Don't Panic, But Do Reflect: Lessons From The Iowa Democrat Debacle

(Mis)Uses of Technology

from the don't-throw-in-the-towel dept

As I write this post, party officials in Iowa are still trying to figure out the results of last night’s Democratic Caucus, while pundits and political opponents have wasted no time in tearing into the Democratic Party, technology, and the very idea of democracy itself. Although there is plenty of reason for criticism, much of what there has been overwrought. Professor Ed Felten’s twitter thread here provides plenty of useful perspective on this:

Nevertheless, there are many lessons to draw from last night’s Iowa caucus mess, but first and foremost is this one: don’t panic.

The biggest reason not to panic is that only the technology supporting the tabulation of the results broke. Whereas the systems for recording the results appear to be working just fine: there was paper, instruments for marking the paper, and (eventually) a way to transmit the recorded results to a place where they could all be added together. Sure, it will take a bit longer to add up all the results for the night, but that’s ok. Because there’s a paper trail we can still know what they are.

That said, there are still some cautionary lessons we as Americans can learn from the experience, not just for primaries and caucuses but for every election for every office, no matter how partisan, run in America.

First, of all the things elections need to be optimized for, speed is not one of them. And should not be one of them. The most important thing elections need is accuracy. While, sure, it would certainly be nice to have accurate results also quickly (and perhaps caucus officials thought the perceived convenience of the app would help towards that end), adding to the list of things it might be nice for election results to be tends to take away from what election results MUST be. And it is not worth the compromise. Which we can understand if we take a moment to think about the consequences if something does go wrong, as something almost always will. The reality is that if election results are delayed, we can still cope. As it is, in general elections results are often not certified until weeks or even months later. There is rarely a legitimate governance need requiring us to know electoral outcomes within hours of polls closing. We can wait a while, and it would probably behoove us to all expect to wait a while, because it’s our hunger for instant gratification that is creating pressure to rely on tabulation systems that are not optimized for what we can’t live without: accurate results we can all believe in. Ultimately nothing else matters, and we should be dubious about any voting solution that offers to deliver anything other than that.

Next is a lesson that applies to more than just elections, which is that just adding tech does not magically make everything better. Often, and as this situation illustrates, it can instead make things worse. Which is not to say that there is no room for technological innovation at all. The world would not be a better place if it were artificially locked into a 20th, or even 19th, century technological environment. Certainly there is a role for technology to play, and for technology innovation. However, not all new technology is necessarily an improvement over what came before. Sometimes what we already had was perfectly fine, or at least good enough for what we needed. And even if a new innovation might someday be better, it won’t necessarily be right away. In the case of the Iowa caucus, it appears they replaced the traditional tried-and-true telephonic support for reporting results with a brand new cell phone driven app, and it is not apparent why. True, sometimes older systems can have their own quirks and imperfections and things we might want to be better, and it’s not a bad thing to want to improve them. But where systems have at least been proven and reliable, any innovation needs to at least be able to deliver that same degree of trustworthiness before it can be considered a replacement, much less an improvement. No technology is magic; no innovation will suddenly solve all of our problems. Even the technology that is now regarded as proven and reliable was once new and untested itself. It takes a lot of time and a lot of work to get a technology to a point where we can depend on it for our most important functions, like elections. Expecting any new technology to immediately be able to be an adequate substitute for what came before, let alone an improvement, is therefore a mistake.

Next, paper matters. Again, the most important thing we need any electoral system to deliver is accuracy. Having a paper record helps deliver it. First, it affords redundancy. Even if our digital systems were perfect and bug-free, it would still be a good idea to have a back-up analog record. But especially while our digital technology is still evolving so rapidly and is still so decidedly NOT bug-free, it is critical to have a way to validate the results it gives us. Even when we think technology is working properly we still need to be able to audit it to make sure it is, and paper lets us do that. And then in those instances, like this one, when the technology has come up short, as it is still very much prone to do, paper means that we can still get what we need: accurate results of how people voted.

There is also one more critical lesson for now, and it’s one mentioned above: something will always go wrong, and so, by anticipating that something will go wrong, we can make it ok when it does. The more complex the system, the more likely something will go wrong. But on the other hand, a system complex enough to have problems is also complex enough to contain solutions.

Elections are complex systems, so there are plenty of opportunities for hiccups. Which is not to say that we want hiccups; naturally we should try to minimize them. The issue though is that no matter how much you try to eliminate them, hiccups will still happen, so what we need to do is be ready. And that means at least two things: one, that we design our systems to have the resiliency needed to overcome those hiccups. For instance, in the case of Iowa, the results could not be transmitted to a central accounting system via the app, and that was a problem. But because there was a paper record they will be able to overcome the problem.

And it means something else, which goes back to the first point made in this post: don’t panic. Don’t rush to throw away faith in our electoral systems when things go wrong, because it is almost a near certainty that something will go wrong. But just because something has gone wrong does not mean it’s the end of the world. Yet if, instead of allowing our systems to self-correct, we instead jump immediately to panic, we create a new problem. The reason we need accurate election results is because we need to be able to believe in them. It won’t matter how accurate they are though if we simply refuse to.

When things go wrong, as they are wont to, we need to take a deep breath and give them a chance to right themselves, which is what will happen in Iowa since they have paper records. Furthermore, as the saying goes, never attribute to malfeasance what can be more readily explained by incompetence. What happened in Iowa was the result of a poor deployment of a technological solution. It was not a conspiracy. It was not the Russians. It was plain old human error to have so heavily depended on this untested mobile app system. And the good news is that we can deal with these sorts of bad decisions. Sure, there are plenty of things we can and should do to minimize these sorts of problems and improve our electoral systems, like better fund and train local election authorities, end reliance on unproven new technologies for supporting elections, ensure we always have paper records, reinforce the franchise, and so on. But there’s no point in attacking our own democracy with distrust. Especially when it is so likely so misplaced.

Filed Under: apps, caucuses, data, election security, elections, iowa, voting, voting security