How the Economics of Journalism Explains 2016's Information Bubbles
Jay Hamilton, a Stanford professor who studies media business models, sees similarities between some of today's outlets and the partisan press of the 1850s.
Despite frequent criticism from the president-elect and despite drawing historically low levels of trust from the public, the media—or at least parts of it—actually did quite well in serving the public good during the run-up to this year’s election. That’s the view of Jay Hamilton, a professor of communication at Stanford and the director of the school’s journalism program, who says that throughout the campaign, enterprising national reporters (particularly those at The New York Times and The Washington Post) were consistently turning out stories with new information that was relevant to voters.
But Hamilton, whose recent book Democracy’s Detectives is about the market for investigative journalism, is not as sanguine about the performance of much of the media industry in the past year. As reporting at large national publications has gotten ever more sharp-eyed—and as these publications have amassed an increasing share of reporting awards as a result—jobs at regional outlets that aren’t based in prosperous coastal cities have been drying up, limiting those papers’ ability to scrutinize local politicians and organizations. (Hamilton also counts Facebook among “the category of media institutions that failed us” for its lack of response to the viral circulation of blatantly inaccurate articles during the campaign.)
An economist by training, Hamilton thinks about these developments in part as a product of the way that many media organizations make money online. I talked to him recently about the incentives that news outlets have for digging up new (and true) information, and how those incentives have changed in the past century and a half. The interview that follows has been edited and condensed for the sake of clarity.
Joe Pinsker: Given what you know about the history of the media industry, do you think that Trump could have gotten elected with the media infrastructure of 10 years ago? 50 years ago?
Jay Hamilton: That's an interesting question, because if we went back 50 years ago, the media-bias question was, "Why is the media so conservative?" Newspapers were predominantly endorsing Republican candidates. I think it would have been interesting with them as gatekeepers—there are so many things that Donald Trump has said that wouldn't make it into the newspapers because of language.
As for 10 years ago or 15 years ago, I talked with Marty Kaiser, who's the former editor of The Milwaukee Journal Sentinel, about this year’s election coverage. He said if you look at the newspapers in the Midwest, 15 years ago they would have had the resources to do many more “What are the voters thinking?” stories. They would have fanned out across the state and talked to people. In Wisconsin, there are 72 counties, so they would have talked to the Democratic county chairman and the Republican county chairman, and they would have perhaps recognized the type of economic issues that may have been driving some of the decisions in those areas.
We’re in a world where, in the last 10 years, 40 percent of newspaper jobs have disappeared, and they have not disappeared evenly across the United States. Jim Tankersley has done great work that shows that media jobs are actually increasing in D.C. and L.A., but they're decreasing substantially in Wisconsin and Pennsylvania and Minneapolis. If you go back to the 1990s, there were people like Barlett and Steele, who got two Pulitzer Prizes at The Philadelphia Inquirer for writing about D.C. issues from a Pennsylvania perspective. They did that series, “America: What Went Wrong,” linking up change in economic policy with what's going on on the factory floor. But all of that has disappeared. People at the metro dailies, the radius of what they write about has become much more restricted. I used to think of that as a local or state problem, but if we’re keeping score through the electoral college, our failure to understand what was going on in Wisconsin or Michigan led people to be very surprised.
Pinsker: One argument that I have heard is that national news organizations “missed the story” on parts of the Trump coalition because their staffs are concentrated in the same large coastal cities. What do you make of that claim?
Hamilton: I got my Ph.D. in economics from Harvard in the 1980s, and when we looked at international trade, we kept score with the Pareto principle, and we would say, “This is a good move for society if the pie gets bigger, if winners could compensate losers.” But the key assumption there is, “if winners could compensate losers.” When running that kind of cost-benefit analysis, jobs are not a consideration, because you assume that people go to their next-best alternative use of their time. If you take that framework, that wouldn't be very helpful if you didn’t also ask the question, “If you're 55 and out of work in West Virginia, what’s the specific connection between the policies being proposed and your life?” I think it’s not enough to drive through and say, “There are opioids here.” It’s a deeper thing about making that connection between policies and people’s lives, and then understanding how people could use their vote as an expression of anger and frustration.
Pinsker: In the attempts to arrive at that kind of understanding, what kinds of stories are likely to get told and which ones get ignored?
Hamilton: I think there are five incentives that lead information to be created. “Pay me”—that’s the subscription model. “I want to sell your attention to somebody else”—that’s advertising. “I want to change how you think about the world”—that’s nonprofit. “I want your vote”—that's partisan. And “I just want to talk”—that's expression. All of those incentives are biased against telling the stories of low-income people. They’re not subscribers, their attention is worth less to advertisers, they vote in lower numbers, and in some cases they’re not linked up to broadband, so you’re not hearing their voices on the internet. And they’re definitely not donors to nonprofits. So there’s a bias against telling stories from their perspective.
Pinsker: I know that you're a fan of journalists’ using computing to help them do their job. What opportunities do you think it offers for solving some of these problems you’ve mentioned with local reporting?
Hamilton: There are two things that are possible that I like to look at. One is supply-side: lowering the cost of discovering stories with better use of data and algorithms. One of the problems with local public-affairs stories is that you're not sure there's a story. If you go to a sports event, you know there's going to be a story. If you go to a city-council meeting, it's a jump ball—you have no idea. I think that using algorithms to discover stories is a way to lower the cost.
It’s almost like electronic tip generation. Open Secrets, the data website, has something called the Anomaly Tracker. Their anomalies are things like, “Show me legislators who have sponsored a bill that is only being lobbied for by one company.” Or, “Show me legislators who have 50 percent or more of their contributions coming from out of state.” Derek Willis, when he was at The New York Times, used to have an alert any time there was a vote in Congress after midnight. As I used to tell my kids, nothing good happens after midnight. He had that programmed.
On the demand side, to me, the dream would be a site that knows you so well that it can tell you a story in a personalized or engaging way. It knows what you’ve read, knows that you know the distinction between Sunni and Shia, knows that you like video and documents. If there were a site that knew you as well as Amazon, it’d allow for product differentiation not in terms of dialing up or down political bias like cable channels do but in terms of offering you what you’re most interested in, in the dimensions that work for you. You might be more likely to engage with a story about public policy if a site could personalize the story by localizing the examples, building on your likely knowledge based on prior reading or watching, and giving you a graph instead of several anecdotes if you had demonstrated a facility with and interest in data. We’re seeing glimmers of that, but to me, once you have that product differentiation, you can charge, and people would pay for it.
Pinsker: Another thing that people have focused on after the election is how easy it is for people to surround themselves with information that supports their worldview. Can you give me a sense of how much of a departure from history this represents?
Hamilton: In the 1850s and 1860s, we had a partisan press, and the words “Democrat” and “Republican” were in the names of the outlets. In the 1870s, you had the very expensive high-speed presses invented, which could print thousands of papers per hour, and it became very important for publications to have a very large circulation, because you could spread the cost of that printing press across many people. At the same time, you started the evolution of national brands, thanks to railroads, and advertisers wanted to be able to deal with one large paper rather than four small ones in the city. So, starting in the 1870s and going through the 1900s, you had this evolution of objectivity as a commercial product. Papers became independent, they took “Democrat” and “Republican” out of the title, and it made business sense to try to speak about politics in a more nonpartisan way.
If you move forward to the cable world or now especially the internet, you see a drop in fixed cost. Think about it in terms of the internet: It’s very cheap to have a website, so you can now target a smaller niche, a more partisan niche. So, high fixed costs gave us objectivity and nonpartisanship; low fixed costs now give us back a world of partisanship. You no longer need an expensive press and trucking system to reach the world. You also don’t need a highly developed reputation built over time through a track record and expensive reporting to draw readers in. With the “fake news” generated in the 2016 election, individuals could make up stories, post them online, and see them circulate rapidly through the Facebook ecosystem.
In the past, many advertisers would shy away from appearing next to demonstrably false content since it would undermine their brand with the broader public. For the fake stories that went viral on Facebook, it is hard to even know which advertisers appeared next to the false stories because the impressions are ephemeral and because the ads can be targeted so that individuals see different products depending on their demographics and past web use.
Pinsker: Given that system of costs, what do you think would need to change in order to pop information bubbles, or at least enlarge them?
Hamilton: That’s one of the things that's interesting about fact-checking. If you look at somebody like Bill Adair at Duke, the creator of Politifact, his dream is to have a fact-checking database and if you have a smart TV, you could have an app so that when an ad comes on or when a program is being played and a statement comes up, it tells you, “That was true,” or “That’s false.” And we’re seeing glimmers of that, with Google experimenting with a tag that appears next to stories in your search results that have been fact-checked. In a world where we’re getting closer to real-time transcription and checking that against previous fact-checks, I think it's going to be easier to see what’s right and what’s wrong.
Pinsker: It’s a cool idea to have a real-time fact-checking bot superimposed on your TV, but wouldn’t most people just turn it off if they had it? Isn’t it more comfortable not to have that?
Hamilton: It depends. I think that’s a great empirical question, because if you have it on, you get to see the person that you detest—you get to see their falsehoods as well as those of your candidate. I think that’s an interesting race, about whose truth you want to know.
But still, yes, there’s always going to be this problem of the market for truth, which relates back to rational ignorance. If I believed my car ran on sand, I wouldn’t be able to drive. But if I believed that Barack Obama was born in Kenya, I might not pay a price. In fact, I might get high-fived or even run for president. So that’s always going to be there, that battle between our information demand as audience members and our information demand as voters. But I do think that computation holds some potential there.
Pinsker: In terms of the industry’s current economics and resources, what do you think the media is particularly well-suited for in covering Trump’s presidency, and what do you think it's particularly poorly suited for?
Hamilton: I think it's well-suited for covering the watchdog function of how policy is made and what the backgrounds are of the people who are coming in, because that's something they've done before.
And then, after missing the local story in the Rust Belt states, I think hopefully people will learn from that. I think that hopefully they won't only go back four years from now. It would be very interesting to check back in and see how the lives are changing or not with the folks who are essential to his election.