Steven Cherry Hi, this is Steven Cherry for Radio Spectrum.
Jonathan Swift in 1710 definitely said, “Falsehood flies, and the truth comes limping after it.” Mark Twain, on the other hand, may or may not have said, “A lie can travel halfway around the world while the truth is putting on its shoes.”
Especially in the context of politics, we lately use the term “fake news” instead of “political lies” and the problem of fake news—especially when it originates abroad—seems to be much with us these days. It’s believed by some to have had a decisive effect upon the 2016 U.S. Presidential election and fears are widespread that the same foreign adversaries are at work attempting to influence the vote in the current contest.
A report in 2018 commissioned by the U.S. Senate Intelligence Committee centered its attention on the Internet Research Agency, a shadowy arm of Russia’s intelligence services. The report offers, to quote an account of it in Wired magazine, “the most extensive look at the IRA’s attempts to divide Americans, suppress the vote, and boost then-candidate Donald Trump before and after the 2016 presidential election.”
Countless hours of research have gone into identifying and combating fake news. A recent study found more than 2000 articles about fake news published between 2017 and 2020.
Nonetheless, there’s a dearth of actual data when it comes to the magnitude, extent, and impact, of fake news.
For one thing, we get news that might be fake in various ways—from the Web, from our phones, from television—yet it’s hard to aggregate these disparate sources. Nor do we know what portion of all our news is fake news. Finally, the impact of fake news may or may not exceed its prevalence—we just don’t know.
A new study looks into these very questions. Its authors include two researchers at Microsoft who listeners of the earlier incarnation of this podcast will recognize: David Rothschild and Duncan Watts were both interviewed here back in 2012. The lead author, Jennifer Allen, was a software engineer at Facebook before becoming a researcher at Microsoft in its Computational Social Science Group and she is also a Ph.D. student at the MIT Sloan School of Management and the MIT Initiative on the Digital Economy. She’s my guest today via Skype.
Jenny, welcome to the podcast.
Jennifer Allen Thank you, Steven. Happy to be here.
Steven Cherry [[COPY]] Jenny, Wikipedia defines “fake news” as “a type of yellow journalism or propaganda that consists of deliberate misinformation or hoaxes spread via traditional print and broadcast news media or online social media.” The term made its way into the online version of the Random House dictionary in 2017 as “false news stories, often of a sensational nature, created to be widely shared or distributed for the purpose of generating revenue, or promoting or discrediting a public figure, political movement, company, etc.” Jenny, are you okay with either of these definitions? More simply, what is fake news?
Jennifer Allen Yeah. Starting off with a tough question. I think the way that we define fake news really changes whether or not we consider it to be a problem or the magnitude of the problem. So the way that we define fake news in our research—and how the academic community has defined fake news—is that it is false or misleading information masquerading as legitimate news. I think the way that, you know, we’re using fake news is really sort of this hoax news that’s masquerading as true news. And that is the definition that I’m going to be working with today.
Steven Cherry The first question you tackled in this study and here I’m quoting it: “Americans consume news online via desktop computers and increasingly mobile devices as well as on television. Yet no single source of data covers all three modes.” Was it hard to aggregate these disparate sources of data?
Jennifer Allen Yes. So that was one thing that was really cool about this paper, is that usually when people study fake news or misinformation, they do so in the context of a single platform. So there’s a lot of work that happens on Twitter, for example. And Twitter is interesting and it’s important for a lot of reasons, but it certainly does not give a representative picture of the way that people consume information today or consume news today. It might be popular among academics and journalists. But the average person is not necessarily on Twitter. And so one thing that was really cool and important, although, as you mentioned, it was difficult as well, was to combine different forms of data.
And so we looked at a panel of Nielsen TV data, as well as a desktop panel of individual Web traffic, also provided by Nielsen. And then finally, we also looked at mobile traffic with an aggregate data set provided to us by ComScore. And so we have these three different datasets that really allow us to triangulate the way that people are consuming information and give sort of a high-level view.
Steven Cherry You found a couple of interesting things. One was that most media consumption is not news-related—maybe it isn’t surprising—and there’s a big difference across age lines.
Jennifer Allen Yes, we did find that. And so—as perhaps it might not be surprising—older people consume a lot more television than younger people do. And younger people spend more time on mobile and online than older people do. However, what might be surprising to you is that no matter whether it’s old people or younger people, the vast majority of people are consuming more news on TV. And so that is a stat that surprises a lot of people, even as we look across age groups—that television is the dominant news source, even among people age 18 to 24.
Steven Cherry When somebody looked at a television-originating news piece on the web instead of actually on television, you characterized it as online. That is to say, you characterized by the consumption of the news, not its source. How did you distinguish news from non-news, especially on social media?
Jennifer Allen Yes. So there are a lot of different definitions of news here that you could use. We tried to take the widest definition possible across all of our platforms. So on television, we categorized as news anything that Nielsen categorizes as news as part of their dataset. And they are the gold standard dataset for TV consumption. And so, think Fox News, think the Today show. But then we also added things that maybe they wouldn’t consider news. So Saturday Night Live often contains news clips and touches on the topical events of the day. And so we also included that show as news. And so, again, we tried to take a really wide definition. And the same online.
And so online, we also aggregated a list of, I think, several thousand Web site that were both mainstream news and hyper-partisan news, as well as fake news. And we find hyper-partisan news and fake news using these news lists that have emerged in the large body of research that has come out of the 2016 elections / fake news phenomenon. And so there again, we tried to take the widest definition of fake news. And so not only are things like your crappy single-article site but also things like Breitbart and the Daily Wire we categorize as hyper-partisan sites.
Steven Cherry Even though we associate online consumption with young people and television with older people, you found that fake news stories were more likely to be encountered on social media and that older viewers were heavier consumers than younger ones.
Jennifer Allen Yes, we did find that. This is a typical finding within the fake news literature, which is that older people tend to be more drawn to fake news for whatever reason. And there’s been work looking at why that might be. Maybe it’s digital literacy. Maybe it’s just more interested in news generally. And it’s true that on social media, there’s more fake and hyper-partisan news than, you know, on the open web.
That being said, I would just emphasize that the dominant ... that the majority of news that is consumed even on social media—and even among older Americans—is still mainstream. And so, think your New York Times or your Washington Post instead of your Daily Wire or Breitbart.
Steven Cherry You didn’t find much in the way of fake news on television at all.
Jennifer Allen Yes. And so this is sort of a function, as I was saying before, of the way that we defined fake news. We, by definition, did not find any fake news on television, because the way the fake news has really been studied and in the literature and also talked about sort of in the mainstream media is as this phenomenon of Web sites masquerading as legitimate news outlets. That being said, I definitely believe that there is misinformation that occurs on television. You know, a recent study came out looking at who the biggest spreader of misinformation around the coronavirus was and found it to be Donald Trump. And just because we aren’t defining that content as fake news—because it’s not deceptive in the way that it is presenting itself—doesn’t mean that it is necessarily legitimate to information. It could still be misinformation, even though we do not define it as fake news.
Steven Cherry I think the same thing would end up being true about radio. I mean, there certainly seems to be a large group of voters—including, it’s believed, the core supporters of one of the presidential candidates—who are thought to get a lot of their information, including fake information from talk radio.
Jennifer Allen Yeah, talk radio is unfortunately a hole in our research; we were not able to get a good dataset looking at talk radio. And indeed, you know, talk radio. And, you know, Rush Limbaugh’s talk show, for example, can really be seen as the source of a lot of the polarization in the news and the news environment.
And there’s been work done by Yochai Benkler at the Harvard Berkman Klein Center that looks at the origins of talk radio in creating a polarized and swampy news environment.
Steven Cherry Your third finding, and maybe the most interesting or important one, is and I’m going to quote again, “fake news consumption is a negligible fraction of Americans’ daily information diet.”
Jennifer Allen Yes. So it might be a stat that surprises people. We find that fake news comprises only 0.15 percent of Americans’ daily media diet. Despite the outsized attention that fake news gets in the mainstream media and especially within the academic community: more than half of the journal articles that contain the word news are about fake news in recent years. It is actually just a small fraction of the news that people consume. And also a small fraction of the information that people consume. The vast majority of the content that people are engaging with online is not news at all. It’s YouTube music videos. It’s entertainment. It’s Netflix.
And so I think that it’s an important reminder that when we consider conversations around fake news and its potential impact, for example, on the 2016 election, that we look at this information in the context of the information ecosystem and we look at it not just in terms of the numerator and the raw amount of fake news that people are consuming, but with the denominator as well. So how much of the news that people consume is actually fake news?
Steven Cherry So fake news ends up being only one percent or even less of our overall media diet. What percentage is it of news consumption?
Jennifer Allen It occupied less than one percent of overall news consumption. So that is, including TV. Of course, when you zoom in, for example, to fake news on social media, the scale of the problem gets larger. And so maybe seven to 10 percent—and perhaps more, depending on your definition of fake news—of news that is consumed on social media could be considered what we say is hyper-partisan or fake news. But still, again, to emphasize, the majority of people on Facebook are not seeing any news at all. So, you know, over 50 percent of people on Facebook and in our data don’t click on any news articles that we can see.
Steven Cherry You found that our diet is pretty deficient in news in general. The one question that you weren’t able to answer in your study is whether fake news, albeit just a fraction of our news consumption—and certainly a tiny fraction of our media consumption—still might have an outsized impact compared with regular news.
Jennifer Allen Yeah, that’s very true. And I think here it’s important to distinguish between the primary and secondary impact of fake news. And so in terms of, you know, the primary exposure of people consuming fake news online and seeing a news article about Hillary Clinton running a pedophile ring out of a pizzeria and then changing their vote, I think we see very little data to show that that could be the case.
That being said, I think there’s a lot we don’t know about the secondary sort of impact of fake news. So what does it mean for our information diets that we now have this concept of fake news that is known to the public and can be used and weaponized?
And so, the extent to which fake news is covered and pointed to by the mainstream media as a problem also gives ammunition to people who oppose journalists, you know, mainstream media and want to erode trust in journalism and give them ammunition to attack information that they don’t agree with. And I think that is a far more dangerous and potentially negatively impactful effect of fake news and perhaps its long-lasting legacy.
The impetus behind this paper was that there’s all this conversation around fake news out of the 2016 election. There is a strong sense that was perpetuated by the mainstream media that fake news on Facebook was responsible for the election of Trump. And that people were somehow tricked into voting for him because of a fake story that they saw online. And I think the reason that we wanted to write this paper is to contradict that narrative because you might read those stories and think people are just living in an alternate fake news reality. I think that this paper really shows that that just isn’t the case.
To the extent that people are misinformed or they make voting decisions that we think are bad for democracy, it is more likely due to the mainstream media or the fact that people don’t read news at all than it is to a proliferation of fake news on social media. And you know, one thing that David [Rothschild]—and in one piece of research that David and Duncan [Watts] did prior to this study that I thought was really resonant was to say that, let’s look at the New York Times. And in the lead-up to the 2016 election, there were more stories about Hillary Clinton’s email scandal in the seven days before the 2016 election than there were about policy at all over the whole scope of the election process. And so instead of zeroing in on fake news, really push our attention to really take a hard look at the way the mainstream media operates. And also, you know, what happens in this news vacuum where people aren’t consuming any news at all.
Steven Cherry So people complain about people living inside information bubbles. What your study shows is fake news, if it’s a problem at all, is really the smallest part of the problem. A bigger part of the problem would be false news—false information that doesn’t rise to the level of fake news. And then finally, the question that you raise here of balance when it comes to the mainstream media. “Balance”—I should even say “emphasis.”
Jennifer Allen Yes. So I think, again, the extent to which people are misinformed, I think that we can look to the mainstream news. And, you know, for example, it’s overwhelming coverage of Trump and the lies that often he spreads. And I think some of the new work that we’re doing is trying to look at the mainstream media and its potential role and not reporting false news that is masquerading as true. But, you know, reporting on people who say false things without appropriately taking the steps to discredit those things and really strongly punch back against them. And so I think that is an area that is really understudied. And I would hope that researchers look at this research and sort of look at the conversation that is happening around Covid and, you know, mail in voting and the 2020 election and really take a hard look at mainstream media, you know, so-called experts or politicians making wild claims in a way that we would not consider them to be fake news, but are still very dangerous.
Steven Cherry Well, Jenny, it’s all too often true that the things we all know to be true aren’t so true. And as usual, the devil is in the details. Thanks for taking a detailed look at fake news, maybe with a better sense of it quantitatively, people can go on and get a better sense of its qualitative impact. So thank you for your work here and thanks for joining us today.
Jennifer Allen Thank you so much. Happy to be here.
We’ve been speaking with Jennifer Allan, lead author of an important new study, “Evaluating the Fake News Problem at the Scale of the Information Ecosystem.” This interview was recorded October 7, 2020. Our thanks to Raul at Gotham Podcast’s Studio for our engineering today and to Chad Crouch for our music.
For Radio Spectrum, I’m Steven Cherry.
Note: Transcripts are created for the convenience of our readers and listeners. The authoritative record of IEEE Spectrum’s audio programming is the audio version.