Jump to ContentJump to Main Navigation
Network PropagandaManipulation, Disinformation, and Radicalization in American Politics$

Yochai Benkler, Robert Faris, and Hal Roberts

Print publication date: 2018

Print ISBN-13: 9780190923624

Published to Oxford Scholarship Online: October 2018

DOI: 10.1093/oso/9780190923624.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2021. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use. date: 07 December 2021

Can the Internet Survive Democracy?

Can the Internet Survive Democracy?

(p.341) 12 Can the Internet Survive Democracy?
Network Propaganda

Yochai Benkler

Robert Faris

Hal Roberts

Oxford University Press

Abstract and Keywords

This chapter examines whether the internet can—or cannot—contribute to democratization, and under what conditions. This chapter discusses five major failure modes that limit the benefits of decentralized digitally-mediated collective action. The first is the failure to convert from a moment’s surge of decentralized passion into a longer-term, sustained effort with competence to engage political institutions systematically over time. The second is the failure to sustain the decentralized openness in the transition to more structured political organization. The third failure mode of the internet and democracy refers to the power of well-organized, data-informed central powers to move millions of people from the center out, instead of the other way around. The fourth failure mode is that precisely what makes decentralized networks so effective at circumventing established forms of control can also make them the vehicles of repressive mobs. The final failure mode is the susceptibility to disinformation and propaganda.

Keywords:   internet, democratization, decentralization, democracy, decentralized networks, repressive mobs, disinformation, propaganda, central powers, political organization

AS WE WRITE this book in early 2018, the prevailing zeitgeist seems to be that all the promise of the internet has been swept away in a cloud of manipulation and abuse. Facebook finds itself under questioning on both sides of the Atlantic for its uses of the personal data of its users. If there are battles over who did what, they seem to be about who used the internet as one big manipulation platform, whether it was the Russians, Cambridge Analytica, or the Trump campaign, helped by Facebook and Google targeted advertising experts. If there was a sigh of relief after the German elections that the Russians did not interfere, it was explained as a function of the central role and high trust that traditional public broadcasters enjoy in the German media ecosystem.

It cannot be, however, that “the internet democratizes” when it enables people who think as we do to challenge institutionalized power that protects institutions we would rather challenge, but “the internet threatens democracy” when it allows people with whom we disagree to challenge institutionalized power that protects institutions we would rather protect. It strains credulity that the Trump candidacy or the Leave campaign were entirely the product of manipulation, Russian or otherwise, rather than the surprising political success of a campaign that tapped into attitudes and beliefs held by millions of people in the United States and the United Kingdom, respectively. One of the reasons it is so difficult, for example, to identify Russian campaigns in the American media ecosystem is that they were so often congruent with the organic online mobilization of people who in fact wanted to go out and protest against immigration or share memes that denigrate mainstream media. These views may be illiberal, but a media ecosystem that effectively suppresses widely held beliefs that are contrary to elite opinions is not more (p.342) “democratic” if elite opinions happen to be pluralistic and egalitarian than it is when elite opinions merely protect oligarchy.

The fact that Donald Trump was able to spin up a campaign with little support from the Republican Party, is democratizing—in the sense that it enabled an insurgent campaign that was far from mainstream consensus to emerge and give voice to the anxieties and ambitions of millions in Americans who had not been heard so clearly before in the halls of power. Much of his success, as we documented earlier in the book, was due to the interaction between partisan media on the right and extensive mainstream media coverage; but his own campaign organization and spending suggest that the campaign itself was focused on using Facebook, Twitter, and online mobilization to attract millions. The fact that Bernie Sanders, a party outsider, was able to run a credible campaign and raise $230 million for his campaign, 60 percent of which was from small individual contributions, is democratizing in the sense of allowing outsiders to challenge the consensus establishment candidate.

The core argument that the internet was a democratizing technology had to do with the fact that in the mass mediated environment, agenda setting and framing of what mattered and what was a credible move in political contestation was centralized and depended on access to organizational and institutional power. The cost of production and distribution of news and opinions; the cost of mobilizing on a national scale; the credibility to assert in public what was, and was not, news or fact; all these were associated with a relatively concentrated power structure. Government, parties, professional commercial media, organized corporations, organized labor, to some extent national-scale membership organizations—all these were necessary to move an idea or a demand from the peripheries of late twentieth-century democracy to the core of its political debate. The internet promised to open new pathways for agenda setting, framing, and mobilization. People, with diverse viewpoints, were able to find each other.1 Citizens and residents could mobilize into new, flatter organizational forms.2 More of us could publish, and more of us could read a broader set of sources and viewpoints, and we could be more engaged when we did so.3 These effects could then spill over into broader media because journalists would also read these decentralized media and their coverage could be shaped not only by political and corporate insiders but by a broader range of public authors.4 These voices, amateur and professional, commercial and noncommercial, mobilized and otherwise, could together form a networked public sphere where power to set the agenda and frame political discourse was more widely distributed than it had (p.343) been in the latter twentieth century.5 Critically, the organizational literature focused on the distributed, nonhierarchical nature of these mobilization efforts, ushering in a new era where collective action did not depend on large political machines.6

All of these features are still true, and the affordances continue to be available for people to organize themselves around establishment organizations, rather than having to run through them. It is impossible to ignore the fact that the national debate on police shootings of black men was fundamentally altered not because one of the major parties took it up; not because the number of shootings increased; but because bystanders shot videos of police shootings, shared them online, and the Black Lives Matter movement developed, at first primarily online, around these shocking images. After the election of Donald Trump, activists organized through a Google Doc to organize the Indivisible movement, which coordinated citizens to appearance at town hall meetings to put pressure on congressional representatives. The Women’s March on Washington could not have developed except through online engagement. The high school students who organized around school shootings and gun control follow the same pattern.

The second decade of the twenty-first century has taught us that while decentralization can support democratization, it is susceptible to five major failure modes. The first of these became clear after the Arab Spring and Occupy: leaderless, decentralized movements can form relatively easily online and focus on distinct great moments, like Tahrir, Gezi Park, or Zuccotti Park. But converting protest into action requires more structure, discipline, and longevity than many of these movements were able to develop, as Zeynep Tufekci captured so well in Twitter and Teargas.7 The first failure mode is simply the failure to convert from a moment’s surge of decentralized passion into a longer-term, sustained effort with competence to engage political institutions systematically over time.

The second failure mode is the failure to sustain the decentralized openness in the transition to more structured political organization. It turns out that marrying the energy of decentralized online action to the discipline of a party or government is far from easy. Howard Dean’s 2004 primary campaign was the first significant political effort that leveraged the possibilities of the radically distributed network of activists to raise a campaign in a truly decentralized, peer-driven way.8 That particular, fully decentralized, user-generated campaign was later foundational to the Ron Paul presidential primary campaign in 2008. While both campaigns benefited from enormous distributed energy, neither was able to harness and organize (p.344) it. By 2008 “MyBarackObama.com” took over from the decentralized peers of the Dean and Paul campaigns and integrated it into this amazing new technology—the social network platform. MyBO provided a centralized platform, but still emphasized the decentralized capabilities—people creating their own meetups; raising their own campaign bundling efforts; communicating their own news. As Micah Sifry documented, however, the Obama transition team made a conscious choice to dismantle the grassroots army that had helped propel the new president into the White House in 2008.9 By the end of the campaign, MyBO had two million active members, 70,000 of whom had started their own fundraising campaigns. Sifry describes how several key actors, most importantly Christopher Edley and Mitch Kapor, pushed hard for the new presidency to introduce a new, perpetual grassroots-engaged democracy, where the presidency and the people use the internet for continuous engagement and participation on a mass scale. Instead, Democratic Party insiders, threatened with loss of control, persuaded the incoming president to fold his online campaign into the DNC, turning MyBO into a new Organizing for America site. As Sifry describes it,

Shunted into the DNC, MyBO’s tools for self-organizing were dismantled within a year. Instead of calling on supporters to launch a voter registration drive or build a network of small donors or back state and local candidates, OFA deployed the campaign’s vast email list to hawk coffee mugs and generate thank-you notes to Democratic members of Congress who backed Obama’s initiatives. As a result, when the political going got rough, much of Obama’s once-mighty army was AWOL. When the fight over Obama’s health care plan was at its peak, OFA was able to drum up only 300,000 phone calls to Congress. After the midterm debacle in 2010, when Democrats suffered their biggest losses since the Great Depression, Obama essentially had to build a new campaign machine from scratch in time for his reelection effort in 2012.

Hierarchical, centralized organizations have very different interests and concerns than decentralized networks of activists and citizens. And Obama’s 2012 campaign was the first test run of the third failure mode of the internet and democracy—the power of well-organized, data-informed central powers to move millions of people from the center out, instead of the other way around. It was there that the tide first turned from the internet as decentralized enablement platform to internet as data-driven control platform.

(p.345) Hailed by many at the time as a success, the Obama 2012 campaign represented the first systematic use of big data and individualization for a campaign to target individual voters, particularly in the get-out-the-vote efforts. Instead of a campaign site like MyBO that had leveraged supporters own social networks to organize events, volunteers, and fundraising, it was the data geeks who were the new heroes. And the internet they used was more about surveillance through social media systems developed for marketing than about participatory mobilization. The title and subtitle of Alexis Madrigal’s celebratory Atlantic story right after the election captures it all: “When the Nerds Go Marching In: How a Dream Team of engineers from Facebook, Twitter, and Google built the software that drove Barack Obama’s reelection.”10 The triumphalism only gradually gave way to a more sober assessment of the implications of this shift toward the capacity of a campaign to control mass mobilization and how much it inverted the dynamics we had originally celebrated as the democratizing effect of the internet.

Soon thereafter, scientific papers from Facebook researchers—documenting how Facebook could manipulate the moods of Facebook users by altering the prevailing sentiment in their newsfeed or increase voter turnout with notifications that a friend had voted—launched a new set of concerns with algorithmic manipulation of politics.11 Zeynep Tufekci’s “Engineering the public: Big data, surveillance, and computational politics” became the first academic essay that highlighted the core affordances that made the American internet such a powerful vehicle for political manipulation.12 Tufekci identified the ability of social media and internet platforms to leverage big-data techniques to provide individually tailored, experimentally validated, targeted communication that could leverage the most cutting-edge behavioral science to manipulate the beliefs and attitudes of users, using algorithmic processes that are entirely opaque to external review and accountability. In this story Facebook and Google, much more than Russia or any state, were the primary culprits, building platforms for refined, scientifically informed manipulation of human individuals at population scales and doing so in pursuit of profit under the newly emerging surveillance capitalism13—but with clear risks for the very possibility of democratic politics. The major threat to the internet as a democratizing force is not the abuse of the system but its intended use. Because the internet had been reengineered, primarily by Google and Facebook, into a preference manipulation platform, the two companies emerged as behemoths whose entire business model was built on producing fine-grained data at the individual level, running mass-scale experiments on millions of users, and developing the capacity to update the (p.346) online interaction in real time to achieve behavioral manipulation made to order and sold to advertisers. As we described in Chapter 9, the evidence that these techniques were critical and effective in the 2016 election is sparse. But the fact that something has not yet been shown to have had the critical marginal impact is no reason to ignore it as a major threat or constraint on how we use the internet in a democratic society. And of all the major threats to democracy that we have encountered in this decade, it is the one most amenable to regulatory or policy intervention.

The fourth failure mode is that precisely what makes decentralized networks so effective at circumventing established forms of control can also make them the vehicles of repressive mobs. Here, the failure is not inefficacy, subversion by organizations, or reconcentration in the hands of centralized organizations that turn the network into a control system. It is rather that the methods or goals of the distributed network of actors are themselves repressive, rather than participatory. This critique of decentralized networks as democratizing followed the Gamergate controversy, as networked mobilization was turned into a harassment and intimidation campaign against women, game developers, and media critics, in the name of preserving geek masculinity and fighting for free speech against “social justice warriors” out to destroy it.14 Adrienne Massanari analyzed the design and algorithms of Reddit, showing how these affordances enabled the Gamergate campaign to take off,15 while Shira Chess and Adrienne Shaw focused on how the Gamergaters’ attack on academic feminist studies of gaming culture exhibited the classic characteristics of the conspiracy theory-driven, paranoid style, in American politics.16 What was different was not that earlier studies of the networked public sphere focused on left or liberal mobilization. These had always included examples for both left and right sites circumventing mainstream media to shape the agenda.17 As Alice Marwick and Rebecca Lewis show, what Gamergate exposed was a new set of distributed attack techniques that went well beyond what could be considered still within democratic practice: doxxing (disclosing private documents) private individuals who expressed opposing positions; intimidation through personal death and rape threats; revenge porn and social shaming; all aggressively gendered and aimed at women. Moreover, Gamergate was the proving ground for a set of techniques that later developed as the core techniques of disinformation in the 2016 election: organized brigades of online harassment organized around hashtags or specific memes; use of sockpuppets to push topics and stories to prominence, and in particular to put them in front of mainstream influencers to propagate the message to society at large; and a networked coalition of (p.347) loosely affiliated groups around core objects of hatred of the kind we saw more in the Unite the Right campaign in Charlottesville, Virginia. It was also where some of the core activists of the alt-right got their start.18 Again, though, it is important to distinguish elements of the strategy that are expressive—like propagating memes or putting them in front of mainstream influencers, and elements that are about intimidating opponents. We focus here only on the latter, unless the expressive actions fall within what we defined and described throughout this book as disinformation. This mode is a “failure” even if it does not actually affect the outcomes of election or policy battles. Because intimidation campaigns can be personal and relatively nihilistic—not caring about discrete policy outcomes—even if they do not affect an outcome at the aggregate level, they can and do harm their targets.

The final failure mode is the susceptibility to disinformation and propaganda and is the motivating force behind this book. The critical thing to understand is that the internet democratizes, if it does, only through its interaction with preexisting institutions and organizations—working with them and around them, and creating new alternatives that interact with them. As one of us put it in an article in 2009, “Like other information goods, the production model of news is shifting from an industrial model—be it the monopoly city paper, IBM in its monopoly heyday, Microsoft, or Britannica—to a networked model that integrates a wider range of practices into the production system: market and nonmarket, large scale and small, for profit and nonprofit, organized and individual. We already see the early elements of how news reporting and opinion will be provided in the networked public sphere.”19 It would be the interaction between “surviving elements of the old system,” changed, small-scale new, digital native commercial media, volunteer media like Daily Kos and Townhall, more effective nonprofits (more effective because they could do much more with less online than in the mass media era), and mobilized individuals that would form the networked fourth estate. The media ecosystem needed all these elements, including the healthy surviving element of traditional professional media and nonprofit professional journalism, to be able to benefit from the edgier, more partisan distributed, mobilized action. What we saw repeatedly in our case studies throughout the book was that this integrated system worked like a networked public sphere or fourth estate outside the right-wing media ecosystem thanks to the reality-check dynamic that characterized the interactions among professional media and newer online media, professional and amateur, for profit and nonprofit, fact-checking organizations and activists. But when the professional, commercial, and nonprofit think tanks, (p.348) like Fox News, Breitbart, the Government Accountability Initiative, the Center for Immigration Studies, or Judicial Watch, all functioned to reinforce the disinformation campaigns, it was then that the networked public sphere turned into a networked propaganda system.

We have only presented data here from the United States. It is entirely possible that observing the patterns in other countries will offer better evidence that it was, in fact, the internet or social media or just Facebook that has created the epistemic crisis we observe in the United States, Europe, and other democratic societies. But if our experience in America is representative, then any analysis of a perceived epistemic crisis anywhere must consider how the internet interacts with that country’s entire media ecosystem, and how that system in turn interacts with that country’s political-institutional system more generally.

We have presented in this Part Four an approach based on a broader sense of political economy than simply a media- or technology-centric view. We have suggested that many factors have contributed to and fed off of the asymmetric polarization in the positions of the two parties in America, with a conservative wing particularly focused on ideological purity and symbolic action. These factors include the long-term patterns of identity threat born of race relations and relatively open immigration policy; the economic insecurity born of policies aimed to reduce taxes and services, reduce labor power, and reduce regulatory oversight over businesses; and an increasing mismatch between a large, deeply religious population and an increasingly pluralistic and gender-egalitarian mainstream. These patterns laid the ground work for the 30-year ascendance of a commercially successful strategy of news and opinion media outlets committed to serving identity-confirming news and views for that conservative wing, while denigrating the veracity and honesty of all other outlets. Changes in technology (satellite and FM radio, cable transmission, and only later the internet), institutions (the deregulation of cable, repeal of the fairness doctrine, reduction in ownership limits, and reduced antitrust enforcement) and political culture have shaped the emergence of a network of media outlets, including radio, cable television, and internet sites, that operate in a distinctly partisan propagandist mode. While there were efforts of liberals to create a parallel but functionally similar system, these failed because the coalition that formed the rest of the polity simply did not have the same cohesion to support ideologically pure, reality-agnostic media. There were certainly elements of this coalition who were (and are) more than happy to consume and produce bias-confirming news and opinion. We have (p.349) seen, from MSNBC and online, such examples in several of our chapters. But the coalition as a whole was too focused on functional results to support a dominant propagandist model. Its members were not willing to concentrate on bias confirmation to the exclusion of what professional journalism and expertise, whether commercial, governmental, or nonprofit, could offer, and once the conservative propagandist outlets took off, fact-based journalism offered a valuable source of credible, independent, but belief-consistent rebuttal for liberals. And so the media and information consumption habits of the two parts of the media ecosystem developed separately and distinctly. As we in the United States, and anyone else in any other democracy, turn to look at solutions, we must consider these background political institutional dynamics to understand where they are and where they are likely to go, what their susceptibilities to disinformation are, and what nationally specific sources of resilience they may be able to harness. It would be particularly unfortunate if countries that do not have the same decades-long processes that made the United States susceptible to propaganda and disinformation, foreign and domestic, will adopt measures that will undermine the democratizing aspects of the internet and social media even though they do not, in fact, face the same risks, or even though they have sources of resilience that are more robust than those we appear to have in America. (p.350)


(1.) Henry Farrell, “The Consequences of the Internet for Politics,” Annual Review of Political Science 15, no. 1 (June 15, 2012): 35–52, https://doi.org/10.1146/annurev-polisci-030810-110815.

(2.) Bruce A. Bimber, Information and American Democracy: Technology in the Evolution of Political Power, Communication, Society, and Politics (Cambridge, UK; New York: Cambridge University Press, 2003); David Karpf, The MoveOn Effect: The Unexpected Transformation of American Political Advocacy, Oxford Studies in Digital Politics (New York: Oxford University Press, 2012).

(3.) Eszter Hargittai, Jason Gallo, and Matthew Kane, “Cross-Ideological Discussions among Conservative and Liberal Bloggers,” Public Choice 134, no. 1–2 (November 20, 2007): 67–86, https://doi.org/10.1007/s11127-007-9201-x; Eric Lawrence, John Sides, and Henry Farrell, “Self-Segregation or Deliberation? Blog Readership, Participation, and Polarization in American Politics,” Perspectives on Politics 8, no. 01 (March 2010): 141, https://doi.org/10.1017/S1537592709992714; Matthew Gentzkow and Jesse M. Shapiro, “Ideological Segregation Online and Offline,” The Quarterly Journal of Economics 126, no. 4 (2011): 1799–1839.

(4.) Henry Farrell and Daniel W. Drezner, “The Power and Politics of Blogs,” Public Choice 134, no. 1–2 (November 20, 2007): 15–30, https://doi.org/10.1007/s11127-007-9198-1; Kevin Wallsten, “Agenda Setting and the Blogosphere: An Analysis of the Relationship between Mainstream Media and Political Blogs: Agenda Setting and the Blogosphere,” Review of Policy Research 24, no. 6 (December 17, 2007): 567–587, https://doi.org/10.1111/j.1541-1338.2007.00300.x.

(5.) Yochai Benkler, The Wealth of Networks: How Social Production Transforms Markets and Freedom (New Haven, CT: Yale University Press, 2006).

(6.) Id. Benkler; Clay Shirky, Here Comes Everybody: The Power of Organizing without Organizations: [With an Updated Epilogue], Nachdr. (New York, NY; Toronto; London: Penguin Books, 2009); Bruce Bimber, Andrew Flanagin, and Cynthia Stohl, “Collective Action in Organizations: Interaction and Engagement in an Era of Technological Change: Cambridge University Press, New York, 2012, 224 Pp., Appendices, References, Index, $29.99 (Paperback),” Voluntas: International Journal of Voluntary and Nonprofit Organizations 25, no. 3 (June 2014): 847–848, https://doi.org/10.1007/s11266-013-9413-2; W. Lance Bennett and Alexandra Segerberg, “THE LOGIC OF CONNECTIVE ACTION: Digital Media and the Personalization of Contentious Politics,” Information, Communication & Society 15, no. 5 (June 2012): 739–768, https://doi.org/10.1080/1369118X.2012.670661.

(7.) Zeynep Tufekci, Twitter and Tear Gas: The Power and Fragility of Networked Protest (New Haven, CT; London: Yale University Press, 2017).

(8.) Micah L Sifry, The Big Disconnect Why the Internet Hasn’t Transformed Politics (Yet) (New York: OR Books, 2014); “The Decembrist: Dean’s Penguin, or Technology and the Nature of Political Interaction,” accessed January 29, 2018, http://markschmitt.typepad.com/decembrist/2003/12/deans_pengun_or.html.

(9.) Micah L. Sifry, “Obama’s Lost Army,” The New Republic, February 9, 2017, https://newrepublic.com/article/140245/obamas-lost-army-inside-fall-grassroots-machine.

(10.) Alexis C. Madrigal, “When the Nerds Go Marching In,” The Atlantic, November 16, 2012, https://www.theatlantic.com/technology/archive/2012/11/when-the-nerds-go-marching-in/265325/.

(11.) Adam D.I. Kramer, Jamie E. Guillory, and Jeffrey T. Hancock, “Experimental Evidence of Massive-Scale Emotional Contagion through Social Networks,” Proceedings of the National Academy of Sciences 111, no. 24 (June 17, 2014): 8788–8790, https://doi.org/10.1073/pnas.1320040111.

(12.) Zeynep Tufekci, “Engineering the Public: Big Data, Surveillance and Computational Politics,” First Monday 19, no. 7 (July 7, 2014), https://doi.org/10.5210/fm.v19i7.4901.

(13.) Shoshana Zuboff, “Big Other: Surveillance Capitalism and the Prospects of an Information Civilization,” Journal of Information Technology 30, no. 1 (March 2015): 75–89, https://doi.org/10.1057/jit.2015.5.

(14.) For a brief overview, see Alice Marwick and Rebecca Lewis, “Media Manipulation and Disinformation Online” (Data & Society, May 15, 2017), 7–9, https://datasociety.net/pubs/oh/DataAndSociety_MediaManipulationAndDisinformationOnline.pdf. For the leading academic treatments, see Andrea Braithwaite, “It’s About Ethics in Games Journalism? Gamergaters and Geek Masculinity,” Social (p.439) Media + Society 2, no. 4 (November 2016): 205630511667248, https://doi.org/10.1177/2056305116672484.

(15.) Adrienne Massanari, “#Gamergate and The Fappening: How Reddit’s Algorithm, Governance, and Culture Support Toxic Technocultures,” New Media & Society 19, no. 3 (March 1, 2017): 329–346, https://doi.org/10.1177/1461444815608807.

(16.) Shira Chess and Adrienne Shaw, “A Conspiracy of Fishes, or, How We Learned to Stop Worrying About #GamerGate and Embrace Hegemonic Masculinity,” Journal of Broadcasting & Electronic Media 59, no. 1 (March 11, 2015): 208–220, https://doi.org/10.1080/08838151.2014.999917.

(17.) The Free Republic Forum was one of the first such examples on the right. See Yochai Benkler, “Free As the Air to Common Use: First Amendment Constraints on Enclosure of the Public Domain,” New York University Law Review 74, no. 2 (May 1999): 354–446. The Dan Rather takedown was every bit as prominent as Trent Lott’s demise in the early discussions of online mobilization. See Yochai Benkler, The Wealth of Networks: How Social Production Transforms Markets and Freedom (New Haven, CT: Yale University Press, 2006), chap. 7.

(19.) Yochai Benkler, “CORRESPONDENCE: A New Era of Corruption?,” The New Republic, March 4, 2009, https://newrepublic.com/article/61997/correspondence-new-era-corruption.