It seems like e-voting snafus are like weather: there’salways a bit of a storm somewhere, and now and then you get a big one. Although we can thank our lucky stars that we haven’t had a real hurricane, an electronic equivalent of Florida in 2000, the recent Arkansas vote-flipping snafu might qualify as a force 9 gale.
And because this time it is clear the outcome of the race was also flipped, this case of Arkansas State House District 45 in 2008 might even be a higher water mark than the case of Florida’s 13th US Congressional District in 2006, one of the most infamous and well investigated snafus.
But this time, the software glitch was not so subtle, and it’s clear that the electronic tally was wrong. The actual outcome of the race is believed correct, though, because of another little software oddity. The touch screen machines printed a paper trail which seems to be an accurate record of the voter’s intent, while the electronic record was not.
Wired’s Kim Zetter has good coverage of the full story (Arkansas Election Officials Baffled by Machines that Flipped Race), so I'll comment only what struck me the most in her story: how election officials responded before the election. Here again I’m harping on the complexity of these voting systems. Election officials noticed the day before the race that the District 45 race was missing from the touch-screen ballots in one precinct. Their response was to print paper ballots for that one race for that one precinct, and have voters use both that paper ballot and the touch screen for the other contests.
Why? The voting systems were too complex for the election officials to know how to fix. There was no guidance on how to respond to the problem. So they did what seemed most expedient to them, and made the election process that much more complex and hard to audit. And certainly if it hadn’t been for the existence (and use) of the paper trail, the incorrect tally might have been an incorrect election result. It’s tempting to fault the election officials for insufficient pre-election testing, not noticing the glitch until the day before the election, not making an ideal (in retrospect) choice of response.
Some of that may be fair. But what I observe, without judgment or blame, is that the unreliability and complexity of these election systems outstripped the ability of election officials to use them - or to use them well to produce a high-confidence result. You might say that the officials should have been more diligent, or you might say that the election systems should have been more reliable, trustworthy, and simpler to use. Either way, it’s a mismatch, and the result is potential failure of the mission: to deliver high confidence election results.