:format(webp)/cdn.vox-cdn.com/uploads/chorus_image/image/53727005/si_02_720.0.jpg)
“Big data revolutionized the way American politicians win elections. In the process, it broke American politics.”
This was the central claim in a recent column by NBC’s Chuck Todd and Carrie Dann. The argument is simple enough: Politics “broke” because the system is paralyzed by polarization, and it’s paralyzed by polarization because technology and demographic data have made it easier (and less risky) for campaigns to target their base instead of appealing to a broad swath of voters.
It’s an interesting thesis.
Big data — defined by Todd and Dann as a “combination of massive technological power and endlessly detailed voter information” — has certainly changed the way campaigns are conducted. Like corporations, campaigns now know far more about their constituents than ever before — what they read, which movies they stream, which shows they watch, where they shop, which products they buy.
This allows campaigns to identify their most likely voters and target them with ads and favorable content. The result, increasingly, is that candidates talk only to voters disposed to agree with them, as opposed to persuading those who don’t.
In the end, Todd and Dann write, mobilizing likely supporters is “faster and far less expensive than persuading their neighbors.” So that’s what campaigns do: ignore the center and concentrate on the most fervent supporters.
All of this creates what Todd and Dann call a “crisis of governing,” in which the halls of Congress are “populated by lawmakers who feel beholden not to all their constituents, but only to their supporters.”
Todd and Dann’s argument is probably too simple. As Seth Masket points out, they’re slightly confused about the applications of “microtargeting,” and they seem to conflate correlation (increasing polarization) with causation (the birth of big data and computing power). But questioning the utility and implications of big data in politics is still worthwhile.
In this interview, I talk to Eitan Hersh, a professor of political science at Yale University and the author of Hacking the Electorate: How Campaigns Perceive Voters. Hersh’s research focuses on how campaigns use data to mobilize voters. He also studies the impact of new technologies on candidate behavior.
Here, I ask him what the big data revolution has meant for American politics, and if it’s really as bad as Todd and Dann claim it is.
Our conversation has been edited for length and clarity.
Sean Illing
Tell me about your research background.
Eitan Hersh
Much of my work relies on using voter registration data and campaign databases to study American politics. My most recent book, Hacking the Electorate, is about how campaigns use public data and microtargeting to isolate and appeal to specific voter groups.
So I’m interested primarily in how campaigns use data to win elections, and more broadly in how data is altering our political process.
Sean Illing
How do politicians “hack” the electorate?
Eitan Hersh
The electorate in America is huge and hard to understand. The idea that any campaign or organization can truly perceive the attitudes and interests of 300 million people is crazy. What campaigns have done over time is gather more and more information so that they can better understand voters. And that's really the hack.
Sean Illing
So it’s about using data — the books people read, the shows they watch, the causes they support — to understand and drive political behavior?
Eitan Hersh
Right. The hack consists in asking — and in discovering — what data can be used to manipulate or mobilize voters. If you have enough data, you can predict how people will behave, how they will vote. Campaigns have developed sophisticated ways of doing this, and the growing availability of data is accelerating that process.
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/8168205/57d54546_0541_4d1a_963a_210d85c1b478_1.e7ad398fbf99cbe362112348a65a86ca.jpeg)
Sean Illing
When we talk about “big data,” what are we talking about exactly?
Eitan Hersh
It means population data as opposed to sample data. Until fairly recently, campaigns relied on survey data, which is obviously limited. But big data has changed that. Campaigns now have access to data about the entire population, not just a narrow sample.
Sean Illing
Where does this data come from? Who collects it?
Eitan Hersh
As far as data in campaigns, individual-level data comes from voter registration databases, other state-by-state public records, and consumer databases. Commercial online targeting data is different — that can come from firms like Facebook and Google. But campaigns have access to plenty of public data sets. Lots of small-area data, for example, come from the Census Bureau (e.g., neighborhood-level income, race, industry data). Precinct data come from state records as well.
Sean Illing
You said a minute ago that campaigns have used this data to develop sophisticated ways of targeting and engaging prospective voters. What does that look like in practice?
Eitan Hersh
For get-out-the-vote mobilization, campaigns use information that publicly listed about most voters like their party affiliation, history of participation in general and primary elections, their age and gender. In some states, the public record indicates their race. In other states, campaigns will predict a voter’s race based on their name and location. They use these data to figure out which voters are their supporters (and which are their opponents). For example, if you are a Democratic campaign and you know someone is a registered Democrat or voted in a Democratic primary, or is black or Latino, or is under 25 years old, you can make a very good guess that they are a supporter.
Once campaigns use the data to find supporters, they can then use canvassers, phone bankers, direct mail, and other tools to encourage them to vote.
Sean Illing
And how have these methods altered the strategic approach to campaigning?
Eitan Hersh
Campaigns can now do a lot more targeting based on individual-level characteristics or even neighborhood-level characteristics, as opposed to thinking generally about what the typical American or the typical constituent wants to hear. So a candidate now knows what her typical voter looks like, how old they are, what gender they are, where they live, where they shop, etc.
So there's a lot more nuance in terms of how candidates understand who their supporters are, and therefore more nuance in terms of how they engage people.
Sean Illing
In other words, candidates are thinking of constituents less in terms of people who live in their districts and more as people who have — or are likely to — vote for them?
Eitan Hersh
Think of the famous Romney 47 percent line or Hillary Clinton's "deplorables" comment. Both of these statements are forms of representation where you dismiss the people who are not your supporters. When I think of my job as a politician, I can think of it as serving the entire population, or I can think of it as serving my people, my voters. So I'm looking for 51 percent, not 60 or 70 percent.
Big data makes it easy for candidates to dismiss their opponents. They now know, with greater and greater precision, how people voted and how they're likely to vote in the future, and their campaigns reflect that.
Sean Illing
I take that to mean that candidates can now circumnavigate the whole process of persuasion. They're not looking to persuade people with ideas or arguments; they're just identifying the people most likely to vote for them and excluding everyone else.
Eitan Hersh
That's exactly right. I think this is a big reason why Hillary Clinton lost. They banked on a campaign focused on mobilizing voters who voted for Obama and did not engage in the kind of persuasion that might have shifted some of these rural voters to their side.
Sean Illing
Implicit in this Big Data approach to politics is the assumption that most voters are essentially unreachable. Perhaps that’s true. But I wonder if operating on this assumption has the effect of making us more polarized and more entrenched than we otherwise would be.
Eitan Hersh
I think a lot of this comes from the voters. The easiest way for the average voter to engage with politics is through party or team affiliation. So I think the voters want this, or they respond to it, in any case. And politicians prefer this because they know who's on their team, and they're not worried about defections.
So this isn’t necessarily conducive to good government or policymaking, but it does simplify the process for both voters and candidates.
:format(webp):no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/8168215/imgres.jpg)
Sean Illing
So is it too heavy-handed to say that big data has rendered persuasion unnecessary?
Eitan Hersh
I think that's a little too heavy-handed. A lot depends on who your opponent is and what kind of strategy they employ. Look at a state like Pennsylvania, which leans Democratic in terms of voting registration. If both sides played exclusively to their sides, the Democrats would win every time. But that didn't happen this year. Clearly, some Democrats were persuadable, and Trump managed to persuade enough of them.
I take your fundamental point, but it's not quite true to say that persuasion is unnecessary or irrelevant.
Sean Illing
How do these changes at the campaign level impact actual policymaking? If candidates are talking more and more to their own teams, and if they’re only responsive to their people, does that not lead to uncompromising governing strategies?
Eitan Hersh
There is much more competition in American politics than it seems. If there's one thing we learned in 2016, it's that there is no such thing as a firewall. There is still a price to be paid for inefficient governance.
But the strategy of focusing on the base is visible both in campaigning and in governing. In the governing setting, individual members of Congress seem to think that their best strategy for keeping their jobs is dig in on partisan fights rather than to compromise. It’s possible that new data about the preferences of their voters informs them that such a strategy is a winning one.
Sean Illing
Has big data improved our politics in any way? Has it made politicians more responsive to their constituencies?
Eitan Hersh
People who run microtargeting firms have told me that these tools allow them to better understand their voters. In the same way that Amazon wants to understand more about us in order to provide us with the products we want, campaigns want to understand what voters want in order to speak to those issues.
This, at least, is the logic you'll hear from campaign staffers.
Sean Illing
I understand that logic, but there's an obvious problem: It assumes the only job of a politician is to tell voters what they want to hear, not what's true or what's necessary. Treating citizens like consumers works, but it doesn't produce wise policy outcomes; instead, it reinforces biases.
Eitan Hersh
I agree, and this is a big problem. What's happening in the political marketplace is very similar to what's happening in the media marketplace. All the incentives encourage publishers to go hyperpartisan, to find a niche market and appeal exclusively to it. This approach will get clicks, but it won't produce more informed readers. The same is true of the political marketplace.
Sean Illing
So what’s the endpoint here? Will big data reduce politics to mechanics, where algorithms are God and voters are ones and zeros?
Eitan Hersh
Perhaps. We're moving in that direction. But the market is always open to disruption. Trump's victory is an example of this. His campaign, in many respects, exploded the prevailing model. Clinton had an incredibly sophisticated data operation, and she still lost.
So politics is still more than ones and zeros.
Sean Illing
Technology develops faster than culture. We develop tools before we fully understand their implications. Is big data an example of technology outpacing politics and changing things in ways we didn’t expect and now can’t control?
Eitan Hersh
Yeah, I think that's mostly right. Although I'd rephrase it: Any kind of technology is a tool, and a tool can always be used for purposes for which it wasn't originally intended. We don't really know how ad targeting online or the use of big data for marketing or campaigns will be used or how they will affect outcomes.
So there’s a lot of uncertainty here, and we’re all living through the experiment.
Sean Illing
The emergence of public relations in the mid-20th century profoundly changed our politics; it turned campaigns into marketing contests. Politicians became products, and the techniques used to sell cars or toothpaste were adapted to campaigns. Does the emergence of big data represent a similar revolution?
Eitan Hersh
That's a tough question. I'm not sure I think about it in those terms. There are big differences between commercial marketing and campaign marketing. For one, if you get 49 percent of the toothpaste market, you're a really rich company. But if you get 49 percent of the market share in an election, you have no power.
So the incentives are different: Every voter matters more than an individual consumer in a commercial market, and candidates are limited in terms of their ability to exclude segments of the population.
Sean Illing
That's a fair point. My concern is that the more candidates target voters disposed to agree with them, the less likely they are to make arguments and defend positions. Instead, we just get an exercise in meat tossing and choir preaching.
Eitan Hersh
I understand that concern. Still, I think the panic over big data in politics is a little overblown. There is still an open political process with market incentives, and when a politician is too misaligned with her district, the market will deliver an alternative in the next election.
Big data hasn’t changed that fact.
