from the wow dept
The new megafile didn’t just tell the campaign how to find voters and get their attention; it also allowed the number crunchers to run tests predicting which types of people would be persuaded by certain kinds of appeals. Call lists in field offices, for instance, didn’t just list names and numbers; they also ranked names in order of their persuadability, with the campaign’s most important priorities first. About 75% of the determining factors were basics like age, sex, race, neighborhood and voting record. Consumer data about voters helped round out the picture. “We could [predict] people who were going to give online. We could model people who were going to give through mail. We could model volunteers,” said one of the senior advisers about the predictive profiles built by the data. “In the end, modeling became something way bigger for us in ’12 than in ’08 because it made our time more efficient.”Compare that, then, to the data driven efforts on the Romney side. First, the campaign ignored all the public polls that turned out to be fairly accurate and plugged in a bunch of their own assumptions when looking at key variables, all of which made them believe they had a stronger position than it turned out they had. But, much more interesting are the stories coming out about ORCA, the Romney campaign's secret computerized weapon in the "get out the vote" effort. It's like the exact opposite of the description of the Obama campaign's data tool.
Early on, for example, the campaign discovered that people who had unsubscribed from the 2008 campaign e-mail lists were top targets, among the easiest to pull back into the fold with some personal attention. The strategists fashioned tests for specific demographic groups, trying out message scripts that they could then apply. They tested how much better a call from a local volunteer would do than a call from a volunteer from a non–swing state like California. As Messina had promised, assumptions were rarely left in place without numbers to back them up.
Romney campaign volunteer John Ekdahl's description of how poorly planned out ORCA was is a must-read. It really sounds like the team there didn't do much testing, and failed to consider how the system would work under load. From the explanations, it also sounds like they didn't do much usability testing, or even think through some basic use cases. It sounds as though either the Romney team didn't think ORCA was ready or they didn't want to "reveal" it until the last minute to avoid tipping their hand to Obama's campaign. The night before the election, they sent volunteers who were supposed to be poll watchers a huge PDF with instructions -- expecting them to print it out, which isn't so easy for everyone these days. They also forgot to tell them they needed to get and bring their "poll watcher certificate" to polling places. And they didn't release the actual app until 6am on election day, giving volunteers no time at all to learn how it worked (or to report bugs).
Now a note about the technology itself. For starters, this was billed as an "app" when it was actually a mobile-optimized website (or "web app"). For days I saw people on Twitter saying they couldn't find the app on the Android Market or iTunes and couldn't download it. Well, that's because it didn't exist. It was a website. This created a ton of confusion. Not to mention that they didn't even "turn it on" until 6AM in the morning, so people couldn't properly familiarize themselves with how it worked on their personal phone beforehand.A different report points to similarly massive problems with other aspects of ORCA, including handing out the wrong PINs, making the app useless in Colorado (and possibly elsewhere):
Next, and this part I find mind-boggingly absurd, the web address was located at "https://www.whateveritwas.com/orca". Notice the "s" after http. This denotes it's a secure connection, something that's used for e-commerce and web-based email. So far, so good. The problem is that they didn't auto-forward the regular "http" to "https" and as a result, many people got a blank page and thought the system was down. Setting up forwarding is the simplest thing in the world and only takes seconds, but they failed to do it. This is compounded by the fact that mobile browsers default to "http" when you just start with "www" (as 95% of the world does).
By 2PM, I had completely given up. I finally got ahold of someone at around 1PM and I never heard back. From what I understand, the entire system crashed at around 4PM. I'm not sure if that's true, but it wouldn't surprise me. I decided to wait for my wife to get home from work to vote, which meant going very late (around 6:15PM). Here's the kicker, I never got a call to go out and vote. So, who the hell knows if that end of it was working either.
Then at 6PM they admitted they had issued the wrong PINs to every volunteer in Colorado, and reissued new PINs (which also didn't work). Meanwhile, counties where we had hundreds of volunteers, such as Denver Colorado, showed zero volunteers in the system all day, but we weren't allowed to add them. In one area, the head of the Republican Party plus 10 volunteers were all locked out. The system went down for a half hour during peak voting, but for hundreds or more, it never worked all day. Many of the poll watchers I spoke with were very discouraged. Many members of our phone bank got up and left.Once again, I'm sure that the Obama system wasn't nearly as perfect and all-knowing as the Time article describes. Nor was the Romney system a total disaster as described in those links -- but it is fascinating to see these stories emerge following the election. Given the Obama win, along with these stories about that datacrunching effort and the worldwide reverence towards Nate Silver, you can bet that everyone gunning for the 2016 nomination is going to spend some time trying to build up a killer technical/data team.