The internet has been all aflutter with the story of a local Michigan boy done good: Nate Silver. The academic responses have been quite interesting, including a very useful discussion over at Scatterplot of what Nate Silver’s success (and that of other poll aggregating election forecasters) means for social science and its strong distaste for public opinion polls with very small response rates. The humorous responses include the excellent #NateSilverFacts, modeled after Chuck Norris facts, and #DrunkNateSilver, which narrate incredible predictions made by a drunk and disorderly Silver.*
But there were other nerdy election stories. For example, a Maine State Senate candidate vilified for her World of Warcraft hobby won her election. My favorite, however, concerns the differential success of Get Out the Vote technology by the two campaigns.
President Obama’s reelection campaign apparently took to hear the message of Big Data, and they did it well. Time has excellent post-election coverage of the Obama campaign’s fundraising and voter turnout efforts. The campaign adopted such staples of the high tech revolution as massive integrated databases, A/B testing, and social media apps that directed volunteers to contact their friends and encourage them to participate (i.e. vote). The Big Data team apparently made the call to suggest President Obama answer questions on Reddit as a cheap way to reach a large group of persuadable voters.
The Romney campaign’s use of data does not seem to have gone so smoothly. For example, Romney’s campaign piloted a massive “Project ORCA”, an automated tool designed for 29,000 election day volunteers to help get out the vote in key swing states. The tool was apparently a disaster. Sociologists will sympathize with this tech-savvy Romney’s supporters frustrations:
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.
The predictable problems ensued. The “app” was not stress tested, and numerous failures occurred, and the whole system apparently went down around 4pm.
Based on pretty simple forecasts derived from economic fundamentals, Obama was a small favorite to win. And, indeed, Obama won by a small margin (in the popular vote). So it’s hard to say that the Romney campaign did a terrible job. As usual, both campaigns were massive and presumably mostly canceled each other out, moving the results back to the fundamentals-based prediction.** That said, I wonder if Obama’s integrated data analysis and ad buying made his campaign’s ad spending much more effective than Romney’s Super PAC-fueled ad buys, which could not rely on any internal targeting, and thus Obama had a significant advantage despite a smaller total amount of fundraising (and I think others have made this argument as well). Similarly, his superior GOTV efforts might have netted a small advantage in the key swing states, giving him a bit of a cushion in the electoral college (though one he ended up not needing). If so, then I think we can see in this presidential election another triumph of the nerds: Obama’s Silicon Valley-esque “Victory Lab” over Romney’s .. whatever it is they were. Bad web developers, I guess?
* My personal favorite #NateSilverFact, for its appropriate level of nerdery, is “Nate Silver once walked over each of the bridges in Konigsberg exactly once.”
** This problem is similar to one named “Friedman’s thermostat” in economics:
Everybody knows that if you press down on the gas pedal the car goes faster, other things equal, right? And everybody knows that if a car is going uphill the car goes slower, other things equal, right?
But suppose you were someone who didn’t know those two things. And you were a passenger in a car watching the driver trying to keep a constant speed on a hilly road. You would see the gas pedal going up and down. You would see the car going downhill and uphill. But if the driver were skilled, and the car powerful enough, you would see the speed stay constant.
So, if you were simply looking at this particular “data generating process”, you could easily conclude: “Look! The position of the gas pedal has no effect on the speed!”; and “Look! Whether the car is going uphill or downhill has no effect on the speed!”; and “All you guys who think that gas pedals and hills affect speed are wrong!”
Similarly, you can look at a series of elections that all converge mostly to the fundamentals-based prediction and say, “Campaigns don’t matter!” Or, you could say that most presidential campaigns are well-funded and competent, broadly speaking, and thus cancel out. The two are very different claims.