Publius Project
Essays & conversations about constitutional moments on the Net collected by the Berkman Center.

Not the Digital Democracy We Ordered

Essay by Matthew Hindman, December 9, 2008 in response to Internet and Politics 2008: Moving People, Moving Ideas

Additional responses include: The New Activism: Why Volunteering Declined in Campaign 08, by Ari Melber, Participation and Polarization in the Networked Public Sphere, by Henry Farrell, A Response to Working Hypothesis for Internet and Politics 2008, by Sunshine Hillygus, The Revolution of the Online Commentariat, by Peter Daou, and From the Bottom-Up: Using the Internet to Mobilize Campaign Participation, by Dana Fisher.

Without the Internet, Barack Obama would not have been elected President of the United States. But even as information technology has assumed unprecedented importance in American democracy, its effects on the public sphere and on campaign practice remain misunderstood.

Let us start with the obvious: Obama raised $500 million online. Without the infusion of online cash, Obama would almost certainly have lost to Hillary Clinton in the primaries. And while Obama might have won the general election without boatloads of money, his campaign organization and overall strategy would have looked radically different.

Yet Obama’s fundraising success owed much to the fact that online audiences are not, in fact, highly fragmented. Forget the Internet caricatures offered up by journalists, campaign staffers, and even the candidates themselves. The contemporary media environment many not be as concentrated as it was in the Cronkite era, but online news audiences are far more concentrated than (rapidly declining) newspaper readership.

Moreover, as the 2008 campaign ramped up, online audience concentration—already potent—increased substantially. The largest news organizations, along with the largest left-leaning political sites, were the biggest beneficiaries.

The increase in traffic to top sites coincided with greater institutionalization and commercialization in the online public sphere. Media organizations both old and new—from The Atlantic and Time to Politico and The Huffington Post —have rounded up stables of bloggers that function much the way op-ed columnists have long functioned at elite papers. The individual blogs that got most political traffic in 2004 seem quaint by comparison.

What we did not see in 2008, though, was the professionalization of the blogosphere. That is because professionalization was present from the very start. There was never a moment—never—when the majority of blog traffic didn’t go to highly-educated professionals with degrees from Ivy League-caliber schools.

Institutionalization of blogging has brought some benefits, particularly in making the blogosphere marginally less male and less white. In much the same way that many newspapers seem to have a “quota of one” for female and minority voices on their op-ed pages, several top online publications have been forced to expand beyond one gender and a single ethnicity.

Institutionalization has also inspired some prominent efforts to dilute the elitist character of online discourse, though with only mixed success. The Huffington Post’s “Off the Bus” project, for example, has consistently produced quality content. A close look at the bios of those participating, though, suggests that these new voices are not the sort of folks who bring a lunch pail to work.

The increase in financial resources available to campaigns is inseparable from another shift. In addition to all of the other firsts it involved, the 2008 cycle will be remembered for taking database-driven campaigning far beyond previous heights.

Both campaigns constructed statistical models of turnout and partisanship for every voter on the rolls in every swing state. While this was not unprecedented, the Obama campaign’s dedication to information-centric campaigning was a difference in degree so profound that it became difference in kind.

Every aspect of the campaign information infrastructure was dramatically better than 2004. Data from myriad sources—state voter files, the candidates’ Web sites, field contacts, private information vendors—was integrated into a single system. In just a couple of presidential cycles, the statistics used to model voter attitudes and behavior have gone from rudimentary cross tabs, to the same sorts of cutting-edge learning algorithms Amazon and eBay use to predict consumer behavior.

As the election approached, the Obama campaign constantly pushed the latest models back out to field staff, and even the volunteers knocking on doors and manning the phones. And the models worked. As DNC chair Howard Dean told the National Press Club, “We can predict with 85 percent accuracy how you’re going to vote based on your credit card data without bothering to see what party you’re in.”

In truth, Dean’s claim oversimplifies the power of these models (and glosses over some key weaknesses). But his comments do force us to ask some uncomfortable questions.

Campaign databases improve with every election cycle, and the voter models get better every year. If 85 percent is the correct number this year, at what point will the number be 95 percent? How do voters feel about having their political leanings inferred from their subscription to Golf Digest, or—hypothetically—their purchase of condoms at the local pharmacy?

Whatever the implications for democratic theory, there is no denying that database-driven campaigning represents an enormous shift at both the tactical and the strategic level. For any campaign, the key metric is dollars per vote. Information-centric campaigning allows campaign professionals to measure and track the return on their spending almost in real time.

Better information on voters also allows use of the old school techniques that are most effective at getting people to the polls. Databases can provide knowledge about voter preference that neighbors, ward captains, and shoe leather provided in an earlier era. Campaigns can mobilize their base without also energizing their opponents.

Such database-driven campaigning has complex, even contradictory, implications for citizenship. The more predictable the leanings of a voter, the greater the incentive one campaign or the other has to get him or her to the polls. Better modeling of partisanship, turnout, and campaign tactic effectiveness should result in a large and durable increase in voter participation.

Yet more information can also produce new forms of digital redlining. Obama improved over Kerry’s performance partly by being able to reach solidly blue voters in otherwise red parts of swing states.

Now instead of unfriendly counties or regions being ceded to the opposition, it is the unpredictable voters—those for whom data are sparse, or for whom the models don’t predict well—who are being ignored.

Long term, this means that true swing voters are likely to shrink as a portion of the electorate. Among other things, information-centric campaigning is a recipe for increased political polarization.

For some, the notion of political candidates keeping extensive data on every citizen is alarming and creepy. Though a few safeguards exist (particularly with the use of credit information), there remains potential for abuse. Some worry that microtargeting might be Orwellian in its effectiveness, and that telling citizens exactly what they want to hear crosses the line from persuasion to manipulation.

Such fears are, as of yet, overblown. Thus far it is much easier to predict a voter’s preferences than to convince her to change them.

But the single most important thing about information-centric campaigning is that it provides a framework for constant, incremental improvement.

Statistical models of voter behavior will continue to get more accurate with each passing election. This is certainly progress, of a sort. But it is becoming increasingly clear that database-driven citizenship is not exactly the digital democracy that many have been expecting.

Matthew Hindman is an assistant professor of political science at Arizona State University. His first book, The Myth of Digital Democracy, has just been published by Princeton University Press.