ITP Techblog

The race to become for the US Democratic presidential nomination has had many in the tech sector shaking their heads in disbelief following the Iowa caucuses last week. Forget whether it's Sanders, Warren, Buttigieg or Biden out in front, what was with the technology?

The app was created by an unknown company called Shadow Inc, whose founders' main qualification appears to be that they had previously worked on Hilary Clinton's campaign. It was supposed to provide a way for 1700 precincts in the Iowa caucus to report the results of the Democratic Party's first state caucus. It failed spectacularly. The results were not only very late, the New York Times estimates at least 10% of the precincts have improperly allocated their delegates.

The publication has myopically followed the debacle in several articles over the past week, and it's clear that while there were multiple reasons for the meltdown, at the heart of the it is a story that tech-watchers in New Zealand will be very familiar with.

To sum it up - rushed to market, inadequate user training, no plan b.

The app, which cost about US$60,000, was apparently developed over a couple of months and was so new that there wasn't time to get it approved by the Apple store. "Instead, the app had to be downloaded by bypassing a phone's security settings, a complicated process for anyone unfamiliar with the intricacies of mobile operating systems, and especially hard for many of the older, less tech-savvy caucus chairs in Iowa," notes the New York Times.

This, from an app that required each precinct leader to report 36 different figures, along with two separate six-digit verification numbers. That is 1700 people expected to input a reasonably complex dataset using an app they had probably just downloaded. That doesn't even account for the potential connectivity issues that might occur.

When a bug or "coding issues" (as they have been described) meant that the app froze when some of the precinct leaders tried to report the results, calls to the state party hotline jammed. This was partly because precinct leaders had switched to manual reporting, and also due to reporters phoning in to get results. It didn't help either that a pernicious website posted the hotline number and encouraged people to "clog the lines".

Meanwhile, some leaders were told to photograph their results and email them through to HQ. Trouble was that no one was monitoring the inbox that night, and when they discovered 700 unread emails the next day, they opened them up to find photos sent to them sideways and "volunteers had to crane their necks to decipher the handwritten forms."

It is this last image that sticks in my mind as a kind of visual metaphor for the entire disaster.

So, what can we learn from Iowa's shame? Here's a starter for five:

1. Technology is hard - there are numerous things to consider, not least of which is usability and security. There is a reason why those big tech companies employ thousands of very smart people.
2. Don't rush to market if you are a large organisation where trust and credibility matter. Tech failures can ruin hard-earned reputations overnight.
3. Never overestimate user ability. The more complex the tech, the more training required.
4. Start simple - what if those 1700 precinct leaders only had to send one data point via the app (to be counted on the night), and the rest of the data via email (to be counted in the days that followed?
5. Always have a plan b, that is itself well tested. In this case, a way to quickly revert back to manual reporting might have helped, but one of the issues was that the Democratic Party had decided on a much deeper level of reporting than in previous years, so there was no old system to fall back on.

The Iowa app example will be used by those against online voting here as an example of why it shouldn't be in implemented in New Zealand. And really, it's hard to argue with them.