In addition to the $2 billion spent this election cycle, the Barack Obama and Mitt Romney campaigns have also been feverishly crunching numbers that have nothing to do with dollars. There's another entity that's proving just as big in this presidential election: big data. However, considering both campaigns have spent a combined total of $13 million on data mining and analysis, it may not be all that different.
There was a time when political campaigns only cared about which broad swath of the population you fell into: were you a middle-aged white male, a young college student, an elderly African American woman, a highly-educated Latino, and so on and so forth. But now, they are increasingly concerned about you, the individual.
The Internet provides a vast storehouse of information on each of our online habits and hangouts, all of which can be as useful to campaigns as they are to online marketers. Whether you're a dyed-in-the-wool partisan on either side or a persuadable centrist can determine which voters campaigns want to go after.
Using algorithms that match the attitudes of voters on specific issues with individual behaviors and tendencies, even the most trivial little tidbit becomes significant: where you shop, which team you root for, which petitions you sign, who your friends are, what mobile device you use etc. This has led to seemingly irrelevant factoids like Obama supporters frequenting Red Lobster and Romney proponents preferring Olive Garden becoming much-hyped stereotypes.
As Sasha Issenberg explained on NPR's On the Media, "They're running algorithms that are basically looking for patterns between that big mass of data they have about each of you with the information that their polling tells them about specific attitudes about the election that's underway."
When campaigns allow a user to log into their campaign web sites via Facebook Connect, they are doing more than just providing ease of use. They are also acquiring consent to harvest the user's Facebook information. With an open door to Facebook profiles, the sky's potentially the limit to personalized political messages.
In addition, campaigns are beginning to use concepts from behavioral science -- such as embarrassing citizens by releasing lists of their voting history, or simply coaxing them to perform their civic duty through peer pressure.
A study during the 2010 midterm elections showed that just posting an "I voted" banner on voters' news feeds pressured 60,000 people directly -- and several thousand more indirectly -- to vote.
There is something innately democratic about communities on the Internet inspiring their friends and relatives to perform a civic action in the real world. Bragging rights, as exemplified by the "I voted" banner can go a long way in the online world. By the same token, so can badges and testaments of accomplishment.
The Obama campaign's digital campaign network -- already stellar in 2008 -- has only grown since. As Tim Murphy notes, Harper Reed, the chief technology officer for the president's reelection campaign, has brought his online community expertise from his days at the t-shirt company, Threadless, which thrives on a brand of crowdsourcing that inspires people to contribute time and effort into making a product while making it fun.
Dashboard, the online organizing community developed by Reed for the Obama campaign works on a similar principle. Data collected for a voter's profile is collected by breaking down information gathering into a series of fun and engaging tasks. Campaign volunteers can also use Dashboard to track their real-time impact on fundraising and canvassing relative to their peers.
Just by viewing a user's social graph on Facebook, campaigns have also been able to urge volunteers to contact their long-lost pals, coworkers or other people in their real lives to spread the message.
As opposed to virtual election events held on social media (think the MySpace Primary in 2008), which may give a false sense of doing your duty just by clicking a button, such strategies actually translate social media action into real-world consequences.
Even the more old school Romney campaign has rebounded from past missteps (most notably misspelling America) on its mobile app to establish a more effective data-gathering machine, recruiting former employees from Apple and Google Analytics, and tapping into consumer data from third-party firms, making this campaign a veritable digital data war.
With fewer and fewer undecided voters remaining, this sort of microtargeting becomes all the more important. While campaigns focus exclusively on a few battleground counties in a handful of swing states, online targeting can help reach voters who would normally have no access to traditional campaigning. It's also useful in an age where computers, tablets and even smartphones are replacing televisions, especially among young voters.
Is it scary that presidential campaigns have so much information about us? Experts say that most of the data provided to campaigns by credit bureaus and market research firms is anonymous and provided in aggregate form, so individual data cannot be extracted. Data from InfoUSA, for instance, can only be read by computers, which generate a targeting score for a particular voter. This score, in turn, instructs a volunteer on the issues to focus on while soliciting. If you're one of several thousand IDs in a database, a lot of people probably wouldn't care about their information being out there.
Nevertheless, it is still troubling to not know how companies collect this information and how campaigns use them. Can campaign staff pinpoint a particular voter by matching consumer data with other information in his or her files? When friends of friends on Facebook are asked to solicit voters, are they given sensitive information about their friends, such as financial history?
Privacy practices here are not regulated like in the commercial sector due to protections afforded by political speech, as Murphy points out. Just as the federal Do Not Call list doesn't apply to political campaigns, it is unlikely that such methods of regulation (like Do Not Track) would apply to political targeting.
It is one thing to use information that a voter willingly provides either on a campaign Web site or on his or her Facebook page. But combining this with voting records, political donations, Web-browsing behavior, and consumer data can produce a composite identity of a person. In addition, cookies that automatically download on your computer upon visits to candidate Web sites almost function like virtual spies, tracking your online behavior and reporting it back to campaigns.
No wonder a majority of Americans are against behavioral targeting and tailored advertising for political or other purposes, as a study by the University of Pennsylvania's Annenberg School of Communication revealed.
Some also worry that such data could be used by certain groups to feed the ugly side of political elections: not merely to rally voters for their cause, but also to thwart those on the other side by providing erroneous election information. This may sound a little far-fetched, but with repeated attempts at disenfranchising voters this season, it's not altogether unlikely.
A report by the Electronic Privacy Information Center speculates on these possibilities, emphasizing the need to educate and protect voters against exploitation and misinformation.
However, even while the mainstream media obsessively covers poll after poll with slight differences in statistically insignificant margins this election, there is almost no mention of data mining and the resulting concerns about voter privacy. As PJ Vogt pointed out on On the Media, journalists tend not to report on stories that they don't have a lot of information on. But even this lack of information needs to be covered. "Campaign reporters need to focus on this stuff. This is the real battlefield," he said.
It's true. If this election goes down to the wire, like most predictions and polls suggest it will, knowing if someone chooses Red Lobster over Olive Garden could mean the difference between winning and losing.
Campaigns are not the only ones tracking your online behavior. Simply by measuring the number of Google searches for voting information in various demographic areas, Economist Seth Stevens-Davidowitz is predicting voter turnout.
By comparing the relationship between Google searches for voting information in 2004 and 2008, Stevens-Davidowitz makes a strong case for the method's reliability in not only predicting voter turnout in various demographic groups including African American, youth and Hispanic, but also voter sentiment during key events during an election season, such as rumors of Obama being a Muslim and the selection of Sarah Palin as McCain's running mate.
Regardless of who ends up winning this election, we have one uncontested winner already: big data. From getting citizens into polling booths to vote to predicting if they actually will, its influence is everywhere.