GLTR 2.2 MASTER PDF.PDF

GEORGETOWN LAW TECHNOLOGY REVIEW

WHY DO PEOPLE SHARE FAKE NEWS?

A SOCIOTECHNICAL MODEL OF MEDIA

EFFECTS

Alice E. Marwick

CITE AS: 2 GEO. L. TECH. REV. 474 (2018)

INTRODUCTION

In 2017, Peter Daou launched “Verrit,” a partisan news site

targeted to Democratic voters disappointed with the results of the 2016

election. The site consists of single quotations, facts, and statistics, each

formatted as a graphic and labeled with a unique “identification code” to

indicate authenticity and accuracy. For instance, a Verrit article titled

“Where Is the Outcry Over Republicans Sabotaging Health Care for

Children?” leads with a Pearl S. Buck quote, “The test of a civilization is

in the way that it cares for its helpless members,” helpfully verified with

the number 0443076, and followed by a stack of infographics, tweets, and

news articles supporting the title’s proposition. The site explains:

Each “verrit” is marked with an identification code and

contextualized with supporting material. The purpose of the

code is to confirm that the content originated at Verrit.com.

To authenticate a verrit, enter the code in the search bar. No

result = fake.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

! Assistant Professor, Department of Communication, University of North Carolina

Faculty Advisor, Media Manipulation Initiative, Data & Society Research Institute. I am

indebted to the members of the Media Manipulation Initiative at the Data & Society

Research Institute, whose research and insight has been fundamental to my own thinking

on these issues: danah boyd, Becca Lewis, Francesca Tripodi, Caroline Jack, Robyn

Kaplan, Matt Goerzen, and Joan Donovan. I would also like to thank Joshua Tucker,

Alex Leavitt, Adam Berinsky, Stacy Blasiola, and many others for conversations about

this paper at Social Science Foo Camp (2018), Reece Peck for sharing his manuscript-in-

progress about Fox News, the students of Julie Cohen’s Technology Law and Policy

Colloquium at Georgetown Law for their memos on an earlier draft of this paper, and

Francesca Tripodi for sending me an early draft of her paper “Alternative News,

Alternative Facts: Deconstructing the Realities of ‘Fake News’” (2018) which uses

ethnographic methods to take the type of active audience approach to “fake news” for

which I am advocating in this paper.!

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 475

The site met with immediate mockery upon launch, with Abby

Ohlheiser of the Washington Post describing it as “something that’s useful

for Clinton supporters who like to argue online about politics.”

Verrits are

rarely, if ever, shared in the wild, which makes it unlikely that people

would go to the trouble of creating fakes. But Verrit is built upon a series

of premises: first, that when confronted with “correct” information, people

will change their political opinions; second, that what is “correct” and

what is “incorrect” are objective truths; and third, that people share

political viewpoints online in an attempt to inform others, or at least

convince others with different opinions. All of these presumptions are

debatable.

Verrit, like Snopes, Politifact, and a host of other fact-checking

sites, reflect fundamental misunderstandings about how information

circulates online, what function political information plays in social

contexts, and how and why people change their political opinions. Fact-

checking is in many ways a response to the rapidly changing norms and

practices of journalism, news gathering, and public debate.

In other

words, fact-checking best resembles a movement for reform within

journalism, particularly in a moment when many journalists and members

of the public believe that news coverage of the 2016 election contributed

to the loss of Hillary Clinton.

However, fact-checking (and another

frequently-proposed solution, media literacy) is ineffectual in many cases

and, in other cases, may cause people to “double-down” on their incorrect

beliefs, producing a backlash effect.

This paper uses active audience approaches to media consumption

to investigate and critique the phenomenon known as “fake news.” The

term “fake news” has been used in scholarly circles for some time to

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Abby Ohlheiser, What Even is Verrit, The News Source Endorsed by Hillary Clinton?,

WASH. POST (Sept. 5, 2017), https://www.washingtonpost.com/news/the-

intersect/wp/2017/09/05/what-even-is-verrit-the-news-source-endorsed-by-hillary-

clinton/ [https://perma.cc/CS2G-JFUZ].

LUCAS GRAVES, DECIDING WHAT’S TRUE: THE RISE OF POLITICAL FACT-CHECKING IN

AMERICAN JOURNALISM (2016).

See, e,g., Philip Bump, Assessing a Clinton Argument That the Media Helped to Elect

Trump, WASH. POST (Sept. 12, 2017),

https://www.washingtonpost.com/news/politics/wp/2017/09/12/assessing-a-clinton-

argument-that-the-media-helped-to-elect-trump/ [https://perma.cc/NF65-CN4Q].

Michelle A. Amazeen, Journalistic Interventions: The Structural Factors Affecting the

Global Emergence of Fact-Checking, JOURNALISM (Sept. 8, 2017),

https://doi.org/10.1177/1464884917730217 [https://perma.cc/K22A-UNQK]; James H.

Kuklinski et al., “Just the Facts, Ma’am”: Political Facts and Public Opinion, 560

ANNALS AM. ACAD. POL. & SOC. SCI. 143, 143–154 (1998); Brendan Nyhan & Jason

Reifler, When Corrections Fail: The Persistence of Political Misperceptions, 32 POL.

BEHAV. 303, 303–330 (2010).

476 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

describe a variety of content: satirical news sites like The Onion,

manipulated photography, fabricated news items, propaganda, and press

releases, to name a few.

During the 2016 election season, “fake news”

emerged first as a way to characterize cheaply-produced sites full of

sensational information that emulated the visual conventions of online

news, but existed solely to capitalize on Americans’ interest in the election

and generate online advertising dollars.

The term expanded to include

hyper-partisan news sites like Breitbart, DailyCaller, and Occupy

Democrats, which provide ideologically-slanted but not necessarily

incorrect coverage.

It was then seized upon by then-candidate Donald

Trump to describe unflattering mainstream news coverage. Not only is the

term “fake news” both vague and value-laden (making it analytically

useless), it does not include other types of problematic information, such

as political memes, YouTube videos, and podcasts produced by far-right

extremist groups that have contributed to mainstreaming white

supremacist, xenophobic, and anti-Semitic ideas.

Regardless of what “fake news” actually means, it is typically tied

up with anxieties about the democratic ramifications of the shift from

consuming news from broadcast television and newspapers to consuming

news on social platforms. Thus, platforms like Facebook and Twitter have

been heavily criticized for their role in spreading, facilitating, and even

encouraging “fake news.”

However, today news spreads through digital

networks as only one element of a constant feed of information. Whether

people are likely to trust a story has less to do with who published it than

who shared it.

Moreover, many “fake news” or hyper-partisan stories

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Edson C. Tandoc, Jr. et al., Defining “Fake News” A Typology of Scholarly Definitions,

6 DIG. JOURNALISM 137, 5, 5–11 (2017).

Craig Silverman & Lawrence Alexander, How Teens in the Balkans Are Duping Trump

Supporters with Fake News, BUZZFEED (Nov. 3, 2016, 7:02 PM),

https://www.buzzfeed.com/craigsilverman/how-macedonia-became-a-global-hub-for-pro-

trump-misinfo [https://perma.cc/7TKF-8PWA].

Yochai Benkler et al., Study: Breitbart-Led Right-Wing Media Ecosystem Altered

Broader Media Agenda, COLUM. JOURNALISM REV. (Mar. 3, 2017),

http://www.cjr.org/analysis/breitbart-media-trump-harvard-study.php

[https://perma.cc/8RWA-MR4J].

Alice E. Marwick & Rebecca Lewis, Media Manipulation and Disinformation Online,

DATA & SOC’Y RES. INST. 107 (May 15, 2017),

https://datasociety.net/pubs/oh/DataAndSociety_MediaManipulationAndDisinformation

Online.pdf [https://perma.cc/AHD8-SPXQ].

Robyn Caplan et al., Dead Reckoning: Navigating Content Moderation After “Fake

News”, DATA & SOC’Y RES. INST. (Feb. 21, 2018),

https://datasociety.net/pubs/oh/DataAndSociety_Dead_Reckoning_2018.pdf

[https://perma.cc/G5AE-Y998].

Mary Madden et al., How Youth Navigate the News Landscape, DATA & SOC’Y RES.

INST. (Mar. 1, 2017), https://kf-site-

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 477

reinforce narratives about race, class, and gender that help build and

reinforce collective identity, especially on the right.

Sharing fake news

must be understood within this context of self-presentation and

reinforcement of group identity. This paper examines the social roles of

various types of problematic information, including “fake news,” hyper-

partisan news coverage, misinformation, and disinformation,

proposing a

sociotechnical model of media effects to understand how and why such

content spreads through social media.

Fact-checking sites and media literacy campaigns presume that

people will not share news if they know it is inaccurate, painting users as

cultural dupes at the mercy of media elites. But this is simply a newer

form of the “magic bullet” media effects model popular in the first half of

the 20

Century.

This theory conceptualized media “messages as magic

bullets capable of mesmerizing listeners who passively received and

responded to communicative stimuli in an essentially uniform manner.”

In contrast, active audience approaches require understanding how and

why people make meaning from media, viewing media use within a

particular sociocultural context.

Using a sociotechnical approach to

understand how and why people share fake news instead reveals complex

social motivations that will not be easily changed.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

production.s3.amazonaws.com/publications/pdfs/000/000/230/original/Youth_News.pdf

[https://perma.cc/P3AC-PBDN]; MEDIA INSIGHT PROJECT, ‘WHO SHARED IT?’ HOW

AMERICANS DECIDE WHAT NEWS TO TRUST ON SOCIAL MEDIA (Am. Press Inst., Mar. 20,

2017),

http://www.mediainsight.org/PDFs/Trust%20Social%20Media%20Experiments%202017

/MediaInsight_Social%20Media%20Final.pdf [https://perma.cc/5XAW-AQ8A].

Francesca Polletta & Jessica Callahan, Deep Stories, Nostalgia Narratives, and Fake

News: Storytelling in the Trump Era, 5 AM. J. CULTURAL SOC. 392 (2017).

Caroline Jack, Lexicon of Lies: Terms for Problematic Information, DATA & SOC’Y

RES. INST. (Aug. 9, 2017), https://datasociety.net/output/lexicon-of-lies/

[https://perma.cc/9QVJ-6A8Y].

SHEARON LOWERY & MELVIN L. DEFLEUR, MILESTONES IN MASS COMMUNICATION

RESEARCH: MEDIA EFFECTS (3d ed., 1995).

J. Michael Sproule, Progressive Propaganda Critics and the Magic Bullet Myth, 6

CRITICAL STUDS. MEDIA COMM. 225, 225 (1989).

HENRY JENKINS, TEXTUAL POACHERS: TELEVISION FANS AND PARTICIPATORY

CULTURE (1992); David Morley, Active Audience Theory: Pendulums and Pitfalls, 43 J.

COMM. 13 (1993); JANICE A. RADWAY, READING THE ROMANCE: WOMEN, PATRIARCHY,

AND POPULAR LITERATURE (1984).

478 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

I. DETERMINING THE EFFECT OF FAKE NEWS

A. Problematic Information Beyond Fake News

This paper uses the term “fake news” as a kind of shorthand. I do

this even knowing that the term has significant problems. Media historian

Caroline Jack, in her piece The Lexicon of Lies, argues that the term

problematic information works best for the current information ecosystem.

This is because other frequently-used terms rely on understanding the

intent of the information creator. Misinformation is unintentionally

incorrect, such as a newspaper printing an erroneous fact and subsequently

issuing a correction. Disinformation, on the other hand, is intentionally

incorrect.

Derakhshan and Wardle suggest adding the term mal-

information to describe “information, that is based on reality, used to

inflict harm on a person, organization or country.”

However, internet

content is ambivalent, as communication scholars Whitney Phillips and

Ryan Milner theorize; the networked nature of the internet and the ability

to replicate and remix images, text, and video makes it impossible to

determine where a particular idea, image, or meme originated, let alone

pinpoint the intent of the author.

This is particularly true considering the

dominance of irony as an expressive and affective force in native internet

content. Thus, using intent to distinguish between types of information is

not only extremely difficult (if not impossible), but potentially misleading.

Equally troublesome is the term propaganda, which has such a

negative valence that it is often used simply to characterize persuasive

information that the speaker does not like. The difference between

propaganda, public relations, and advertising is not a matter of scope,

scale, or information content, but of whether or not the speaker approves

of the originator.

And, as previously mentioned, given the networked

and constantly shifting nature of internet content, pinpointing an originator

may be impossible.

Finally, the term fake news is simultaneously too broad and too

narrow. It brings to mind websites architected to look like “real,”

mainstream news sources, filled with clickbait content in order to make

money from online advertising. The Oxford Internet Institute’s

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Jack, supra note 12.

Hossein Derakhshan & Claire Wardle, Information Disorder: Definitions, in

UNDERSTANDING & ADDRESSING THE DISINFORMATION ECOSYSTEM 5, 8 (2017),

https://firstdraftnews.org/wp-content/uploads/2018/03/The-Disinformation-Ecosystem-

20180207-v2.pdf [https://perma.cc/WFY4-9Z35].

WHITNEY PHILLIPS & RYAN M. MILNER, THE AMBIVALENT INTERNET: MISCHIEF,

ODDITY, AND ANTAGONISM ONLINE (2017).

Jack, supra note 12.

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 479

computational propaganda project classifies these websites as

“counterfeit” news sites and their content as “junk news.”

Many of these

sites were famously investigated by Buzzfeed reporter Craig Silverman in

the run-up to and aftermath of the 2016 election.

However, an array of

research has found that this is by no means the only way in which

problematic information spreads. For instance, First Draft defines seven

types of mis- and dis-information: satire or parody, misleading content,

imposter content, fabricated content, false connection, false context, and

manipulated content.

Using the term “fake news” ignores the fact that hoaxes, memes,

YouTube videos, conspiracy theories, and hyper-partisan news sites are

equally common ways of spreading problematic information.

For

example, Figure 1 is an image tweeted by an account called NatSocPagan

on January 24

, 2018. It shows the headshots of 33 Fox News employees

that the image labels as Jewish or married to Jews, along with a picture of

Executive Chairman Rupert Murdoch, who is labeled “Christian Zionist.”

Under the collage is the caption, “Out of the 151 News related personnel

working at Fox, 32 of them are Jewish. That is roughly 20%. If Jews only

make up 2% of American’s total population, that means they have a

1000% rate of over-representation at 21st Century Fox.” The thinly veiled

implication is that even Republican stalwart Fox News is part of the

globalist media—“evidence” for a long-standing White Supremacist

conspiracy theory that the media is controlled by Jews for the purpose of

destroying Aryan culture. This image suggests two things. First, that

politically or ideologically problematic information often spreads through

categories of information that do not fit the fake news rubric: this includes

memes; images; YouTube videos; and lengthy text files that shore up

evidence supporting some fringe theory (such as the one linked in the

image used by NatSocPagan, which includes links to biographies of Fox

employees). Second, the image suggests that even information which is

factually correct, such as the religious commitments of various Fox News

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Vidya Narayanan et al., Polarization, Partisanship and Junk News Consumption Over

Social Media in the US: COMPROP Data Memo 2018.1, COMPUTATIONAL PROPAGANDA

PROJECT (Feb. 6, 2018), http://comprop.oii.ox.ac.uk/wp-

content/uploads/sites/93/2018/02/Polarization-Partisanship-JunkNews.pdf

[https://perma.cc/TPJ5-WWF7].

Craig Silverman, This Analysis Shows How Viral Fake Election News Stories

Outperformed Real News on Facebook, BUZZFEED (Nov. 16, 2016, 5:15 PM),

https://www.buzzfeed.com/craigsilverman/viral-fake-election-news-outperformed-real-

news-on-facebook [https://perma.cc/SVJ4-QNLQ]; Silverman & Alexander, supra note

Claire Wardle, Fake News. It’s Complicated., FIRST DRAFT NEWS (Feb. 16, 2017),

https://firstdraftnews.com/fake-news-complicated/ [https://perma.cc/NT9V-SL7J].

Marwick & Lewis, supra note 8.

480 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

staffers, can be used to spread deceitful ideas, agendas, news frames, and

conspiracy theories.

Figure 1: “Fox News” Jewish meme. Tweeted by “National Socialist” @NatSocPagan on

January 24, 2018.

Therefore, throughout the document, I use Jack’s term problematic

information as a catch-all for a wide variety of false and misleading

content, and “fake news” as a heuristic given public engagement with the

term. It is important to note that the problematic information that I am

most concerned with is political. People share false stories about Beyoncé

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 481

having twin boys, or a man being sexually assaulted by a sasquatch, but

the civic and democratic ramifications of such false information is

probably negligible (I could be convinced otherwise).

For example,

Buzzfeed identified the two most-shared false news stories on Facebook in

2017 as “Babysitter transported to hospital after inserting a baby in her

vagina” and “FBI seizes over 3,000 penises during raid at morgue

employee’s home,” both from World News Daily Report.

However,

while Buzzfeed identified only eleven of the top fifty stories as political, I

would argue that many other stories that appear in their top fifty are

political, such as “17-year-old teenager sues his parents for being born

white” and “Texas church shooter was Antifa member who vowed to start

civil war” (see Table 2). Such stories and those like them involve power

relationships, structural inequality, or sensational anecdotes that illustrate

such; they shore up or argue against particular partisan identities and

positions. In other words, “fake news” stories may include the types of

problematic political positions that I find most concerning, or they may

simply be sensationalist tabloid content. In Part 2B, I analyze a sample of

“fake news” stories and a sample of problematic information to see the

similarities and inconsistencies between the two.

However, before delving into the research on fake news, I want to

examine in more depth how we can understand the effects that fake news

may have on individual users and citizens. This requires a bit of

Communication disciplinary history.

B. Models of Media Effects

The rapid industrialization and urbanization of the Northeastern

United States at the end of the 19

century caused a great deal of anxiety

about modern life. The theory of mass society held that urban life—

tenement apartments, movie theaters, and factory work—would engender

a faceless throng of atomized, alienated people particularly susceptible to

demagogues or manipulation by mass media.

This theory, influenced by

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

These are both from the Buzzfeed analysis mentioned in the next sentence. If I were to

argue that these stories have civic and political ramifications, I would say that widespread

hoaxes and falsehoods may undermine overall trust in the media. On the other hand, these

are the types of stories that tabloids like the Weekly World News trafficked in for decades.

Craig Silverman et al., These Are 50 of the Biggest Fake News Hits on Facebook in

2017, BUZZFEED (Dec. 28, 2017, 2:31 PM),

https://www.buzzfeed.com/craigsilverman/these-are-50-of-the-biggest-fake-news-hits-

on-facebook-in [https://perma.cc/TK69-3SE2].

Raymond A. Bauer & Alice H. Bauer, America, ‘Mass Society,’ and Mass Media, 16 J.

SOC. ISSUES 3 (1960); Irene Taviss Thomson, The Theory That Won’t Die: From Mass

Society to the Decline of Social Capital, 20 SOC. F. 421 (2005).

482 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

European sociologists such as Tönnies and Durkheim, held that old ties of

nation, ethnicity, or family would be broken down, replaced by whatever

was propagated by media messaging. This was especially salient in post-

World War I America.

As a result, a number of American public

intellectuals, such as Walter Lippmann and Harold Lasswell, began to

seriously investigate the effects of mass media messaging.

In Communication studies, discussions of media effects theory

typically portray propaganda scholars of the 1930s–1950s as adhering to a

so-called “magic bullet” or “hypodermic needle” theory. The magic bullet

theory presumes that people—the audience—make up an undifferentiated

mass; that media affects all people the same way; and that it has the

impact that the creators intended, whether that is to buy dish soap or

support a political party. The audience is comprised of “cultural dupes”

who are easily led by distracting entertainment.

Despite its straw-man

nature, and the fact that the theory is clearly incorrect, the magic bullet

theory still pops up both in scholarship and in popular understanding of

media.

Regardless, the magic bullet theory quickly fell out of favor among

researchers, as empirical and experiment-based frameworks failed to find

evidence for lasting or consistent media effects.

For instance, Lazarsfeld

and Merton found that in order for mass media to have a persuasive effect,

it had to be coupled with in-person, positive contact with someone holding

the same opinion.

Katz and Lazarsfeld’s Personal Influence argued that

any media message was strongly mediated by opinion leaders and peers.

A host of behavioral studies that followed indicated that media stimuli had

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

LOWERY & DEFLEUR, supra note 13.

MAX HORKHEIMER & THEODOR W. ADORNO, DIALECTIC OF ENLIGHTENMENT:

PHILOSOPHICAL FRAGMENTS (Gunzelin Schmid Noerr ed., Edmund Jephcott trans., Stan.

Press 2002) (1944). Subsequent research has shown that this is an unfair characterization

of scholars like Lasswell, who were more critical and nuanced than the magic bullet

theory might suggest. See W. Russell Neuman & Lauren Guggenheim, The Evolution of

Media Effects Theory: A Six-Stage Model of Cumulative Research, 21 COMM. THEORY

169, 169–196 (Apr. 8, 2011). It does, however, describe the moral panic around juvenile

violence in the 1950s, which was causally linked to comic books and television, see

JANET STAIGER, MEDIA RECEPTION STUDIES (2005), and subsequent panics around

violent movies and video games.

Kevin Arceneaux & Martin Johnson, More a Symptom Than a Cause: Polarization and

Partisan News Media in America, in AMERICAN GRIDLOCK: THE SOURCES, CHARACTER,

AND IMPACT OF POLITICAL POLARIZATION 309 (James A. Thurber & Antoine Yoshinaka

eds., Cambridge U. Press 2015); Lowery & DeFleur, supra note 14.

Paul F. Lazarsfeld & Robert K. Merton, Mass Communication, Popular Taste, and

Organized Social Action, MEDIA STUDS. 18 (1948).

ELIHU KATZ & PAUL F. LAZARSFELD, PERSONAL INFLUENCE: THE PART PLAYED BY

PEOPLE IN THE FLOW OF MASS COMMUNICATIONS (Transaction Publishers 2006) (1966).

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 483

negligible or temporary effects.

These perspectives suggested that, in

contrast with the magic bullet theory, that the effects of media were

minimal—a limited effects model of media.

“Limited effects” media

theories were heavily criticized by critical scholars like Todd Gitlin, who

argued that their inability to find direct causal relationships between media

viewing and outcomes ignored the “power of the media to define normal

and abnormal social and political activity, to say what is politically real

and legitimate and what is not; to justify the two-party political structure;

to establish certain political agendas for social attention and to contain,

channel and exclude others; and to shape the images of opposition

movements.”

More recently, researchers have sought to understand people’s

engagement with media. Active audience theory emerged from several

paradigms: British cultural studies;

the “uses and gratifications” tradition

of Communication studies;

and audience researchers, who increasingly

rejected the idea of the audience as an undifferentiated mass.

The new

“active audience” paradigm examined how people make meaning from

media, often incorporating sophisticated analyses of the relationship

between individual positionality, social context, and hegemonic ideologies

expressed through mass media.

For instance, Janice Radway’s Reading

the Romance used ethnography to understand women who read romance

novels, a genre often dismissed as silly or escapist.

Her participants

valued the escape that reading provided from their busy lives as

homemakers and mothers, but they also enjoyed seeing the strong,

independent-minded heroines get emotional gratification from the male

characters in the novels. Radway deftly showed how her subjects, lower-

middle class women who had not found emotional fulfillment in

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

STAIGER, supra note 28.

For a very thorough examination of media effects theories, see id.

Todd Gitlin, Media Sociology: The Dominant Paradigm, 6 THEORY & SOC’Y 205

(1978).

JOHN FISKE, TELEVISION CULTURE (Routledge, 2d ed. 2011) (1987); Stuart Hall,

Encoding and Decoding in the Television Discourse (1973), reprinted in CULTURE,

MEDIA, LANGUAGE: WORKING PAPERS IN CULTURAL STUDIES 1972–79 128 (Stuart Hall

et al. eds., 1st ed. 1980); DAVID MORLEY, THE NATIONWIDE AUDIENCE: STRUCTURE AND

DECODING (1980).

Tamar Liebes & Elihu Katz, Patterns of Involvement in Television Fiction: A

Comparative Analysis, 1 EUR. J. COMM. 151 (1986); Denis McQuail et al., The Television

Audience: A Revised Perspective, in SOCIOLOGY OF MASS COMMUNICATIONS (Denis

McQuail ed., 1972), reprinted in MEDIA STUDIES: A READER 271, 284 (Paul Marris &

Sue Thornham eds., 2d ed. 1996).

IEN ANG, DESPERATELY SEEKING THE AUDIENCE 2 (2006); Virginia Nightingale,

What’s Happening to Audience Research?, 39 MEDIA INFO. AUSTL. 18, 18–20 (1986).

Morley, supra note 15.

RADWAY, supra note 15.

484 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

heterosexual partnerships, were able to maintain their belief in traditional

roles by reading about heroines who had. At their best, such qualitative

studies explored the complex interplay between individual agency and

hegemonic power expressed in mass media; at their weakest, media texts

were portrayed as intrinsically polysemic and open to any interpretive

resistance.

In the 1990s, cultural studies scholars began to move beyond a

general “audience” model to deconstruct the popular notion of the “fan” as

brainless or pathologized, instead arguing that fans have affective

investments in particular types of media.

Many of these studies

emphasized the participatory nature of fandom. Fans did not simply

consume content: they produced their own in the forms of fan art, fan

films, and fan fiction. Henry Jenkins used participation in contrast with

the notion of mere spectatorship, in conversation with scholars like Janice

Radway, Ien Ang, and Camille Bacon-Smith. He described fans “not only

as consumers of mass-produced content, but also as a creative community

that took its raw materials from entertainment texts and remixed them as

the basis for their own creative culture.”

With the rise of the internet,

these processes extended beyond fan cultures, who were deeply networked

and productive long before computer-mediated communication. Jenkins

extrapolated his theory in the book Convergence Culture and a number of

essays on the topic, writing:

Patterns of media consumption have been profoundly

altered by a succession of new media technologies, which

enable average citizens to participate in the archiving,

annotation, appropriation, transformation and recirculation

of media content. Participatory culture refers to the new

style of consumerism that emerges in this environment.

To Jenkins and other media theorists like Larry Lessig and Clay

Shirky, participatory culture represented a shift from top-down cultural

production to “user-generated content.” The internet made it easier for

everyone, fans or not, to create their own types of media, comment on

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

CAMILLE BACON-SMITH, ENTERPRISING WOMEN: TELEVISION FANDOM AND THE

CREATION OF POPULAR MYTH (1992); Lawrence Grossberg, Is There a Fan in the

House? The Affective Sensibility of Fandom, in THE ADORING AUDIENCE: FAN CULTURE

AND POPULAR MEDIA (Lisa A. Lewis ed., 1992); JENKINS, supra note 15.

HENRY JENKINS, MIZUKOH ITO & DANAH BOYD, PARTICIPATORY CULTURE IN A

NETWORKED ERA 1 (2016).

Henry Jenkins, Quentin Tarantino’s Star Wars?: Digital Cinema, Media Convergence,

and Participatory Culture, in RETHINKING MEDIA CHANGE: THE AESTHETICS OF

TRANSITION 286 (David Thorburn and Henry Jenkins eds., 2003).

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 485

mass media, and collaborate with each other on projects like Wikipedia or

citizen journalism. Thus, the productivity of online fans was extended to

all. The only roadblocks to these egalitarian forms of participation were

grasping to maintain control in a rapidly-changing world.

In many cases, these early studies of online participation reflect a

utopian characterization of the early internet as “fandom writ large.”

These studies have generally not aged well. It is no longer the broadcast

media industries that concern most internet scholars, but data-mining

operations, government surveillance, and huge platforms like Google and

Facebook. And while in 2010, Clay Shirky was unable to come up with a

harmful form of user-generated content—instead picking the “LOLCat” as

the lowest form of online creation

—the current moment has brought into

stark relief just how harmful user-generated content can be. Regardless,

what the active audience and participatory paradigms suggest is that it is

not enough to see how many people were exposed to a fake news story or

YouTube video; we must understand what these viewers do with it.

C. Theorizing the Effects of Fake News

While the magic bullet theory has been repeatedly debunked, it

remains useful to think with, given that one simplistic but popular

narrative of fake news is that people voted for Trump in the 2016 election

because they believed a passel of conspiracy theories and falsehoods about

Hillary Clinton that were mostly spread on Facebook.

This narrative is

unsurprising; researchers have found that “magic bullet-like” language is

often used to discuss partisan media such as fake news.

Once again, we

find ourselves in a period of rapid social change, which produces great

anxiety over the effect of new forms of media on the masses. The current

conversation over fake news and disinformation struggles with the balance

between disinformation as a “magic bullet” (i.e. duping foolish Facebook

users into believing that Barack Obama was not born in the United States),

and a model that prioritizes causal effects on user activity while

disregarding the structural influence of problematic patterns in media

messaging and representation.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

HENRY JENKINS, MIZUKOH ITO & DANAH BOYD, PARTICIPATORY CULTURE IN A

NETWORKED ERA 3 (2016).

CLAY SHIRKY, COGNITIVE SURPLUS 18–19 (2010).

A related theory is radicalization via far-right media, aka the “Red Pill.” This theory is

widely believed by far-right members, particularly white supremacists.

KEVIN ARCENEAUX & MARTIN JOHNSON, CHANGING MINDS OR CHANGING

CHANNELS?: PARTISAN NEWS IN AN AGE OF CHOICE (2013).

486 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

For examples of the latter, a report from the Quello Center at

Michigan State University argued against the “filter bubble” theory of

personalized search results,

instead finding that internet users interested

in politics “search for and double check problematic political information,

and expose themselves to a variety of viewpoints.”

While the study

advocates against a technologically determinist viewpoint, its findings

could be used to avoid pursuing interventions. Similarly, Bakshy et al.

used a very large Facebook dataset to determine how often users were

exposed to political viewpoints and news stories outside their ideological

orientation. They conclude “[a]lthough partisans tend to maintain

relationships with like-minded contacts, on average more than 20% of an

individual’s Facebook friends who report an ideological affiliation are

from the opposing party, leaving substantial room for exposure to

opposing viewpoints.”

Finally, Allcott and Gentzkow determined that

the overall effect of fake news was limited, estimating that “the average

US adult might have seen perhaps one or several news stories in the

months before the election.”

This would suggest that fake news had less

impact than television advertising. Boxell et al. created an index of

political polarization in the United States, finding that the group most

likely to be polarized were adults over sixty-five, which is the

demographic group least likely to use the internet. They conclude that

“these facts can be shown to imply a limited role for the Internet and

social media in explaining the recent rise in measured political

polarization.”

Such studies suggest that fake news or problematic

information has limited effects. However, each of them operationalizes

fake news narrowly, ignores the relationship between problematic

information and mainstream media, and attempts to draw causal

relationships between viewing fake news and a particular outcome.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Pariser’s theory of the filter bubble holds that personalization algorithms will result in a

lack of exposure to ideologically diverse content. Instead, users will only be shown

content that they already agree with, or that matches their previous online activity and

interests. ELI PARISER, THE FILTER BUBBLE: HOW THE NEW PERSONALIZED WEB IS

CHANGING WHAT WE READ AND HOW WE THINK (2011).

William Dutton et al., Social Shaping of the Politics of Internet Search and

Networking: Moving Beyond Filter Bubbles, Echo Chambers, and Fake News (Quello

Ctr., Working Paper No. 2944191, 2017).

Eytan Bakshy et al., Exposure to Ideologically Diverse News and Opinion on

Facebook, 348 SCIENCE 1130 (2015).

Hunt Allcott & Matthew Gentzkow, Social Media and Fake News in the 2016 Election,

31 J. ECON. PERSP. 211, 213 (2017).

Levi Boxell et al., Greater Internet Use Is Not Associated with Faster Growth in

Political Polarization Among US Demographic Groups, 114 PROCS. NAT’L ACAD. SCIS.

10612, 10616 (2017).

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 487

Finally, Figure 1—and thousands like it—call into question many

of the studies of fake news dissemination, which must operationalize “fake

news” in order to measure it. They do this in a variety of ways, but often

researchers rely on a corpus of URLs that represent media outlets that

spread “fake news” such as DailyCaller or Breitbart. For example, one

study from the Oxford Internet Institute uses a list of ninety-one websites

that they categorize as “junk news.”

These include sites like Breitbart

and DailyCaller, but also Mediaite and the New York Daily News, which

are arguably legitimate media sources.

Allcott & Gentzhow’s study of

the spread of fake news on social media used a list of 156 stories identified

by Buzzfeed, Snopes, and Politifact as demonstrably false.

The Public

Data Lab’s Field Guide to Fake News used a list of twenty-two “fake

news” stories based on a Buzzfeed list similar to the one discussed

above.

None of these studies would attend to images such as Figure 1;

nor would they attend to, say, a YouTube “documentary” on white

genocide. Thus, many of the studies previously discussed may

underestimate the engagement that people have with problematic or

ideologically-driven information online.

While these studies are important and valuable, they are limited.

Research on online disinformation is in its infancy, and there is a need for

more sophisticated models to truly understand the effects of “fake news”

that take into account the complex interplay of people, media, and

technology.

II. A SOCIOTECHNICAL MODEL OF MEDIA EFFECTS

In order to fully understand why people share fake news, we need

to adopt a sociotechnical model of media effects with three parts (Table

1): first, that people make meaning from information based on their social

positioning, identity, discursive resources, and skill set; second, that media

messaging is often structured in particular ways to further a variety of

agendas—whether it be increasing consumption of goods, increasing time

on a website, or furthering a political viewpoint; and third, that the

material settings of media consumption (for instance, newspapers, cable

television, or social media) have particular technical affordances that

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Narayanan et al., supra note 20.

Erik Wemple, Opinion, Study Bashes Trumpites for Promoting ‘Junk’ News. But

What’s That?, WASH. POST (Feb. 7, 2018), https://www.washingtonpost.com/blogs/erik-

wemple/wp/2018/02/07/study-bashes-trumpites-for-promoting-junk-news-but-whats-that/

[https://perma.cc/239R-32X8].

Allcott & Gentzkow, supra note 50.

LILIANA BOUNEGRU ET AL., PUBLIC DATA LAB, A FIELD GUIDE TO FAKE NEWS AND

OTHER INFORMATION DISORDERS 20–22 (2017).

488 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

affect both meaning-making and messaging. In other words, people can

and do make meaning from media, but they cannot simply make any

meaning. In networked settings, this is complicated both by the presence

of connected others, and by the algorithms and advertising models that

drive social media.

The term sociotechnical comes from organizational development;

it describes a way of understanding work practice as both a set of social

systems (human behavior) and a set of technical systems (complex

infrastructures). Organizations are successful when their social and

technical systems work together smoothly.

In Science and Technology

Studies (STS), the term is used loosely to describe the social construction

of technology, or the way that human agency and technical affordances

mutually shape artifacts. As Wiebe Bijker writes, any conceptual

framework of the sociotechnical “must combine the strategies of actors

with the structures by which they are bound.”

In other words, a

sociotechnical theory of media effects must examine actors, preferably

taking an ethnographic approach to understanding cultural practice and

group identity, and media, as both patterns of messages and sets of

technological affordances which constrain or enable certain meanings and

actions. This will require multiple methodologies and interdisciplinary

thinking.

This model works with more modern effects theories such as:

agenda-setting, which holds that media coverage largely determines what

people think of as significant issues, legitimizing or de-legitimizing certain

political viewpoints;

priming, which says that the amount of coverage

media devotes to an issue makes audiences receptive to particular

themes;

and framing, which analyzes how news organizations construct

a story to further a particular point of view.

In these theories, the

messages of media are influenced by larger hegemonic ideologies and

political economies, but may be interpreted in a variety of ways by the

recipients. However, these theories were formulated in the age of

broadcast media and presume traditional journalistic practices. They also

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

THE SOCIAL ENGAGEMENT OF SOCIAL SCIENCE, VOLUME 2: THE SOCIO-TECHNICAL

PERSPECTIVE (Eric Trist, Hugh Murray & Beulah Trist eds., 1993).

WIEBE E. BIJKER, OF BICYCLES, BAKELITES, AND BULBS: TOWARD A THEORY OF

SOCIOTECHNICAL CHANGE 15 (1995).

Maxwell E. McCombs & Donald L. Shaw, The Agenda-Setting Function of Mass

Media, 36 PUB. OPINION Q. 176 (1972).

Shanto Iyengar et al., Experimental Demonstrations of the “Not-So-Minimal”

Consequences of Television News Programs, 76 AM. POL. SCI. REV. 848 (1982).

ERVING GOFFMAN, FRAME ANALYSIS: AN ESSAY ON THE ORGANIZATION OF

EXPERIENCE (1974).

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 489

do not consider the role of the individual in creating and spreading media

messaging.

Actors. First, the researcher must determine how people make

meaning from media. Drawing from and inspired by media

anthropology,

active audience theories,

ethnography of social

technologies,

and human-computer interaction,

such approaches

typically use ethnographic or qualitative methods to understand media use

in situ

—as part of daily life.

This approach requires defining specific

groups of actors (i.e. users, audiences), such as “American mainstream

conservatives” or “open-source software hackers,” observing their on- and

offline activities, understanding their social contexts, and listening to how

they describe their use of media. From active audience theory, it presumes

that actors do not simply receive media messages but decode or interpret

them based on their social position and discursive resources available to

them.

Celeste Condit argues that audiences do not often interpret texts in

radically different ways; for example, they often agree on the basic

plotline of a television show. Rather, “[i]t is not that texts routinely feature

unstable denotation but that instability of connotation requires viewers to

judge texts from their own value systems.”

In other words, the more we

know about the actors involved, the more we can understand the roles and

meanings that media takes in their lives.

Messages. Second, the model requires an understanding of the

media messages under analysis. Using methods such as content analysis,

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

FAYE D. GINSBURG ET AL., MEDIA WORLDS: ANTHROPOLOGY ON NEW TERRAIN

(2002).

Morley, supra note 15; RADWAY, supra note 15.

NANCY K. BAYM, TUNE IN, LOG ON: SOAPS, FANDOM, AND ONLINE COMMUNITY

(2000); DANAH BOYD, IT’S COMPLICATED: THE SOCIAL LIVES OF NETWORKED TEENS

(2014); WHITNEY PHILLIPS, THIS IS WHY WE CAN’T HAVE NICE THINGS: MAPPING THE

RELATIONSHIP BETWEEN ONLINE TROLLING AND MAINSTREAM CULTURE (2015).

LOUIS L. BUCCIARELLI, DESIGNING ENGINEERS (1994); LUCY A. SUCHMAN, PLANS AND

SITUATED ACTIONS: THE PROBLEM OF HUMAN-MACHINE COMMUNICATION (1987).

MARY L. GRAY, OUT IN THE COUNTRY: YOUTH, MEDIA, AND QUEER VISIBILITY IN

RURAL AMERICA 126–127 (1st ed. 2009).

DANIEL MILLER & DON SLATER, THE INTERNET: AN ETHNOGRAPHIC APPROACH 65

(2000).

Hall, supra note 35.

Celeste Michelle Condit, The Rhetorical Limits of Polysemy, 6 CRITICAL STUDS. MASS

COMM. 103, 107 (1989).

Hsiu-Fang Hsieh & Sarah E. Shannon, Three Approaches to Qualitative Content

Analysis, 15 QUALITATIVE HEALTH RES. 1277 (2005); Marilyn Domas White & Emily E.

Marsh, Content Analysis: A Flexible Methodology, 55 LIBR. TRENDS 22 (2006).

490 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

discourse analysis,

or computational analysis of social media data,

the

researcher can determine reoccurring patterns, themes, and underlying

presumptions of a particular set of media messages; for instance, “alt-right

memes,” “Donald Trump’s tweets,” or “superhero blockbuster films.”

This approach presumes that media texts are polysemous, in that they can

be interpreted multiple ways. However, this does not presume an infinite

array of meanings. As rhetorician Leah Ceccarelli advises, “[p]olysemy

indicates a bounded multiplicity, a circumscribed opening of the text in

which we acknowledge diverse but finite meanings.”

These limits or

boundaries are often determined by the nature of the platform through

which the messages are expressed. Network television, for instance,

adheres more tightly to dominant ideologies than does YouTube. Using a

qualitative or critical approach to content analysis enables an

understanding of the relationship between media messages, power, and

social control.

Affordances. Finally, the researcher must determine the features,

functionality, and—if possible—affordances of the communication

medium through which messages are expressed. An affordance is a

popular concept in technology studies; it refers to what a user perceives a

particular object can do.

Evans et al. argue that an affordance is a

possibility for action; it is determined by the user, the features of the

technology itself, and the outcome for which the technology is used.

For

example, for an American teenager, Instagram may afford posting a small

amount of highly edited and curated photos, while Snapchat may afford

sending silly photos to friends to create intimacy and a sense of

“backstage” interaction. There is nothing technical that stops the teenager

from posting the same silly photo to Instagram, but her perception of what

the technology lets her do is quite different. This is, of course, influenced

by social norms and what Ilana Gershon calls “media ideologies”—

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

DEBORAH CAMERON, WORKING WITH SPOKEN DISCOURSE (2001); Norman Fairclough,

Jane Mulderrig & Ruth Wodak, Critical Discourse Analysis, in DISCOURSE STUDIES: A

MULTIDISCIPLINARY INTRODUCTION 357 (Teun A. Van Dijk ed., 2d ed. 2011).

Seth C. Lewis et al., Content Analysis in an Era of Big Data: A Hybrid Approach to

Computational and Manual Methods, 57 J. BROAD. & ELEC. MEDIA 34 (2013); Rob

Procter et al., Reading the Riots on Twitter: Methodological Innovation for the Analysis

of Big Data, 16 INT’L J. SOC. RES. METHODOLOGY 197 (2013).

Leah Ceccarelli, Polysemy: Multiple Meanings in Rhetorical Criticism, 84 Q. J.

SPEECH 395, 398 (1998).

Sandra K. Evans et al., Explicating Affordances: A Conceptual Framework for

Understanding Affordances in Communication Research, 22 J. COMPUT.-MEDIATED

COMM. 35 (2017); James J. Gibson, The Theory of Affordances, in THE ECOLOGICAL

APPROACH TO PHYSICAL PERCEPTION 119 (Classic ed. 2015) (1986).

Evans et al., supra note 73.

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 491

people’s beliefs about the right way to use a specific type of media.

However, using affordances helps to avoid a technologically determinist

approach to communication technology, acknowledging how material

conditions affect user agency. A variety of methods can be used to analyze

affordances; for example, researchers might walk users through a

particular application

or interview users about how they understand

technology.

Such methods aim to understand use in context. Clearly, the

affordances of a newspaper or letter will be less complex than an

application like Snapchat, but in both cases, materiality matters.

The rest of this paper tests the sociotechnical model of media

effects in order to answer the question: why do people share fake news?

This study is merely preliminary; future research should investigate the

model using multi-disciplinary and multi-methodological studies.

A. Active, Conservative Audiences

To determine the actors—those who share fake news—this paper

focuses on conservative and far-right Americans. This is due to the

scholarly studies that point to a partisan asymmetry of information

consumption. The interdisciplinary research group Public Data Lab

tracked “fake news” spreading through Facebook and found that pro-

Trump and anti-Clinton pages were most likely to spread fake news, and

that the fake news stories with the most traction were anti-Clinton.

recent Guardian investigation found that anti-Clinton YouTube videos,

which often trafficked in conspiratorial thinking or outright lies, were far

more likely to be recommended than anti-Trump videos.

Faris et al.

determined that in the months up to the election, centrists and left-leaning

citizens were likely to get information from mainstream news sources like

CNN and the New York Times, while right-leaning citizens were far more

likely to consume a dense network of “hyper-partisan” sources like

Breitbart and The Daily Caller.

While these sources may not be fake

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

ILANA GERSHON, THE BREAKUP 2.0: DISCONNECTING OVER NEW MEDIA 16–50 (2010).

Ben Light, Jean Burgess & Stefanie Duguay, The Walkthrough Method: An Approach

to the Study of Apps, 20 NEW MEDIA & SOC’Y 881 (2018).

Joshua McVeigh-Schultz & Nancy K. Baym, Thinking of You: Vernacular Affordance

in the Context of the Microsocial Relationship App, Couple, 1 SOC. MEDIA & SOC’Y 1

(2015).

Bounegru et al., supra note 55.

Paul Lewis, “Fiction Is Outperforming Reality”: How YouTube’s Algorithm Distorts

Truth, GUARDIAN (Feb. 2, 2018, 7:00 AM),

http://www.theguardian.com/technology/2018/feb/02/how-youtubes-algorithm-distorts-

truth [https://perma.cc/DWR5-ZKER].

ROB FARIS ET AL., BERKMAN KLEIN CENTER, PARTISANSHIP, PROPAGANDA, AND

DISINFORMATION: ONLINE MEDIA AND THE 2016 U.S. PRESIDENTIAL ELECTION (2017),

492 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

news per se, the same study describes them as “combining

decontextualized truths, repeated falsehoods, and leaps of logic to create

fundamentally misleading view[s] of the world.”

Because these sites do

not adhere to the same norms of objectivity and nonpartisanship as

traditional journalism, they are able to amplify messaging that supports

partisan goals even if it is untrue. However, Faris and Benkler clearly

point out that it is not that Republicans are more credulous than

Democrats; it is that they are inhabiting an information system that is full

of inaccurate information.

A similar study from the Computational Propaganda project at the

Oxford Internet Institute found that Trump supporters and “Hard

Conservatives” (people espousing anti-liberal, anti-immigration, and/or

pro-militia views) were far more likely to share what the project calls

“junk news” on Facebook or Twitter than their left-wing and centrist

equivalents.

In fact, the Trump group on Twitter shared more junk news

than all the others combined.

However, this may be partially explained

by a variety of studies which find that people who are more ideologically

extreme share more on social media.

Since the conservative media

sphere is infested with disinformation, very partisan conservatives would

then be more likely than very partisan liberals to share disinformation.

Finally, my previous research on online disinformation finds that much of

it is ideologically motivated by far-right extremist groups.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

https://dash.harvard.edu/bitstream/handle/1/33759251/2017-

08_electionReport_0.pdf?sequence=9 [https://perma.cc/M5G7-NNMD].

Benkler et al., supra note 7.

“Junk news” is defined as sources which “deliberately publish misleading, deceptive or

incorrect information purporting to be real news about politics, economics or culture.

This content includes various forms of extremist, sensationalist, conspiratorial, masked

commentary, fake news and other forms of junk news. Narayanan et al., supra note 20, at

2. Sources were evaluated in terms of professionalism, style, credibility, bias, and

counterfeit nature.

Id.

Daniel Preoţiuc-Pietro et al., Beyond Binary Labels: Political Ideology Prediction of

Twitter Users, PROCEEDINGS OF THE 55TH ANNUAL MEETING OF THE ASSOCIATION FOR

COMPUTATIONAL LINGUISTICS, VOL. 1: LONG PAPERS 729 (2017); Jesse Shore et al.,

Network Structure and Patterns of Information Diversity on Twitter (2017); Magdalena

Wojcieszak, ‘Don’t Talk to Me’: Effects of Ideologically Homogeneous Online Groups

and Politically Dissimilar Offline Ties on Extremism, 12 NEW MEDIA & SOC’Y 637

(2010).

Rebecca Lewis & Alice E. Marwick, Taking the Red Pill: Ideological Motivations for

Spreading Online Disinformation, in UNDERSTANDING & ADDRESSING THE

DISINFORMATION ECOSYSTEM (2017), https://firstdraftnews.org/wp-

content/uploads/2018/03/The-Disinformation-Ecosystem-20180207-v4.pdf?x37029

[https://perma.cc/S674-8HKT]; Marwick & Lewis, supra note 8.

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 493

However, this is not to say that liberal or left-leaning individuals

do not share problematic information online. For example, a variety of

studies have suggested that very liberal individuals are more likely to

adhere to anti-vaccination beliefs.

Some conspiracy theories appeal

primarily to left-leaning individuals.

While it is clear that false

information is more prevalent on the right, more research is needed to

determine what sharing behaviors look like amongst left-wing Americans.

While ethnographic studies of communities where fake-news

sharing is prevalent are limited,

we can gain some insights into this

community as an audience by examining qualitative work on conservative

communities and media consumption. Though this paper does not hew to a

particular political ideology, I will also note that discussions of fake news

that, in an attempt to seem nonpartisan, do not identify patterns in

problematic information—or treat all fake news as interchangeable false

“content”—will be very unlikely to be able to answer “why people share

fake news.”

Francesca Tripodi’s ethnographic research on mainstream

American conservatives engages directly with questions of partisanship

and ideology. She argues that because conservatives consider the

mainstream media to be “fake,” they turn to alternative media sources

such as Breitbart and the video channel Prager U for information.

Her

participants used the term “fake news” solely to refer to mainstream media

sources like CNN, MSNBC, the New York Times, and The Washington

Post. When asked for examples of false narratives, they mentioned the

sexual assault allegations against Trump, the Trump campaign’s purported

collusion with Russia, and coverage of the Charlottesville rally which

ignored leftist agitation. In these instances, their beliefs are counter-

factual. The mainstream media is not reporting “fake news,” and the

hyper-partisan sources her participants favor are far more likely to spread

false information. Ultimately, conservatives and liberals are not only

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Charles McCoy, Anti-Vaccination Beliefs Don’t Follow the Usual Political

Polarization, CONVERSATION (Aug. 23, 2017, 8:34 PM), http://theconversation.com/anti-

vaccination-beliefs-dont-follow-the-usual-political-polarization-81001

[https://perma.cc/892X-PZAT].

McKay Coppins, How the Left Lost Its Mind, ATLANTIC (July 2, 2017),

https://www.theatlantic.com/politics/archive/2017/07/liberal-fever-swamps/530736/

[https://perma.cc/BLH2-XQJY].

For an excellent recent take on mainstream conservatives and their relationship to the

media, see FRANCESCA TRIPODI, DATA & SOC’Y RES. INST., SEARCHING FOR

ALTERNATIVE FACTS: ANALYZING SCRIPTURAL INFERENCE IN CONSERVATIVE NEWS

PRACTICES (2018), https://datasociety.net/output/searching-for-alternative-facts/

[https://perma.cc/D4H6-S6ZZ].

Francesca Tripodi, Alternative News, Alternative Facts: Deconstructing the Realities of

“Fake News", presented at the Am. Soc. Ass’n, Phila. Pa. (Aug. 13, 2018).

494 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

consuming different media, but existing in different epistemological

realities.

Another perspective on partisan news consumption is found in

Reece Peck’s outstanding book Fox News Decoded: Partisanship,

Populism and the Performance of Class, which is based on content

analysis of hundreds of hours of Fox programming and interviews with

Tea Party activists and media practitioners. In the book, Peck

painstakingly analyzes how Fox paints itself as the champion of the

underdog—white, working class, political conservatives, or the “little

guy”—by fusing conservative talking points with tabloid culture. In

contrast, while outlets like the New York Times or National Public Radio

pride themselves on being objective, they are widely viewed by

conservatives as having a left-wing bias (giving rise to the liberal bumper

sticker slogan “reality has a liberal bias”).

To explain this disparity, Peck explains how such outlets signify a

social identity through the stories they cover, the entertainment media they

highlight, and the marketing appeals they make to intellectuals,

cosmopolitan business people, and hip urbanites.

Thus, by believing

stories in the New York Times or the Washington Post, cultural

conservatives risk taking on the mantle of left-wing identity regardless of

what those stories discuss. Peck points out that partisanship is primarily

driven not by any particular party position or platform, but by affinity for

and similarity to one’s fellow party members. A great deal of research

finds that partisanship is identity-based: people identify with the party that

they feel that most members of their social group belong to, and will

adjust their party preferences to match their family, friends, or

neighborhoods.

Just as the New York Times references eating Korean

food, listening to Kendrick Lamar, and flying business class, Peck argues

that Fox News uses a set of cultural referents like country music stars in

their political rhetoric, which interpolates a particular class and race

identity: namely, white, blue-collar, and masculine—a strategy that he

calls cultural populism.

Understanding Fox News is extremely important to understanding

problematic information. Fox, along with its pundits and local affiliates,

often amplifies far-right stories that begin in fringe online communities.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Reece Peck, Fox News Decoded: Partisanship, Populism and the Performance of

Class, at 2 (n.d.) (unpublished manuscript).

POLITICAL PARTIES AND PARTISANSHIP: SOCIAL IDENTITY AND INDIVIDUAL ATTITUDES

57 (John Bartle & Paolo Bellucci eds., 2015).

See Peck, supra note 90, at 93–98.

Marwick & Lewis, supra note 8.

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 495

It contributes to partisanship, which fuels problematic information.

It has

been the most popular cable news network for fifteen straight years, and

cable news is where most Americans get their political information.

Many of the narratives and talking points that appear frequently in

problematic information work seamlessly with the metanarratives and

themes pioneered and honed by Fox.

It would be a mistake to assume that Fox News watchers are

“radicalized” or “brainwashed” by what they watch—a new form of the

magic bullet theory. Arceneaux and Johnson argue that partisan media is a

symptom of polarization, not the cause

and express frustration that

studies of media like Fox News often take a “magic bullet” approach,

assuming uniform exposure, causal effects, and passive news consumers.

Returning to Francesca Tripodi’s research, she finds that conservatives use

close-reading techniques similar to biblical interpretation to read Trump’s

speeches, proposed legislation, and transcriptions of debates.

When they

compare their own interpretations to mainstream media coverage, they

inevitably find inaccuracies, reinforcing their idea of mainstream media as

“fake.” As Tripodi writes, this approach to evaluating news articles and

sources resonates with contemporary conservative critiques of the left,

which hold that “liberal ideology is formed by disputable claims and

emotional appeals instead of fact-based evidence.”

By carefully parsing

the “facts” themselves, conservatives believe they have arrived at a

truthful interpretation of current events.

However, these interpretations are, obviously, not necessarily

correct, as the interplay between individual attitudes and partisan media

are often more complicated. In Arlie Hochschild’s Strangers In Their Own

Land: Anger and Mourning on the American Right, she tells a “deep

story” which underlies the anger and frustration felt by her informants in

southern Louisiana. Her participants felt that they were “in line” for the

benefits promised to them by the American Dream—stable jobs and

financial security, both of which were in short supply in Louisiana. But the

line did not seem fair. She writes:

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Shanto Iyengar et al., Affect, Not Ideology: A Social Identity Perspective on

Polarization, 76 PUB. OPINION Q. 405 (2012).

See generally Peck, supra note 90.

Arceneaux & Johnson, supra note 29.

ARCENEAUX & JOHNSON, supra note 46.

Francesca Tripodi, Alternative Facts, Alternative Truths, DATA & SOC’Y: POINTS (Feb.

23, 2018), https://points.datasociety.net/alternative-facts-alternative-truths-ab9d446b06c

[https://perma.cc/RBL7-KQ5W].

Id.

496 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

Black women, immigrants, refugees . . . all have cut

ahead of you in line. But it’s people like you who

have made this country great. You feel uneasy. It

has to be said: the line cutters irritate you. They are

violating rules of fairness. You resent them, and you

feel it’s right that you do. So do your friends. Fox

commentators reflect your feelings, for your deep

story is also the Fox News story.

100

The “deep story” of waiting in line for long-awaited benefits is a

metaphor that resonates with Hochschild’s informants; it is embedded in

their everyday conversations, and shapes how the Louisianan

conservatives feel. In Polletta and Callahan’s analysis of storytelling by

conservative commentators and in Trump’s campaign material, they find

plenty of evidence for Hochschild’s deep story. Like Hochschild, they are

interested in stories, or narratives, that are told both in media and

conversation. These narratives are coded and moral, reinforcing

ingroup/outgroup feelings. They are so prevalent that even if people have

no personal evidence for something, they hear stories told in conservative

outlets that enforce their own beliefs, which are also repeated in

conversation with others.

101

For instance, one frequent theme of Fox

coverage (which appears throughout Hochschild’s interviews as well) is

that urban liberals look down upon rural and “flyover” state residents and

see them as stupid, ignorant, or rednecks. During the 2016 presidential

campaign, Hillary Clinton’s famous “basket of deplorables” speech

specifically targeted the alt-right and white supremacists, but it was taken

by mainstream conservatives as attacking them. Many Trump supporters

thus wore the badge deplorable with pride. The speech supported their

deep beliefs, and it did no favors for Clinton. While the media gave the

speech a great deal of coverage, this was almost unnecessary for

conservatives to view it as insulting and condescending.

In her study of political viewpoints in Wisconsin, Katherine

Cramer argues that rural residents express a rural consciousness, which

she characterizes as a “politics of resentment.” In other words, rural

residents believe that urban governments and institutions ignore rural

concerns, deprive rural areas of needed resources, and that non-rural

people are getting more than their “fair share.”

102

Cramer’s participants

saw urbanites as lazy, lacking common sense, and undeserving of the

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

100

ARLIE RUSSELL HOCHSCHILD, STRANGERS IN THEIR OWN LAND: ANGER AND

MOURNING ON THE AMERICAN RIGHT139 (2016).

101

Polletta & Callahan, supra note 11.

102

KATHERINE J. CRAMER, THE POLITICS OF RESENTMENT (2016).

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 497

many economic advantages they enjoyed as residents of Madison or

Milwaukee. Not all of the people Cramer interviewed were Democrats,

and she makes it clear that this rural consciousness is not simply racism

tied up in another name—the rural Wisconsinites were very dismissive of

elected officials and wealthy tourists, both predominantly white. But the

idea of resentment, feeling like underdogs, is what is important here and in

Hochschild’s work. Mainstream media represents the out-of-touch, lazy

urbanites trying to cut ahead of the line; conservative media attempts to

affect an “authenticity” which is tied to non-elite, non-urban identity.

103

Thus, according to previous research, successful conservative news

stories often tap into “deep stories” that are common in other forms of

conservative storytelling, such as:

• Conservative values are under attack from liberals (who may

manifest themselves as immigrants, trans people, feminists,

Black Lives Matter activists, Antifa, or any number of other

boogeymen);

• Liberal urbanites look down upon rural conservatives; and

• The mainstream media are left-wing elites who wish to destroy

traditional values, are corrupt and greedy, and, in the most

extreme narratives, are controlled by Jews. This implies that

mainstream media stories cannot be trusted; since they are

usually furthering a narrative completely at odds with whatever

ideology or position the problematic information is taking up,

this makes the problematic information itself seem more

truthful.

Partisan news also produces affect:

• It reflects a cultural identity of blue-collar whiteness, or rejects

liberal urban identity markers;

• It gives a sense of urgency;

• It not only creates in-group solidarity, but reinforces out-group

animus; and

• It expresses resentment towards the undeserving.

In the next section, I analyze the messages of fake news stories to

see if they conform to these patterns identified in mainstream conservative

media.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

103

See generally Peck, supra note 90.

498 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

B. Messages: Deep Stories

To examine the messages of fake news, I conducted two brief

exercises in content analysis. In the first, I used a sample of the top 100

fake news stories on Facebook as determined by Buzzfeed (fifty for 2017

and fifty for 2016). These are strictly false stories posted by websites pre-

determined by Buzzfeed to include fake news, such as yournewswire.com

and tmzhiphop.com. In other words, they adhere to the model of “fake

news” discussed in “Problematic Information Beyond Fake News.” For a

second corpus, I used the “Hot 50” on Snopes.com for March 15, 2018.

104

Snopes.com is a popular fact-checking site that determines the accuracy of

popular email forwards, viral videos, trending social media topics, and

memes in addition to “fake news” stories, so it included a broader range of

problematic information. Rather than providing headlines, Snopes

provides “claims,” since the assertions they analyze are often found in

multiple pieces of media.

I categorized each story, read the original source when available

(many “fake news” stories from 2016 had disappeared, and many of the

Snopes stories referred to memes, tweets, or videos), and coded it as

political or not, using a very broad definition of “political” as pertaining to

systemic power relationships. For stories I determined to be political, I

then coded the political leaning of each story. Political leaning was

determined by examining the underlying themes and messages of each

story, whether they mentioned a politician in favorable or unfavorable

terms, and whether they included “hot-button” issues such as immigration,

Islam, or Black Lives Matter. After coding all the items, I went back and

re-coded them, then resolved any inconsistencies between the two samples

(e.g. I had coded an anti-Hillary conspiracy as “conspiracy” in the first

sample and as “political” in the second—I decided to code both of them as

“political”).

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

104

I am not certain how Snopes defines the “hot” in their “Hot 50” (it is not the “most

searched” or “most shared,” as those have their own pages. However, those categories

indicate the most searched or most shared stories at snopes.com, meaning they were

searched or shared by people who had already visited a fact-checking site, presumably to

determine the veracity of a claim. This is not the same as popular false stories). On their

“Transparency: Topic Selection” page, they write: “The inputs we use for the process of

determining reader interest include the tabulation of terms entered into our search engine,

reader e-mail submissions, comments and items posted to our Twitter and Facebook

accounts, external social media posts, Google Trends, Twitter’s Trending Now,

Facebook’s Trending Topics, and items flagged for review by Facebook users as part of

our partnership with Facebook.” Transparency: Topic Selection, SNOPES (Feb. 28, 2017),

https://www.snopes.com/topic-selection/ [https://perma.cc/4SMZ-S6ER]. Therefore, the

“Hot 50” seems like a good starting point to classify problematic information with a

wider scope than the Buzzfeed lists.

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 499

I began with a very simple coding scheme: None, Left, or Right.

• None: Stories which had no political leaning. They might be

outrageous, sensational, or grotesque, but did not mention partisan

issues.

• Right: Stories that were anti-Obama, anti-Clinton, pro-Trump, or

adhered to right-wing positions on partisan issues such as gun

control, police brutality, and immigration.

• Left: Stories that were anti-Trump, pro-Obama, pro-Clinton, or

adhered to left-wing positions on partisan issues such as gun

control, police brutality, and immigration.

105

I had two hypotheses:

• H1: “Fake news” stories would primarily be right-leaning.

• H2: Right-leaning “fake news” stories would adhere to the deep

stories and affective messaging discussed in the previous section.

1. Buzzfeed Stories

The Buzzfeed stories were surprisingly difficult to code. I was

forced to abandon my simple coding in favor of six categories: none,

undetermined, left, right, conspiracy, or racist.

Undetermined: I used this code if the story had a political valance, but it

was difficult to determine. A story like “Pence: Michelle Obama Is the

Most Vulgar First Lady We’ve Ever Had,” could be interpreted as anti-

Pence (and therefore left-leaning) or anti-Michelle Obama (and therefore

right-leaning).

Racist: Stories that used racist African American stereotypes, but

otherwise did not include any partisan issues.

106

Racism is clearly

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

105

Several stories in both samples dealt with animal welfare and hunting. I spent quite a

bit of time deliberating over this but ultimately coded the animal welfare stories as non-

partisan and the hunting stories as left-wing.

106

There were no racist stories about any other ethnic group, although there were three

Islamophobic and one anti-immigrant story, which I coded as right-leaning since they

played on other partisan issues like voter fraud and Hillary Clinton’s supposed affiliation

with ISIS.

500 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

political, but not something that neatly maps on to partisanship in terms of

who might share such a story.

Conspiracy theories: Stories that supported nonpartisan conspiracy

theories (such as Princess Diana’s death). Partisan conspiracy theories,

such as Pizzagate,

107

were coded as such.

The results of my coding can be seen in Table 2. Most of the

stories had no political leaning (forty-one). Twenty-one leaned right,

twelve I considered undetermined, eleven leaned left, seven focused on

conspiracy theories, and four were racist. Seven dealt with non-US issues

(Mexico, Argentina, and the Philippines) so were not analyzed for content.

2. Snopes Claims

Unlike the Buzzfeed corpus, not all the Snopes stories were false—

ten were “mixed,” three were categorized by Snopes as “out of date”

(accurate information that circulates sometimes years after it happened),

six were true, and thirty-one were false. The political leanings were also

fairly easy to determine—stories were either non-political (“300,000

pounds of rat meat disguised as chicken wings were sold in the U.S.”) or

highly partisan (“Donald Trump's IQ, at 156, is comparable to that of the

smartest U.S. presidents.”). There were no racist stories

108

and many fewer

ambiguous claims. Only one story was coded as “undetermined”—a

sexual harassment claim against the recently deceased Stephen Hawking.

While this might convey a sentiment of feminist support for the #metoo

movement, it is not clear from the original story.

The political leanings of the Hot 50 can be seen in Table 3.

Twenty-five had no political leaning, eighteen leaned right-wing, seven

stories were left-wing, and one was undetermined. Notably, all but one of

the right-leaning stories were false (thirteen false, three mixed, and one

out of date), whereas three of the left-wing stories were true (two false and

two mixed).

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

107

A far-right conspiracy theory holding that prominent Democratic politicians, including

Hillary Clinton, were running a child trafficking ring for pedophiles out of the basement

of Comet Ping Pong pizza in Washington, DC.

108

This may be due to Snopes editorial policy, which does not cover content they think is

inappropriate for the site.

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 501

3. Findings

My first hypothesis—that problematic information would primarily

reflect right-wing sensibilities—is false. Most problematic information

does not involve political issues. Significantly more false stories lean right

than left, although there is still a plethora of false content that appeals to

people with left-wing sensibilities.

My second hypothesis—that right-wing problematic information

would reflect the deep stories and affective messaging discussed in the

previous section—was supported. Both the right-wing “fake news” stories

and problematic content listed on Snopes mapped fairly neatly to the

“deep stories” outlined in the previous section. Thus, my first significant

finding is that problematic partisan information exists on a continuum

with mainstream partisan media. Some stories engaged with long-term

conservative talking points involving Obama’s birth certificate and ties to

ISIS, the dangers of Muslim refugees, celebrities and liberals getting their

comeuppance, and the like, which reinforce themes that are already

omnipresent on Fox News. Others made sensational claims about Hillary

Clinton’s criminality and the violence of Black Lives Matter members.

This suggests that the difference between problematic partisan information

and mainstream partisan information is simply a matter of scale, rather

than a clear line between “true” and “false.” Moreover, some of the stories

played on themes emphasized by extremist groups. For instance, the

headline "Police Find 19 White Female Bodies in Freezers With ‘Black

Lives Matter’ Carved into Skin" is similar to false stories spread by white

supremacist groups to brand Black Lives Matter a terrorist organization.

Indeed, my research with Becca Lewis is concerned with how far-

right extremist groups use social media for two purposes: first, to recruit

more adherents to their points of view, and second, to get media coverage

of their news frames, ideas, slogans, and so forth, thus “opening the

Overton window” (the range of political viewpoints that are socially

acceptable in American society) and furthering their ideological aims.

109

Thus, problematic information is often more extreme than the political

views voiced on Fox News and may include outright racism, misogyny,

Islamophobia, and anti-Semitism, which is signaled through dog-whistle

phrases and coded language in mainstream media. Ideologues may adopt

some of these affective and framing strategies to appeal to conservatives

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

109

Lewis & Marwick, supra note 85; Marwick & Lewis, supra note 8.

502 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

and further their own aims, even if these do not map at all to mainstream

conservativism.

110

The second significant finding is that fake news and problematic

content appeals to different people for different reasons. “Fake news”

content is clickbait. The goal of the fake news producer is to have as many

people spread their content as possible. The easiest way to do that is to

find a news item that will be shared by people of different political

proclivities. This can be a sensational claim about vaccinations or

conspiracies or animals—topics that appeal across party lines—or it can

be a story that includes both conservative and liberal points of view. A

story like “White Baseball Players Kneel in the 50’s to Protest Black

Lynchings,” could be interpreted in support of NFL player Colin

Kaepernick’s position on Black Lives Matter, or it could be a refutation of

the history of White racism. It was impossible for me to determine the

intent of these stories without seeing the context in which they were

shared. The latter stories are polysemous, which is key to their appeal.

This polysemy potentially increased the audience for the stories, in that

people with many different political leanings might be motivated to share

them.

111

Precisely because of the ambiguity, I had a difficult time coding

the “fake news” stories.

On the other hand, only two of the stories in the Snopes sample

also appeared in the Buzzfeed sample. Most of the Snopes stories would

not be classified as “fake news” by any algorithm, but they are still widely

shared, so this analysis shows the utility of examining problematic

information beyond fake news. Moreover, some of the Snopes stories were

true, while all the “fake news” stories were false. Again, this points to a

continuum of “truthiness” rather than a clear bright line between “fake”

and “real” news.

Obviously, this quick exercise is not definitive; it is merely a

starting point for future research, which might use larger samples from a

more diverse set of sources, use multiple coders to increase reliability, or

interview people who shared news items to determine intent. Nonetheless,

in the context of this paper, it is a good demonstration of how messaging

research might be combined with other methodologies in a sociotechnical

approach to determining media effects.

While we may understand the content of problematic information a

bit better and understand both why it resonates with conservatives and

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

110

This is a bit complicated, since Trump does draw frequently from fairly extreme

online content and holds more extreme policy positions than many mainstream

Republicans.

111

John Fiske, Television: Polysemy and Popularity, 3 CRITICAL STUD. IN MASS COMM.

391 (1986), https://doi.org/10.1080/15295038609366672 [https://perma.cc/7U3Y-95FE].

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 503

how it is coherent with more mainstream conservative media, we must

look not only at problematic information itself, but the technical context in

which it is spread in order to fully understand this problem.

C. Affordances: The Impact of Social Sharing

Today, most people get at least some of their news from social

media, whether that be Facebook, Twitter, Snapchat, YouTube, Reddit,

newsletters, blogs, or podcasts. Social media has several significant

differences from traditional media:

1. Anyone can produce and distribute content

2. Content is shared through social networks and in social

contexts

3. Social media platforms promote content algorithmically, based

on complex judgments of what they think will keep you on the

platform

First, anyone can produce and distribute content, rather than just

professional journalists or filmmakers. This means that people holding

viewpoints outside the Overton Window—who would never appear on

mainstream media—can and do produce their own news stories,

tweetstorms, and blogs. This has produced a huge boom in news targeted

to a wide range of political commitments: there are feminist news sites,

anarchist Facebook pages, socialist newsletters, libertarian podcasts, white

supremacist memes, even fan sites for Stalin and blogs written by people

who advocate returning to monarchic feudalism.

112

It has also allowed for

a variety of activist groups to organize around issues that are often absent

from mainstream media. The Black Lives Matter movement, for example,

used hashtag activism to increase coverage of police brutality and bring

attention to biased coverage of black victims. It has also enabled the

creation of information that is false, biased, or both, which is often

indistinguishable from that which is professionally produced. The cost of

and skill involved in creating a YouTube video or Medium post continues

to decline. While most internet users do not post YouTube videos or

political blog posts (although many do), a huge number take part in lower-

overhead online activities, such as liking a Facebook post, reblogging,

retweeting, or commenting on a news story, or wading into a discussion

war on someone else’s Facebook or Instagram account. Others simply

listen, scrolling through a feed or reading a story.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

112

“Tankies” and “neo-reactionaries,” respectively.

504 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

Secondly, most people do not go to individual newspapers or blogs

to consume news. Instead, they consume news as part of their social feed.

This means that political material is served up as simply one ingredient in

a bouillabaisse of photographs, personal stories, advertisements, movie

trailers, celebrity gossip, sports news, or whatever else appears in

someone’s Facebook feed, Snapchat stories, or subscribed subreddits.

Alfred Hermida refers to this constant flow of information as ambient

journalism.

113

Zizi Papacharissi notes that this stream is affective; among

social media participants, news is “collaboratively constructed out of

subjective experience, opinion, and emotion, all sustained by and

sustaining ambient news environments.”

114

In social spaces, the traditional

journalistic value of objectivity no longer makes sense: virtually every

story is augmented with someone’s opinion. Hermida writes, “Professional

publications go to great lengths to distinguish the spaces for commentary

and opinion from those for the news. Social media platforms break down

such boundaries, with facts and fiction, and observations and opinions, in

the mix.”

115

Scholars and journalists are just beginning to understand the

myriad impacts of social sharing. The Media Insight Project found that

when people see a post from a person they trust, they are more likely to

recommend the news source to friends, follow the source on social media,

and sign up for news alerts from the source.

116

Similarly, a Knight

Foundation study found that “[n]ews-seekers depend on friends, contacts

and individuals followed as trusted news sources as much as or more than

they depend on the media outlets themselves.”

117

This may have

something to do with the greatly diminished trust in traditional journalism;

when mistrust of news is widespread, people look for other markers of

trustworthiness. Notably, Republicans are far less likely to trust news than

Democrats, suggesting that the “deep story” of media bias discussed in the

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

113

Alfred Hermida, Twittering the News: The Emergence of Ambient Journalism, 4

JOURNALISM PRAC. 1751 (2010), https://doi.org/10.1080/17512781003640703

[https://perma.cc/4NCQ-5BPN].

114

Zizi Papacharissi, Toward New Journalism(s): Affective News, Hybridity, and Liminal

Spaces, 16 JOURNALISM STUD. 27 (2015).

115

Alfred Hermida, Social Media and Journalism, in THE SAGE HANDBOOK OF SOCIAL

MEDIA (Jean Burgess, Alice E. Marwick, & Thomas Poell, eds., SAGE Publications Ltd.

2017).

116

MEDIA INSIGHT PROJECT, supra note 10.

117

Mobile-First News: How People Use Smartphones to Access Information, KNIGHT

FOUND. (May 11, 2016), https://knightfoundation.org/reports/mobile-first-news-how-

people-use-smartphones-acces [https://perma.cc/YW9Y-8VN5].

2018 GEORGETOWN LAW TECHNOLOGY REVIEW 505

previous section is widespread.

118

But most importantly, within social

environments, people are not necessarily looking to inform others: they

share stories (and pictures, and videos) to express themselves and

broadcast their identity, affiliations, values, and norms.

At a recent workshop on partisan media, my friend “Carly” related

her frustration. Her mother, a strong conservative, repeatedly shared “fake

news” on Facebook. Each time, Carly would send her mother news stories

and Snopes links refuting the story, but her mother persisted. Eventually,

fed up with her daughter’s efforts, her mother yelled, “I don’t care if it’s

false, I care that I hate Hillary Clinton, and I want everyone to know that!”

This anecdote encapsulates how problematic political information

functions online as an identity-signaling mechanism. When someone

chooses to share a fake news story on Facebook, Twitter, via text message,

or on Whatsapp; when they post a conservative meme to their wall; or

when they “like” a YouTube video about a pro-Trump conspiracy theory,

they may well be doing it to signal their identity and affiliate themselves

with like-minded others. Right-leaning citizens are more likely than left-

leaning citizens to do this because the right-wing mediasphere has more

problematic information than that of the left; left-leaning citizens are more

likely to share mainstream media articles, although they are by no means

immune to problematic information. Regardless, what is important is not

the accuracy of the information shared, but the identity that it signals.

Sharing a New York Times story signals something different than a Mother

Jones story than an US Weekly story than a Wall Street Journal story than

a Breitbart story. Media institutions are affiliated with partisan identities,

which map to particular configurations of race and class. People want to

share stories that express their feelings and identities.

Finally, platforms sort or recommend content based on complex

algorithms which serve different videos, images, or stories based on what

they think will keep you on the platform. Just as television programs only

had to be good enough to keep you from changing a channel, social media

content does not have to be good; in many cases, a sensational or even

grotesque result will be more likely to engage viewers than one that is

more thoughtful, but perhaps less lurid. The artist James Bridle wrote a

passionate Medium post critiquing YouTube videos made for children that

contained disturbing imagery of pregnancy, vaccinations, toothache,

children in distress, or beloved characters fighting or hurting each other.

119

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

118

10 Reasons Why Americans Don’t Trust the Media, KNIGHT FOUND. (Jan. 16, 2018),

https://medium.com/trust-media-and-democracy/10-reasons-why-americans-dont-trust-

the-media-d0630c125b9e [https://perma.cc/EQ8E-629B].

119

James Bridle, Something Is Wrong on the Internet, MEDIUM (Nov. 6, 2017),

https://medium.com/@jamesbridle/something-is-wrong-on-the-internet-c39c471271d2

[https://perma.cc/VS33-LQ4Y].

506 GEORGETOWN LAW TECHNOLOGY REVIEW Vol 2.2

His article set off a wave of controversy and forced YouTube to make

sweeping changes to how children’s videos were monetized. Many of

these videos were probably churned out by small video studios using pre-

existing character models and automated software simply for advertising

revenue, but in other cases they were made by independent video creators

who refined their productions based on what they believed the algorithm

prioritized, such as the “Toy Freaks” channel on YouTube, made by a

single dad who featured his two daughters in various upsetting

situations.

120

Since disturbing videos got more traction on YouTube and

thus made more money, more were produced, and more were

recommended to viewers. Similarly, the Guardian study of YouTube

discussed previously found that in many cases, searching for a term like

“dinosaurs,” “aliens” or even “Hillary Clinton” brought up a series of

Recommended Videos based on conspiracy theories or outright nonsense

like Flat Earth theory.

121

Thus, it is quite possible that problematic

information is prioritized on social media sites because it garners more

engagement, even if that is to dispute it or make fun of it. Additionally,

algorithms often serve content based on a user’s social networks and

perceived interests. In an episode of the Gimlet podcast Reply All, a caller

(“Charles”) told a story about noticing an uptick in white supremacist

content in his Facebook feed—something that was antithetical to his own

political positions. He found out later that his brother-in-law had briefly

attended White Pride meetups in his hometown and connected to some

likeminded people via Facebook. Facebook used this information to serve