting process as it adapts to new technologies. At home in Vancouver, we find ourselves in the tail end of a streak of elections, from the federal level to the municipal (and likely the provincial before too long), so I’ve found myself reading and thinking about how elections work as a technology.

When I say technology, I mean it in the sense I learned from reading essays on the subject by scientist, teacher and writer Dr. Ursula Franklin, where technology is not only the material tools that we use, but also the process by which we use them. Elections are about as close to an ultimate expression of that concept as I can imagine: democracy is realized through myriad tools and procedures, and exists only when it is practiced. The rules for handling and counting votes are as important as the ballots themselves, because neither the materials nor the process for operating them can produce the same results on their own.

Note that I’m not talking about the campaigns themselves, which are vastly more complicated than the casting and collecting of votes. While less complicated, election day is where the output of tabulated votes is transformed through observation and belief into an outcome of collective decision-making, spanning vast geography in a small timeframe, scaling up to a hundred million participants or more. And we thought web 2.0 was cool; the electoral mechanism itself is nothing short of awe-inspiring.

But there’s an ugly trend making its way into the technology of voting, one so familiar that we can feel powerless to change its course even though every early sign suggests that we should: polling stations are putting high-tech into the voting booth itself.

There’s enough suspense to go around for one day, so I’ll put my conclusion up front: the current and foreseeable generations of electronic voting machines have no place in democratic elections. None; nada; them and the horse they rode in on. For the rest of this post, I’ll explain how that conclusion came about.

What, Me Worry?

If you’re bracing for a rant on dishonest election officials and the ease of data manipulation that electronic voting can provide, I can’t deliver that. Though integrity and security in electronic voting machines is obviously very important and current, it’s a red herring that sidesteps the whole question of whether the machines should really be in use at all.

So let’s assume honesty in election officials for the sake of discussion. Take a few moments to watch this video showing the calibration process for a touch-screen voting machine that misbehaves across party lines:

Just put the key thing in and tap out a quaint sequence, and then it all works. It’s so simple, except that it’s heinously complicated. Not the actual calibration, but everything involved. The key used by the official is itself likely more complex than any voting machine preceding touch-screen technology. As such, the key, the touch-screen, the paper recording device, are all likely to be beyond the understanding and means to repair or to re-create by anyone present at the voting station. Put simply, the technology makes the material side the exclusive domain of electronics and software specialists.

Deep Disconnects

Rather than a person creating the mark on a ballot – the piece that is counted as a vote – , a voting machine sublimates the process of recording the mark on the ballot into the binary, scrambled and untouchable in the process of making the ballot. It’s like having to whisper your choice to someone else, who then goes around the corner and marks the ballot on your behalf. But that person moves very fast, and lives free of political bias, so it’s all cool?

In the video we see this machine making a printed record for visual verification at the point of voting. This feature is definitely reassuring, but initial counting still originates from a software record, making the paper trail the source of truth only when arbitrating a challenge to the results.

So there are really two serious disconnects in the overall usability of electronic voting machines: the first between non-specialist (polling volunteers and officials), and specialists (who design, assemble and service the machines), and the second between the voter and the actual recording of the vote. On the more practical side, they add expense and error-prone complexity to running elections, pushing up costs even though the trees get a break with less paper. Reliability, accountability and a good user experience are concerns closer to home.

The Long Game

I’ve recently been enjoying Neal Stephenson’s Anathem. The story features a culture of monastic academics, dedicated to preserving knowledge through millennia through traditions of thought and an ascetic lifestyle. While the world outside their walls seems contemporary and turbulent, their anachronistic lifestyle keeps them steady in their service of cultural knowledge. Good times.

Anathem, Stephenson has said, is inspired in large part by the Long Now Foundation and their signature project, the Millennium Clock. It sounds far-fetched, but the clock is built to run for 10,000 years. Given the changes in the last 200 years, we can barely speculate on the highs and lows that await humanity in the centuries to follow ours. Given this goal, the clock has one particularly eye-catching design requirement: it must be built so that it can be maintained and repaired with nothing more complicated than Bronze-Age technology. This affordance for radical, unforeseen circumstances makes the clock’s workings and the knowledge of timekeeping they embody accessible.

Back to elections. Almost. In the last few years I’ve seen the rare electronic till failure in restaurants, retail and grocery stores. Most times, there’s no manual backup in place. That is, there’s no stack of blank receipts for making transactions, no procedure to follow after turning it off and on, and then calling support. Or the manager, who will then call support after asking if they tried turning it off then back on. For a store it’s lost sales. Where electronic voting is in place, it can mean lost votes easily numbering more than would have broken ties in both the 2000 and 2004 elections. It’s worth asking if the moments of awkward register crashes at the store foreshadow a loss of knowledge about how to run an election without electronic voting, should electronic means become the norm from classroom to government for several generations.

The potential benefits of electronic voting are hard to resist: faster results would be hard to give up. I know, because I’m watching early results now and yes, fresher feels better. The prospects of securing votes from interference are fleeting when you start to consider the huge number of vulnerabilities that become possible with electronic voting machines.

The paper vote, for the things it can’t do, has benefits that I think outweigh those of speed and arguable security.

Like the clock built to be understood and maintained for 10,000 years, democracy should be practiced in a way that ensures its availability where electronic assistance isn’t possible. Future generations should be able to re-create the technology of a democratic election without more than a ballot, a way of marking it, and rules for casting and tabulating votes. By keeping the technology of voting simple, we also ensure that it remains accessible, reachable through materials on hand and without the need for specialized knowledge. Closer to home, we as voters know that people should never have to leave the poll at all uncertain that the interface recorded their intention honestly.

By doing so, we’ll be preserving one of society’s key mechanisms for consensus decision-making, and enjoying the certainly of knowing one’s vote has truly been cast.