One of the most comprehensive studies on network propaganda and the 2016 U.S. election

Network analysis from a team of Harvard researchers found the media ecosystem of the right is the primary driver of manipulation, disinformation, and radicalization in American politics.

What is this report?

This report summarizes key findings and recommendations from the book Network Propaganda: Manipulation, Disinformation, and Radicalization in American Politics, which is considered to be one of the most comprehensive empirical studies on the subject of network propaganda regarding the 2016 U.S. presidential election. This study was led by Yochai Benkler (Harvard Law School professor, Director at Berkman Klein Center for Internet & Society at Harvard), Robert Faris (Research Director at BKC), and Hal Roberts (BKC fellow).

Yochai Benkler and his team of researchers argue with empirical findings that the cause of the U.S. network propaganda crisis is institutional, not technological, and focused on the U.S. media ecosystem dynamics. The authors found that the main driver of the current crisis takes on a partisan bent – asymmetric political polarization. This refers to the difference, or asymmetry, between the right’s media ecosystem and the rest of the political landscape (center and left) in their respective media network architectures and attitudes towards professional journalism. Their data support lines of research on polarization in American politics that focus on the asymmetric patterns between the left and the right, rather than studies that see polarization as a general historical phenomenon, driven by technology or other mechanisms that apply across the partisan divide. They found insufficient evidence to conclude that other real elements, such as bots, commercial advertising platforms (i.e., Facebook), or Russia had a significant effect in propagating false information, despite providing a propaganda pipeline.

Looking at over a million articles from the open web and social media activities (Facebook, Twitter) as well as coverages on cable TV, the researchers began by identifying a list of commonly identified threats, mapping out the political media ecosystem using network analysis, and discovering that the ecosystem looks very different on the right than the rest of the media landscape. The book covers in detail the hypotheses, approaches, and statistical findings that make up their central argument, but this report focuses specifically on the empirical takeaways without diving into the statistical approaches and pieces of evidence.

How did they study the media landscape?

  • Cross-linking patterns between media sources offer a view of authority and prominence within the media world.
  • The sharing of media sources by users on Twitter and Facebook provides a broader perspective on the role and influence of media sources among people engaged in politics through Twitter and Facebook.
  • The differential media sharing patterns of Trump and Clinton supporters on Twitter enable a detailed analysis of the role of partisanship in the formation and function of media structures.
  • Content analysis using automated tools supports the tracking of topics over time among media sources.
  • Qualitative media analysis of individual case studies enhances our understanding of media function and structure

What did they find?

The narrative sharing practices of the network follow fundamentally different pathways on the right vs. the rest of the media ecosystem, and this asymmetry is what made Republican voters and politicians particularly susceptible to misperception and manipulation. The researchers observed the following as drivers of asymmetric political polarization:

  • Asymmetric architecture of news media: The authors found that there is no left-right division in the media landscape across platforms, only a division between the right and the rest of the media ecosystem. Dynamics on the right tend to reinforce partisan statements, irrespective of their truth, and to punish actors who insist on speaking truths that are inconsistent with partisan frames and narratives dominant within the ecosystem.
  • Asymmetric attitudes toward professional journalism governed by norms of objectivity: Because of the architecture, the right is insular from the corrective function of professional journalism observed in the other ecosystem. The right-wing reporters who played prominent roles in propagating false stories paid no professional price because the network they work for is a propagandist, not journalistic, enterprise. Whereas the mainstream media ecosystem exhibited repeated efforts to hold themselves to high journalistic standards (removal of content, rapid corrections, and displicining reporters).

Asymmetric network architecture

Using network analysis of articles shared or referenced on the open web, twitter, and Facebook, the authors found that the structure and composition of media on the right and left are quite different. The center of gravity of the overall media landscape is the center-left, and partisan media sources on the left are less important than the center-left (see graph below). On the other hand, the center of attention and influence for conservative media is on the far right, notably more prominent than the center-right, which is of minor importance and least represented.

The Polarization of the Media Landscape

[Network map based on Twitter media sharing from May 2015 to Nov. 2016] Breitbart emerges as the nexus of conservative media. The Wall Street Journal is treated by social media users as centrist and less influential. The rising prominence of Breitbart along with relatively new outlets such as the Daily Caller marks a significant reshaping of the conservative media landscape over the past several years. Source: Partisanship, Propaganda, and Disinformation: Online Media and the 2016 U.S. Presidential Election

Partisan news is more popular on the right than on the left.


[Partisan distribution of top 250 media sites by Facebook shares] The structure of the overall media landscape shows media systems on the left and right operate differently. The asymmetric polarization of media is evident in both open web linking and social media sharing measures. Prominent media on the left are well distributed across the center, center-left, and left. On the right, prominent media are highly partisan. Source: Partisanship, Propaganda, and Disinformation: Online Media and the 2016 U.S. Presidential Election

The leading media on the right and left are also rooted in different traditions and journalistic practices. On the conservative side, more attention was paid to pro-Trump, highly partisan media outlets. On the liberal side, by contrast, the center of gravity was made up largely of long-standing media organizations steeped in the traditions and practices of objective journalism. However, these traditions and practices were also exploited in search of ‘neutrality’ over truth-seeking described next in asymmetric journalistic attitudes.

Asymmetric journalistic attitudes

Propaganda feedback loops on the right: When observing right-wing conspiracy theories, we saw positive feedback loops between the core of that network (Fox News, leading pundits, and Breitbart) and the remainder of the online right-wing network. In those cases we saw repetition, amplification, and circling of the wagons to criticize other media outlets when these exposed the errors and failures of the story. By contrast, the mainstream media ecosystem exhibited intensive competition to hold each other to high journalistic standards, and a repeated pattern of rapid removal of content, correction, and in several cases disciplining of the reporters involved.

  • Fox News actively used its position at the core of the right-wing media ecosystem to support the president in the central political controversy of his presidency: the investigation by Robert Mueller. The data warrant the conclusion that Fox shares little but a few visual trappings with the world of professional journalism at the core of the rest of the U.S. media system. It is, across its online and television properties, America’s leading propaganda outlet.
  • Breitbart played a large role by not only producing its own immigration stories but also acting as a source of stories and authority for other sites on the right and as the center of attention among social media users who shared content from right-wing sites on Twitter or Facebook.)

The left’s search for publicly performing neutrality (and scoops) becomes a vulnerability that right wing propagandists can and do exploit. The internal dynamics of news reporting led mainstream media to emphasize the email investigation over substantive discussion of politics.  If there is fault in the incessant coverage of emails, it is a fault in patterns of compliance with professional media norms of ‘even-handed reporting’, not in their violation. The mainstream media sought to provide balanced coverage to a particularly negative candidate by providing uniformly negative coverage regardless of substance.

What are the possible solutions?

The authors suggest interventions be proposed based on local institutional conditions, rather than context-free explanations based on the nature of the technology.  Their central argument has been that the present source of information disorder in American political communications is the profound asymmetry between propaganda feedback loop that typifies the right-wing media ecosystem and the reality-check dynamic that typifies the rest of the media system. The most important attainable change in the face of that asymmetry would be to reform the practice of professional journalism.

Changes to professional journalism practices

Benkler, Faris, and Roberts argue that the existing approach to neutrality adopted by the professional media is to provide even coverage. However, by maintaining the “one side says x, the other side says y” model of objectivity in the presence of highly asymmetric propaganda efforts, mainstream media become sources of legitimation and amplification for the propagandists. The authors recommend a shift in how journalists practice objectivity, from ‘demonstrative neutrality’, to a concept called accountable verifiability, to counteract some of the reinforcement and legitimation that the present practice creates on the background of highly asymmetric propaganda practices.

Accountable verifiability refers to moving away from engaging in a public performance of neutrality through balanced coverage and moving closer to emphasizing the transparency and accountability of journalists’ sources and practices. Some of this is simply emphasis and extension of existing practices – (e.g., providing public access to underlying documentary materials). They recommend that this could perhaps be done in a network of media outlets, as it is in peer review in science, by designating a set of independent nonprofit fact-checking organizations that will operate as quality assurance for journalistic enterprises and be provided with more access to underlying materials than under current practice.

Addressing challenges posed by new technologies

Benkler, Faris, and Roberts reviewed a series of proposed solutions to address the role technology platforms play (e.g., fact-checking, media literacy improvement, etc.), and considered it directionally helpful but lacks empirical evidence behind proposed benefits. However, they point to The Honest Ads Act, introduced by Senators Amy Klobuchar, Mark Warner, and John McCain, to be helpful in combating network propaganda to a certain extent. The Act seeks to separate paid vs. unpaid internet communications, require disclaimers on online advertising, and create a fine-grained public database of online political issue advertising. The authors consider the third point, the creation of a public political advertising database, to be the most important of the bill’s provisions because it should allow public watchdog organizations to offer accountability for lies and manipulation in almost real time. Benkler and team did note that the proposed solutions today still will not address propaganda (e.g., a covert foreign information operation will not follow new laws, just as it did not follow old ones), but it would ‘certainly bring some of the most effective manipulation tactics into the sunlight.

While technology was not believed to be the main driver of the current discord in American media ecosystem, the authors still believe that it needs to be regulated to an extent. From the perspective of disinformation and misinformation, the trouble with concentrated platforms is that if and when they fail, or become bad actors or beholden to bad actors, the negative effect is enormous. The danger presented by a single company having massive influence over large portions of the population is reason enough to focus either on ensuring greater competition that will diffuse that power or on tightly regulating how that company can use its market power. Nonetheless, they caution against overly restrictive efforts like the one Germany has enacted in 2018, as it incentivizes over-censorship and runs the risk of legitimizing efforts of more restrictive regimes to incorporate more speech-repressive criminal provisions.

Additional resources / coverage: