Understanding Fake News, From a Media Manipulator

An “ordinary citizen” noticed that billboards advertising for the movie I Hope They Serve Beer in Hell were vandalized and sent pictures to the blog Curbed Los Angeles, attracting more attention to the controversial film. In his book Trust Me, I’m Lying: Confessions of a Media Manipulator Ryan Holiday cops to having vandalized the billboards and sending the tip himself—as a publicity stunt for his client, the film’s writer Tucker Max.

Holiday was routinely gaming the media, playing bloggers desperate to make a living off the trickle of revenue they accumulate from page views. The only way many bloggers can make a living, Holiday explains, is by constantly posting content that gets thousands of clicks. This made them happy to create new posts out of cosmetic rewordings of press releases or anonymous tips, without taking time to properly check sources or the accuracy of tips. Viral content from blogs, in turn, can turn into “real news” when it is picked up by better known and more mainstream news sources. The fact that something that is being talked about can serve as justification for further coverage, even if there was no car behind the headlights.

From his memoirs, Holiday makes it sound easy and common for people working for major brands—he worked for American Apparel in addition to Max—to exploit the material insecurity and burnout of bloggers, “trading up” to broader coverage in the mainstream media. The narrative we commonly hear about fake news is that it’s ideologically driven. The awful (left or right wing, pick your poison) media is trying to get us to buy into their evil politics. But in Holiday’s telling, for a large proportion of fake news creators there isn’t a plan. Thousands of writers are hustling to make money off a few minutes of your attention, none of them following a conspiracy any more complicated than “do what seems to be working.” The media landscape governed by these rules is going to have a distinctive character, but it’s not going to be due to a central controller dictating the direction of the news. Instead the global pattern is going to emerge as the product of many interactions between lower level units.

The second half of Holiday’s book is a recanting of his underhanded tactics and a broadside attack at the careless, self-interested bloggers who post sensationalized half-truths. But his conclusions about the need for people to choose to walk away misses what is so compelling and insightful about his own account. The problem he is identifying is one with the ecosystem. Like an environmental ecosystem, all the different parts—bloggers, brand managers, journalists at established outlets—determine the character of the media through their mutual interaction.

When scientists analyze a dynamic system like this, they seek to characterize the patterns that will necessarily recur and stabilize based on the way the variables interact with each other. If you’re concerned about a population of fish falling too low, for example, you don’t go out and scold individual fishers (or, in Holiday’s case bloggers). You understand that fishers are responding to the incentives inherent in the system, and so you try to address those underlying incentives. In evolutionary game theory, you would try to capture the key ways different parts of the system interact, modeling the major dynamics—for example, the tendency for bloggers who spend no time on fact checking to outcompete those who expend energy fact checking. Then you can examine what other strategies can outcompete that strategy, and under which environmental conditions they will do so.

In a scientific paper that “went viral,” evolutionary modeler Paul Smaldino did just this for the scientific publication system. He modeled the pressures of the “publish or perish” academic research system, including the incentives to pump out more research papers. His model demonstrated how cutting corners could provide a material advantage, to the point that it would become the norm for academics to publish shoddy, unreliable work. The system was set up so that, as John Ioannidis put it “most published research findings are false.”

At the time, many researchers were discussing replication as a solution to a culture of sloppy work. His model explored what would happen if you introduced a group of people routinely replicating others’ research—in essence a group of scientists double-checking the work of their colleagues—and instituted some sort of penalty for having done sloppy work. Contrary to expectation, even adding a sizeable number of replicators to the field would not change the underlying incentives. Cutting corners was so profitable that the best play for researchers was still to just do it and hope not to get caught—people who worked more slowly and carefully still couldn’t compete.

The dynamics of the model surprised Smaldino, who thought replication was a useful and sensible strategy. In more recent work modeling the scientific process, Smaldino examines a wider range of possible reforms to the scientific system, finding that many work in conjunction—and in unexpected ways. For example, rewarding methodological rigor (instead of publication history or previous accuracy) and only giving out small grants leads to more reliable work. But when grants are large, random fluctuations in who gets them at early career stages can lead to high rates of false positives (reporting effects that aren’t true), even when reviewers select for rigor.

Based on Holiday’s insider account of media manipulation, what’s needed is an attempt to change the underlying incentives of the system. There are times when Holiday seems to understand this, as when he praises the New York Times move to force readers to pay for a subscription if they want access to more than 20 articles a month. This changes the NYT’s goal to producing 21 or more “must read” articles, he says. From a systems view, that’s precisely the kind of solution that people concerned with the media need to be considering. From his account, convincing a blogger to take more time to check facts is just convincing the blogger to get another job—and opening a slot for someone else who won’t care about fact checking. To be effective, moral entreaties can’t just be made at the individual level. That’s good news for someone as manipulative as Holiday. We don’t need to reproach individuals to solve the problem (although we can still dislike intentional manipulation). To address a systemic problem like fake news, we need to find ways to change the underlying incentives and dynamics of the system. And that’s something that will take more thought.