As you might imagine, it is hard to choose from the manyevents of Election Day 2008 to report and reflect on! But I thought that I’d pick a handful of events that show just how vitally important it is the election equipment be designed carefully – and the consequences of products that aren’t, and vendors that don’t seem to care. I have to say, it’s potentially dire, which is why I’ve picked as many as 3 events to support my claims.
First stop, New Jersey, where some colleagues have reported first hand that the equipment in their home precincts are failing to work ideally, in a seemingly innocent way: "chirp!" That’s the sounds that Sequoia AVC Advantage DRE voting machine should make when a voter casts and electronic ballot, just the same as antique ballot boxes that rang a bell when a ballot was inserted. The point of course is to provide some accountability to voters who might otherwise be tempted to attempt to vote more than once. Whether or not you think that’s a big deal, or even an effective method to help poll workers detect it, consider this: with the antique ballot box, you couldn't '"turn off'"the bell. With the modern electronic voting station, a county official can choose to re-configure to software to be silent. Why was that particular on/off switch a good feature to have? Probably this is a product design mistake, and though simple, it creates a way that a machine can "mal-function" and raise questions about whether it is correctly working in other more important ways.
The vendor simply didn’t consider that designing the machine with many configuration options (bells and whistles if you’ll pardon the pun) simply creates the opportunity for more perceived malfunctions and lower trust. Trust – that’s the missing design concept in NJ.
Next stop, Michigan, where Oakland County Clerk Ruth Johnson complained that their Election Systems & Software M-100 machines where incorrectly tabulating results during pre-election testing. To paraphrase the vendor ES&S's response, the machines were working correctly, but the observed problems were the fault of the people who performed the tests incorrectly. It’s a common refrain from all the vendors, and one that’s often – but hardly always – true. But it misses the point. These machines are so complex and the correct operation so detailed that in thousands of county offices across the country, ordinary people make mistakes that make the machines appear to be malfunctioning.
It’s the same principle in MI as in NJ. The vendor made the machine too complex and user-error prone that the machines do not appear to be worthy of trust. And therefore they are in fact untrustworthy! We don’t trust an important system because somebody else keeps assuring us that it’s really working fine – especially since there have been occasions when similar systems were definitely found to be not working fine.
Last stop, Everywhere USA. You’ll see news reports from all over, starting with early voting many days ago, of voting machines doing "vote flipping." Most of these reports are mistaken, based on voters misunderstanding what they are seeing (as I earlier wrote about Knox County, TN). But again, if the machines are so poorly designed the many ordinary people don’t understand what they are seeing, you’ve got the same missing ingredient.
Now of course many readers know that OSDV is working on trustworthy voting systems (including our first prototype release!) But that isn’t my final point. The really, really big problem here is what happens when the public trust seriously starts to break down, which may be happening right now. Probably the most notable mis-trust factor is potential security problems, which in fact are probably less worrisome than actual reliability problems. But security is the stated reason for threatened litigation to contest election results using the following logic: the results in county XYZ varied from polls; the results contributed to a close race; the machines in county XYZ can be hacked; therefore the results may be invalid and there should be an investigation before election results are finalized.
I am not making this up, just to send chills down your spine just like mine. The title of this news article is enough: Republican National Committee Prepares for Computerized Voting Fraud Legal Battle. Read the details if you dare, but “hackability” is the core argument of threatened lawsuits.
We all know that no computer is perfectly secure, but the voting equipment in use to day has fallen so far from trust – trust which deserve to be able to place in the machineries of democracy – that now they create a potentially huge inroad for the greatest bane of confidence in election outcome – the courts. And I don’t mean one election landing in the Supreme Court like in 2000. It could be many elections landing in many courts across the land, in claims that simply cannot be refuted: “Prove to me that this computer hasn’t been tampered with, and worked perfectly.”
When trust in our election process is reduced to that un-answerable question, then I think that poorly designed, unreliable technology has dragged us almost as far down as we can go down the trust scale. Let’s hope it doesn’t come to pass in 2008.