Once upon a time, only the elite could network globally. David Rockefeller—the grandson of the oil tycoon John D. Rockefeller—was a pioneer networker. According to a recent report, “He recorded contact information along with every meeting he had with about 100,000 people world-wide on white 3-by-5-inch index cards. He amassed about 200,000 of the cards, which filled a custom-built Rolodex machine, a 5-foot high electronic device.” Rockefeller’s contacts ranged from President John F. Kennedy to the shah of Iran, Pope John Paul II and the astronaut Neil Armstrong. Henry Kissinger, a former secretary of state, garnered the most cards—35, describing hundreds of encounters over sixty years. Close behind was Gianni Agnelli, the Italian industrialist who ran Fiat. In his memoirs, Rockefeller recalled how he had learned the importance of contacts while serving as an Army intelligence officer during World War II. “My effectiveness depended on my ability to develop a network of people with reliable information,” he wrote. It was an experience that he took home with him when he joined Chase Manhattan after the war.1
What was once the preserve of a tiny elite is now available to everybody with an Internet connection and a smartphone, tablet or laptop computer. Facebook was founded at Harvard in 2004, before David Rockefeller’s 90th birthday, and rose rapidly to become the world’s dominant social media platform. The company’s foundational premise was and remains that “Simply through sharing and connecting, the world gets smaller and better.”2 Connecting people on Facebook, Mark Zuckerberg declared in 2015, was building a “common global community” with a “shared understanding.” A year later he summed up his utopian vision in a Facebook Live session (with Jerry Seinfeld):
We’re at this next point in human civilization, where we have the next set of tools that we need, things like the internet, that can be this global communication infrastructure ... Just like we went from hunter-gatherers to villages and cities and then nations, I think we now need to come together as a global community. Because a lot of the problems that you’re talking about, whether it’s terrorism or the refugee crisis or climate change or global diseases spreading around the world—these are not things that can be solved by any one city or one nation or one small group of people.
Humanity is certainly connected as never before. One in three people—2.48 billion—worldwide used a social network in 2017, according to eMarketer, nearly nine per cent more than the previous year, thanks to rapid expansion in Asia-Pacific, Latin America, the Middle East and Africa. Three quarters of all users of the Internet on mobile phones used their devices to access social media. In the United States, an estimated 60 per cent of the population will use a social network at least once a month this year, up 2.6 per cent since 2017. The number of U.S. Facebook users is expected to reach 169.5 million in 2018, more than 60 per cent of all Internet users. Penetration is similarly high in the United Kingdom, where half the population is on Facebook.
How far the resulting network can be regarded as a problem-solving global community remains an open question, as we shall see. What is beyond doubt is that network platform companies are astonishingly profitable businesses—not least because users have handed them their personal data for nothing. As a commenter on the website MetaFilter memorably observed in 2010: “If you are not paying for it, you’re not the customer; you’re the product being sold.” In terms of its revenues, Facebook is primarily an advertising company, as is Google’s parent company Alphabet. As a half-smirking Zuckerberg explained to Senator Orrin Hatch at a congressional hearing in 2018, “We sell ads, Senator.”
Eight of the world’s most highly valued companies in 2017 were technology businesses, with a market capitalization equal to nearly a third of the market capitalization of the other 92 companies in the global top hundred. Of these eight companies, five (Apple, Alphabet, Microsoft, Amazon and Facebook) were American, two Chinese (Alibaba) and Tencent) and one South Korean (Samsung). It was not strictly speaking software that “ate the world,” in Marc Andreessen’s famous phrase, because two of these companies sell hardware. It would be more accurate to say that network platforms ate the world, as the market dominance of all these companies arises from network effects and the operation of Zipf’s Law (which can be summed up “winner takes nearly all”).
Amazon ate bookselling. Jeff Bezos’s company today sells 55 per cent of all the books sold in the United States, 82 per cent of all the e-books and 99 per cent of all the audiobooks. Nor is that all. Although its share of total U.S. retail remains small (4 per cent, or half of Walmart’s share), the company’s share of the global cloud business is 34 per cent, its share of U.S. online commerce is 44 per cent, and its share of the voice-activated device market is 71 per cent.3 Google ate search. It accounts for between 87 and 92 per cent of online searches worldwide, processing 63,000 queries a second. Apple ate music (along with Alphabet’s YouTube and Spotify). YouTube ate television. Above all, Google and Facebook ate advertising. According to eMarketer, the two companies will capture a combined 56.8 per cent of U.S. digital ad spending in 2018, though Amazon and Snapchat have been increasing their shares. The revenues from this source seem certain to continue growing as a rising share of total advertising expenditure goes to the Internet.
“For many years,” according to a New Yorker profile, “[Mark] Zuckerberg ended Facebook meetings with the half-joking exhortation ‘Domination!’” He stopped doing this because in European legal systems “dominance” is a term used to describe a business monopoly. However, he remains unabashed about Facebook’s appetite for market share. “There’s a natural zero-sumness,” he told an interviewer in September 2018. Revealingly, the figure he most admires in history is the Emperor Augustus:
You have all these good and bad and complex figures [in ancient Rome]. I think Augustus is one of the most fascinating. Basically, through a really harsh approach, he established two hundred years of world peace. What are the trade-offs in that? On the one hand, world peace is a long-term goal that people talk about today. Two hundred years feels unattainable. [But] that didn’t come for free, and he had to do certain things.4
Facebook is indeed an empire, with as many users as Christianity has adherents, but a workforce of just 23,000. Nor does the new Caesar render up much to the old one: between 2007 and 2015, Facebook paid just 4 per cent of its profits in taxes. Amazon paid only 13 per cent, Google 16 per cent and Apple paid 17 per cent. (The average S&P 500 company paid 27 per cent.)
Market dominance is seldom without its political aspect. Prior to 2016, remarkably little attention was paid to the growing political power of the big technology companies. The posture they adopted might best be summed up as faux naïf. “Don’t be evil” was the motto adopted by Google in July 2001, after a brainstorming session between Eric Schmidt and early employees shortly before he took over as chief executive. As Schmidt recalled, referring to the company’s decision to exit China rather than comply with Beijing’s censorship requirements, “We actually did an ‘evil scale’ and decided [that] not to serve at all was worse evil.” By contrast, close involvement in the administration of Barack Obama—including energetic efforts to secure his reelection in 2012—was deemed to lie at the other end of the evil scale. According to one estimate, there were 252 job moves between Google and the Obama administration from its inception to early 2016, and 427 meetings between White House staff and Google employees from 2009 to 2015. In 2012, Google was the nation’s second-largest corporate spender on lobbying, behind General Electric. By 2017 Google was number one, spending $18 million.5 The closeness of the relationship between Silicon Valley and the Democratic Party was not unknown at the time, but it was remarkably uncontroversial. Little, if any, attention was paid to the political activities of the other big technology companies.
Donald Trump’s victory in the November 2016 presidential election changed that, not just because the leaders of the big technology companies had confidently expected him to lose, but also because it was immediately apparent to them that their core products had either helped Trump win or failed to avert Clinton’s defeat. The fact that Trump had dominated Clinton on both Facebook and Twitter throughout the campaign had been overlooked by most political pundits because his huge lead in follower numbers was at odds with all opinion polls. Likewise, experts ignored or discounted his even larger lead over her in terms of Google searches. Brad Parscale, Trump’s digital media director, put it well: “These social platforms are all invented by very liberal people on the west and east coasts. And we figure out how to use it to push conservative values. I don’t think they thought that would ever happen.” A video recording of a post-election internal meeting at Google confirms this. Senior executives lined up to express their dismay at Trump’s victory and their allegiance to Clinton.6 As President Obama told David Letterman, his own successful use of social media in 2008 had left him and other Democrats with “a pretty optimistic feeling about it. … What we missed was the degree to which people who are in power [sic], special interests, foreign governments, etcetera, can in fact manipulate that and propagandize.” Obama’s analysis of what had gone wrong for his party and its presidential candidate in 2016 is worth quoting in full:
One of the biggest challenges we have to our democracy is the degree to which we don’t share a common baseline of facts. … What the Russians exploited but it was already here is we are operating in completely different information universes. If you watch Fox News, you are living on a different planet than you are if you listen to NPR. … That’s what’s happening with these Facebook pages where more and more people are getting their news from. At a certain point you just live in a bubble. And that’s part of why our politics is so polarized right now. I think it is a solvable problem but it’s one we have to spend a lot of time thinking about.7
Speaking at MIT in February 2017, Obama suggested that “the large platforms—Google and Facebook being the most obvious, but Twitter and others as well that are part of that ecosystem—have to have a conversation about their business model that recognizes they are a public good as well as a commercial enterprise.” It was, he said, “very difficult to figure out how democracy works over the long term” when “essentially we now have entirely different realities that are being created, with not just different opinions but now different facts—different sources, different people who are considered authoritative.”8
Obama was right. The status quo, in which a handful of giant corporations dominate the public sphere but tend not only to divide it but also to pollute it with a torrent of fake news* and extreme views—at least some of it generated by this nation’s foes—is not defensible. Something needs to change. But what? What exactly does it imply to say that the network platforms are “a public good as well as a commercial enterprise”? In short, to quote Nikolai Chernyshevsky (from whom Lenin stole the famous title): “What is to be done?”
The Self-Non-Regulation of the Internet
The starting point for any serious analysis must be the business model of the network platforms. They are, in Tim Wu’s phase, “attention merchants,” the heirs of the big twentieth-century media companies, such as Hearst, which used news and other mostly non-fictional content to attract readers’ attention, selling space alongside articles and photographs to advertisers.9 They have a lot of attention to sell—more than any print publishing company in history. The average American spends 5.5 hours a day using digital media, more than television, radio and print put together.10 More than half that digital media time is spent on mobile devices. An average smartphone user clicks, taps and swipes 2,617 times per day. In April 2016, Facebook said it was capturing on average 50 minutes of every American’s day, up from 40 minutes in July 2014, though usage may have declined slightly in 2017.11
Most of what is said online is inane. The ten most common words used in Facebook status updates are day, loud, word, ticket, nice, long, light, hangover, good and vote.12 A little of what is said is highly cerebral. Some is old-fashioned fiction: stories that do not purport to be true. But a significant proportion is “news,” i.e., content that purports to be true information about current affairs. In 2017, two thirds of American adults said they got news from social media sites. Around three quarters of Twitter users got news from the application, around two thirds of Facebook users, and around a third of YouTube users. In all, 45 per cent of American adults get news from Facebook, 18% of them from YouTube and 11 per cent from Twitter. A significant share of younger users also gets news from Instagram and Snapchat.13 It is a startling fact that Facebook and Google are now responsible for nearly 80 per cent of news publishers’ referral traffic.
The network platforms have set out to act as aggregators of news for the simple reason that it engages users’ attention. If extracts from the novels of Dickens or Pascale’s Pensées had the same appeal, these too would feature in users’ “feeds” or search results. The key point is that the network platforms customize the news that users see in order to maximize their engagement. When Mark Zuckerberg talked in 2013 of making Facebook “the best personalized newspaper in the world,” this was what he meant. News Feed is a “personalized collection of stories,” and a user sees on average of 220 per day. Advertising and Pages, dedicated profiles for groups or causes, are sources of stories in News Feed. Anyone can buy ads to promote Pages, using an automated interface. Whenever one of Facebook’s users opens the Facebook app, a personalization algorithm sorts through all the posts that a person could potentially see, serving up and sorting the fraction it thinks he or she would be most likely to share, comment on, or like. (Shares are worth more than comments, which are both worth more than likes.) Around two thousand pieces of user data (“features”) are used by Facebook’s machine-learning system to make those predictions. A somewhat similar process works when a user enters words in the Google search box. The user’s individual search history, geographic location, and other demographic information affect the content and ranking of the results.
The problem is that the algorithms are not prioritizing truthfulness or accuracy but user engagement. For example, on October 1, 2017, Google directed users towards a false story alleging that the perpetrator of the Las Vegas massacre on that date was a member of the far-left group Antifa. A study by the Wall Street Journal and former YouTube engineer Guillaume Chaslot showed that a user searching for “The Pope” on YouTube was directed to videos with titles such as “How Dangerous is the Pope?”, “What if the Pope was assassinated?” and “BREAKING: They caught the Pope.”14 As we shall see, these and other features of the network platforms had historic consequences in 2016 when they played a decisive role in the election of Donald Trump as the U.S. president.
How did we arrive at this state of affairs—when such important components of the public sphere could operate solely with regard to their own profitability as attention merchants? To answer this question, we must briefly review the history of Internet regulation. A decisive early decision was to define the Internet as a Title I information service, and therefore fundamentally different from the old telephone network, which was governed by Title II’s intrusive monopoly utility regulations. (The Internet was briefly re-classified as a Title II service between 2015 and 2017, but no major regulatory change occurred in that period.) Another important decision was to give Internet companies very lenient treatment when they violated copyright. The Digital Millennium Act’s notice-and-takedown provisions minimized the penalties to the network platforms of making the intellectual property of others available gratis to their users. A third vital decision was enshrined in Section 230 of Title V of the 1996 Telecommunications Act,15 which was enacted after a 1995 New York court held an online intermediary liable for a user’s defamatory posts. Previously, managing content had triggered classification as a publisher—and hence civil liability—creating a perverse incentive not to manage content at all. Thus, Section 230c, “Protection for ‘Good Samaritan’ blocking and screening of offensive material,” was written to encourage nascent firms to protect users and prevent illegal activity without incurring massive content management costs. It states:
1. No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.
2. No provider or user of an interactive computer service shall be held liable on account of:
A. any action voluntarily taken in good faith to restrict access to or availability of material that the provider or user considers to be obscene [etc.]; or
B. any action taken to enable or make available to information content providers or others the technical means to restrict access to material in paragraph 1.
In essence, Section 230 gives websites immunity from liability for what their users post. The net result of this regulatory framework is that technology companies are neither communication utilities nor content publishers.
The argument for Section 230, as articulated by the Electronic Frontier Foundation, is as follows: “Given the sheer size of user-generated websites … it would be infeasible for online intermediaries to prevent objectionable content from cropping up on their site. Rather than face potential liability for their users’ actions, most would likely not host any user content at all or would need to protect themselves by being actively engaged in censoring what we say, what we see, and what we do online.” Senator Ron Wyden puts it even more strongly: “If websites, ISPs, text message services, video game companies and any other type of platform were held liable for every word and deed they facilitated or somehow enabled, the entire system would shut down … collaboration and communication on the internet would simply cease.”16 In effect, it has been argued, Section 230 split the difference between liability, which would have meant restriction, or complete lack of curation, which would have led to a torrent of “filth, racism, insults, and pornography.” Thus, “hobbling 230” would “stifle the competition that got us to today’s rich internet in the first place.”17 According to one recent and influential account:
Platforms moderate content because of a foundation in American free speech norms, corporate responsibility, and the economic necessity of creating an environment that reflects the expectations of their users. Thus, platforms are motivated to moderate by both of §230’s purposes: fostering Good Samaritan platforms and promoting free speech. … [They] should be thought of as operating as the New Governors of online speech. These New Governors are part of a new triadic model of speech that sits between the state and speakers-publishers. They are private, self-regulating entities that are economically and normatively motivated to reflect the democratic culture and free speech expectations of their users.18
Note that under the present dispensation, the network platforms have the power (not the obligation) to “curate” content that they host. They do so, it is argued, “out of a sense of corporate social responsibility, but also, more importantly, because their economic viability depends on meeting users’ speech and community norms.” This curation began some time ago with the exclusion of content that no one would publicly condone. For years, the big technology companies have shared “signatures” of child pornography, so that once an illegal photo or a video is uploaded to one site, it is detected and taken down or excluded from all platforms. Facebook, Twitter, YouTube and Microsoft have a global working group that applies the same kind of technologies to terrorist content. In November 2017, for example, YouTube took down videos of Anwar al-Awlaki, the jihadist cleric killed by a U.S. drone strike in Yemen in 2011. Video fingerprinting technology has not removed al-Awlaki altogether from the platform, but it has substantially reduced the number of videos relating to him.
However, the process of removing or at least downgrading offensive content has not stopped with recognized advocates of pedophilia or jihad. In January 2018 YouTube removed from its Google Preferred platform the channels of Logan Paul, a YouTube star with almost 16 million subscribers, after he posted a video showing a suicide victim in Japan. “Demonetizing” YouTube videos, so that they are not promoted on the platform and their creators receive no share of any advertising revenue, is a powerful sanction short of outright prohibition. Twitter set out to be the “free speech wing of the speech movement,” but in 2015 added a new line to its Twitter Rules that barred “promot[ing] violence against others … on the basis of race, ethnicity, national origin, religion, sexual orientation, gender, gender identity, age, or disability.” Any danger that the platforms might abuse their power has been airily dismissed with a promise that all problems can be addressed simply by making “changes to the architecture and governance systems put in place by these platforms.”19 Yet platforms’ content moderation guidelines are not public, only their more nebulous “community standards.”
The network platforms have been left blissfully unmolested by the Federal Communications Commission, even as they have moved into direct competition with the television radio stations that it regulates and the telecommunications firms that provide most consumers with access to the Internet. The Federal Election Commission until recently considered digital platforms exempt from disclosure rules on political ads. In 2011 Facebook asked the FEC for an exemption to rules requiring the source of funding for political ads to be disclosed, arguing that the agency “should not stand in the way of innovation.”20 The only U.S. regulator that Silicon Valley has had to contend with is the Federal Trade Commission, which has powers to enforce consumer-protection laws. In 2011 the FTC cited Facebook for “engaging in unfair and deceptive practices” with regard to the privacy of user data. The company signed a consent decree pledging to establish a “comprehensive privacy program” and to evaluate it every other year for twenty years—a commitment it seems to have honored mainly in the breach, as we shall see.
In short, to characterize the U.S. regulation of network platforms as laissez faire or “light touch” would be a considerable understatement. The current system is, in essence, one of self-regulation—or, to be more precise, self-non-regulation.
The Political Consequences
The consequences have been profound. The first and best known has been to favor fake news. It has been claimed that social media mainly tend to amplify the content produced by traditional media.21 But that overlooks two things. First, if network platforms choose to promote a message of their own—for example that users should consider organ donation for transplants—the results are far more impressive than when newspapers or television channels make similar appeals.22 Second, social media also disseminate fake news, which traditional media tend not to do. Unfortunately, it appears that false information on Twitter is typically retweeted by many more people, and far more rapidly, than true information, especially when the topic is politics. Researchers at the Massachusetts Institute of Technology tracked 126,000 stories—some true, some false—tweeted by roughly three million people more than 4.5 million times from 2006 through 2017. They then used six different fact-checking sites—including Snopes, Politifact, and FactCheck.org—to rate the truthfulness of each story. It turned out that false claims were 70 per cent more likely than the truth to be shared on Twitter. True stories were rarely retweeted by more than a thousand people, but the top one per cent of false stories were routinely shared by between 1,000 and 100,000 people. And it took true stories about six times as long as false ones to reach 1,500 people. No accurate news item was able to chain together more than ten retweets, whereas fake news could put together a retweet chain 19 links long—and do it ten times as fast as the accurate news item put together its ten retweets. This finding is especially startling, as Twitter users who share accurate information typically have more followers, and send more tweets, than fake-news sharers.23 An important role is evidently played in the dissemination of fake news by “bots”—automated accounts purporting to be humans—who are believed to account for between 9 and 15 per cent of active Twitter accounts and as many as 60 million Facebook users.24 In the words of a recent large-scale study by the Knight Foundation, “A supercluster of densely interlinked, heavily followed accounts plays a large role in the spread of fake news and disinformation on Twitter. Social bots likely make up the majority of the accounts in the supercluster, and accounts in the cluster participate in what appear to be coordinated campaigns to push fake news stories.”25
A second consequence of leaving the network platforms to their own devices has been polarization. Homophily—the tendency of birds of a feather to flock together—has long been a recognized feature of social networks, even those of modest size. Giant online networks were therefore always likely to self-segregate into more or less homogeneous clusters. This was true of the “blogosphere,” for example, prior to the ascendancy of the network platforms.26 We are innately inclined to form into opposing sides over any bone of contention, as the case of “The Dress”—a photograph of a dress taken in England at 3:30pm on a February afternoon—illustrates. (For days the Internet was rent asunder: Was the dress black and blue or white and yellow?)27 Yet the network platforms do more than merely reveal our innate divisions. Because their algorithms are designed to maximize our engagement and, because confirmation bias is one of our many cognitive biases, they tend to accentuate it.
A good illustration of the point is the way that Twitter works. If the network of political retweets is graphed using a force-directed algorithm, two highly segregated communities of users are revealed: liberals and conservatives.28 A similar depiction of the retweet activity of messages containing moral and emotional language on a range of political topics (gun control, same-sex marriage, climate change) looks very similar. The presence in tweets of words that researchers classified as “moral-emotional” increased their diffusion by a factor of 20 per cent for each additional word. Moreover, “moral contagion” was bounded by group membership, in that “moral-emotional language increased diffusion more strongly within liberal and conservative networks, and less between them.”29 Another paper, based on 3,938 Twitter users who together generated 4.8 million tweets in August 2016, suggested that it is politically extreme people who tweet about politics more than centrist users of Twitter.30 In Congress, more ideologically extreme lawmakers get more Facebook followers than centrists.31
This is not to say that people consciously choose to inhabit filter bubbles or echo chambers. A 2015 study found that “most social media users [were] embedded in ideologically diverse networks, and that exposure to political diversity has a positive effect on political moderation.” Nor is it to blame all polarization on social media. The division between liberals and conservatives has deep historical roots. Its reflection in the media—from newspapers to cable television—is not new. And the increase in polarization in recent years seems to have been greatest amongst the elderly, the group least likely to use the internet and social media.32 Yet the network platforms are clearly making polarization worse because of the way they work. A good example is the way the YouTube suggestions algorithm works. In the words of former Google engineer Guillaume Chaslot, “Videos about vegetarianism led to videos about veganism. Videos about jogging led to videos about running ultramarathons. Given its billion or so users, YouTube may be one of the most powerful radicalizing instruments of the 21st century. ... YouTube leads viewers down a rabbit hole of extremism, while Google racks up the ad sales.”33 YouTube assists extremists in other ways, notably by placing advertisements for respectable corporations and other organizations on the websites of Nazis and so-called white nationalists. YouTube channels with over 1,000 subscribers and 4,000 watch-hours over a twelve-month period can apply to receive a portion of YouTube’s revenue from the advertisements running on their videos. Until it was shut down, Brian Ruhe’s Nazi channel featured ads—and therefore earned money—from Nissan, Disney, Mozilla and 20th Century Fox.34 There is also evidence from the 2015 protests against police violence in Baltimore that the degree and volume of moral rhetoric used on social media predicted the number of arrests during protests.35 The “culture war” in the United States—over campus free speech, climate change, healthcare reform, racism, gender fluidity, and sexual harassment—would no doubt be happening without social media. But the network platforms have surely intensified the conflict.
The elections of 2016 brought to light the full extent of the network platforms’ role in the modern public sphere. Like the culture war, the elections would no doubt have been bitterly contested without social media. But the network platforms in 2016 did more than merely intensify the contests. They decisively influenced the results.
There is of course nothing new about a change in the structure of the public sphere having political consequences. Each new communications technology—the newspaper, the telegraph, the radio, the television—has had its effect on the political process. William Randolph Hearst was first a hero and then a pariah for his accumulation of mass-circulation newspapers and magazines, each of which toed his political line. Most recently, cable news stations such as CNN and Fox have had a measurable impact on American elections. According to one study, biased coverage by Fox News during the 2000 presidential election was enough to shift 10,757 votes in Florida, where George W. Bush’s decisive margin of victory was just 537 votes.36 Social media played a modest role in the 2008 election because at that time Facebook and Twitter were still in their infancy, though there is no question that Barack Obama’s use of them was far superior to John McCain’s. But they were more important in the 2012 election. This shift had its roots in 2009, when Dan Wagner began work on a sophisticated model (known as the Survey Manager), which before long was accurately forecasting the “shellacking” inflicted on the Democrats by Tea Party Republicans in the 2010 midterms. At first, much of the data Wagner used were from surveys he and his team carried out. But by 2012 the Democratic National Committee had invested heavily in technology designed to integrate as much information as possible about individual voters. The Obama campaign spent twice as much on online advertisements as Mitt Romney’s. Its approach to television advertising was based on far superior data on the viewing habits of potentially persuadable voters. The Romney campaign mostly out-sourced its data operation; in effect it relied on the technology and methods Obama had used in 2008.37 In 2012, by contrast, the Obama campaign was able to create a voter-outreach app that analyzed users’ Facebook connections and encouraged users to reach out to potential Obama supporters among their Facebook friends. Google also made its data tools (Google Analytics) available to Obama’s reelection campaign.
Google is first and foremost a search engine. So well established is the company’s reputation for fast and efficient ranking of web pages that most users assume more or less uncritically that its search results are an objective measure of citation frequency, if not credibility. That is why nine out of ten clicks are on the first page of Google search results. In fact, the search engine can be manipulated and with significant effects. Three experiments relating to the 2010 Australian election found that the influence of a Google-like search engine on voter behavior was very great. After subjects were left to conduct their own search-based web research, seeing results that were biased by design, the likelihood of their voting for a particular candidate diverged from their initial preferences by between two and four percentage points. The authors term this differential “voter manipulation power” or VMP. Between three quarters and all of the subjects in these experiments showed no awareness of the manipulation. A similar experiment with U.S. voters had similar results. The VMP of voters in the 2014 Lok Sabha Elections in India was 9.4 per cent and as much as 73 per cent amongst unemployed males from Kerala. The authors of a pioneering paper on the “search engine manipulation effect” concluded in 2015 that “If a search engine company optimized rankings continuously and sent customized rankings only to vulnerable undecided voters, there is no telling how high the VMP could be pushed. … Search results favoring one candidate could easily shift the opinions and voting preferences of real voters in real elections by up to 80 percent in some demographic groups with virtually no one knowing they had been manipulated.”38 The same author’s estimated that “Google’s search engine—with or without any deliberate planning by Google employees—was currently determining the outcomes of upwards of 25 percent of the world’s national elections … because Google’s search engine lacks any kind of equal-time rule, so it virtually always favors one candidate over another.”39 Comparably powerful is the “Search Suggestion Effect,” whereby Google search suggestions (“autocomplete” suggestions) have the power to shift opinions and voting preferences:
Negative (“low valence”) search terms can attract 10-to-15 times as many clicks as neutral or positive terms can (an example of “negativity bias”), which means that a simple yet powerful way for a search engine company to manipulate elections is to suppress negative search suggestions for the candidate it supports, while allowing one or more negative search suggestions to appear for the opposing candidate (the “differential suppression of negative search suggestions”) … [T]he higher a suggestion appears in a list of search suggestions, the more impact it has on search, and overall, manipulating search suggestions can shift a 50/50 split among people who are undecided on an issue to a 90/10 split without people’s awareness and without leaving a paper trail for authorities to follow.40
Nearly everything that happened in 2016 had a pre-history. Cambridge Analytica was established in June 2014 by Robert and Rebekah Mercer, who had previously invested in Alexander Nix’s Strategic Communications Laboratories (SCL) Group. The new company was set up as a joint venture with SCL; its name was chosen by Stephen K. Bannon, executive chairman of the right-wing website Breitbart, who became a director. The company’s voter database was purchased from the Cambridge University researcher Aleksandr Kogan, whose “this is your digital life” app represented itself as a research tool used by academic psychologists. Its online questionnaire was hosted by a company called Qualtrics. Respondents were asked to authorize access to their Facebook profiles and, when they did, Kogan’s app harvested their data as well as the data of all their Facebook friends—not only their names, birth dates and location data, but also lists of every Facebook page they had ever liked. The small print accompanying Kogan’s questionnaire told users that their data could be used for commercial purposes, a violation of Facebook’s rules at the time. In 2015, Facebook said that, having learned that Kogan had passed user data to Cambridge Analytica, it had excluded him from access to Facebook data and demanded assurances that the data had been deleted. But the company did not disclose to users—or anyone else—that their data had been misused. Moreover, in November 2015, Facebook brought Kogan in as a consultant to explain the technique he had used.41 In March 2018, when Facebook was forced by news revelations to comment, the company insisted there had not been a data breach but merely “a scam—and a fraud.” In all, some 270,000 people downloaded Kogan’s app. That was enough to give him access to the data of 87 million Facebook users, 71 million of whom were Americans. Around 30 million profiles contained enough information, including places of residence, to enable Cambridge Analytica to build the psychographic profiles that were its main selling point.42 In 2014 the company did work for Texas Senator Ted Cruz, the John Bolton Super PAC, conservative groups in Colorado and the campaign of Senator Thom Tillis, the North Carolina Republican.43
To date, the most controversial aspect of social media’s involvement in the 2016 U.S. election was the way Russian entities, principally the Internet Research Agency (IRA), disseminated inflammatory content through Twitter and Facebook. This, too, had a prelude. According to special counsel Robert Mueller’s indictment of 25 Russian individuals (12 of them identified as Russian military intelligence agents), Moscow began building a U.S. influence operation in 2014. It had already experimented with “information warfare” in a number of countries, including Ukraine, which Russian forces invaded in 2014. During the UK referendum on membership of the European Union, at least 2,752 Twitter profiles appear to have been created and managed by the IRA. Russian content found itself into mainstream media on numerous occasions: 29 different Russian-run accounts were quoted across 73 different news stories. The Daily Telegraph embedded posts from Russian accounts such as @Ten_GOP and @Pamela_Moore13 15 times, for example. Another Russian account, @WarfareWW, accounted for four of the Daily Mail’s seven citations. A recent NBER study estimates that automated tweets added 1.76 percentage point to the “Leave” share of the vote. There was a “massive volume” of tweets from Russian-language accounts in the few days before the referendum.44 When, in October 2018, Twitter released datasets of more than 10 million tweets and more than 2 million images and videos from 3,841 IRA-affiliated (plus 770 others suspected of being Iranian), the material dated back as far as 2009. 45
In short, there was ample reason to expect the network platforms to play both an important and a subversive role in the U.S. election of 2016. And yet almost no precautions appear to have been taken by either the authorities or the big technology companies themselves. On the contrary, Facebook executives continued to repeat the company’s foundational mantra: “Move fast and break things.” On June 18, 2016, a day after the fatal shooting of a Chicago man was captured on Facebook Live, the company’s vice president Andrew “Boz” Bosworth, circulated an internal memorandum that epitomized a culture of indifference to negative externalities:
So we connect more people.
That can be bad if they make it negative. Maybe it costs a life by exposing someone to bullies. Maybe someone dies in a terrorist attack coordinated on our tools.
And still we connect people.
The ugly truth is that we believe in connecting people so deeply that anything that allows us to connect more people more often is *de facto* good. It is perhaps the only area where the metrics do tell the true story as far as we are concerned.
That isn’t something we are doing for ourselves. Or for our stock price (ha!). It is literally just what we do. We connect people. Period. …
… I know a lot of people don’t want to hear this. Most of us have the luxury of working in the warm glow of building products consumers love. But make no mistake, growth tactics are how we got here. …
That’s our imperative. Because that’s what we do. We connect people.
What Exactly Happened in 2016?
Since the November 2016 election, Facebook has admitted that the Internet Research Agency used up to 470 false identities and spent $100,000 on more than 3,000 Facebook and Instagram ads to spread politically divisive posts to Americans before and after the vote. The ads promoted about 120 Facebook Pages set up by the IRA and related groups. These Pages posted 80,000 pieces of content between January 2015 and August 2017, reaching up to 126 million Americans—a number only slightly smaller than the total number who voted. In all, an estimated 146-150 million users saw posts from accounts linked to the IRA, including around 16 million Instagram users. Jonathan Albright has suggested that posts on just six publicly known Russia-linked Pages were shared 340 million times. In addition, Russians with false identities used Facebook Events to promote political protests, including an August 27 anti-immigrant, anti-Muslim rally in a rural Idaho town that was known to welcome refugees. The event was “hosted” by “SecuredBorders,” a phony anti-immigration group that was in fact a Russian front. (It had 133,000 followers when Facebook closed it down.) “Heart of Texas,” a Russian-controlled Facebook group that promoted Texas secession, announced a rally to “Stop Islamification of Texas” in front of the Islamic Da’wah Center of Houston on May 21, 2016. A separate Russian-sponsored group, “United Muslims of America,” advertised a “Save Islamic Knowledge” rally for exactly the same place and time. The Russians did not confine themselves to Facebook. Twitter has admitted that Russian bots tweeted 2.1 million times before the election. So far, 2,700 Twitter accounts have been identified as IRA-run. Of these, 65—notably @WarfareWW, @TEN_GOP and @Jenn_Abrams—featured prominently in a large-scale study of the fake news phenomenon.46
How big an impact did this Russian effort have? Political scientist Kathleen Hall Jamieson goes so far as to argue that the election was in fact decided by Russian content aimed at discouraging potential Clinton voters from turning out, as this was the key variable that tipped Michigan, Pennsylvania and Wisconsin over to Trump.47 It is important, however, to recognize that the impact of Russian was not simply in one, pro-Trump direction. A study of retweet activity reveals, surprisingly, that liberals were more inclined to disseminate this content than conservatives.48 Another analysis of the Twitter network suggests that “clusters of accounts affiliated with Russia serve a brokerage role, serving as a cultural and political bridge between liberal U.S. accounts and European far-right accounts.” 49 Moreover, the Russians only accounted for a part of the fake news disseminated by foreign actors during the election. Buzzfeed traced a hundred pro-Trump sites to a small town in Macedonia. At least one of these was run by the Israeli private intelligence firm Psy-Group. The Iranians also appear to have abused Twitter.50
Yet the key point is that all this foreign-originated content was a drop in the ocean. Between March 23, 2015, and November 2016, an estimated 128 million people in America were responsible for nearly 10 billion Facebook posts, shares, likes, and comments about the election. Hindman and Barash identified 6.6 million tweets or retweets from 454,832 separate accounts that linked to at least one of more than 600 fake news or conspiracy websites during the month before the election. The tweets of Russian operatives and bots represented perhaps one per cent of all electionrelated tweets. The same goes for fake news from any source was only a fraction of total election-related content. True, according to one estimate, the average American encountered between one and three stories from known publishers of fake news during the month before the 2016 election.51 Another study of the browsing histories of 2,525 adult Americans during the run‑up to the 2016 election suggested that one in four Americans saw at least one false story, with the most conservative 10 per cent of the sample accounting for two thirds of visits to fake news sites. Yet false stories were a very small proportion of the total news people consumed, accounting for just 1 per cent of the news intake of Clinton supporters and 6 per cent in the case of Trump supporters.52
It could be argued that even that small a proportion of fake news—foreign and home-grown—might have sufficed to decide a very tight election. Research on the impact of traditional television advertising between 2004 and 2012 concluded that partisan imbalances in advertising had significant effects. (Exposure to an additional ad by one party or the other shifted the partisan vote of approximately two people out of 10,000.)53 Moreover, false election stories tended to attract more attention than true ones. In August 2015, a rumor circulated on social media that Donald Trump had let a sick child use his plane to get urgent medical care. Snopes confirmed almost all of the story as true. However, only about 1,300 people shared or retweeted the story. In February 2016, by contrast, a rumor circulated that Trump’s elderly cousin had recently died and that he had left a message publicly condemning his relative’s presidential bid. Snopes rejected this story as false, but around 38,000 Twitter users shared it. Its retweet chain was three times longer than the one produced by the true sick child story. A false story alleging that the boxer Floyd Mayweather had worn a Muslim head scarf to a Trump rally also reached an audience more than ten times the size of the sick child story.
However, the Russians had real news, too, which they had obtained by means deemed legitimate by the mainstream media outlets that published it. Beginning in March 2016, Russian hackers Fancy Bear sought to phish their way into the Democratic party’s emails, sending fake emails that seemed to come from Google and told recipients to change their passwords. Chairman John Podesta fell for the ruse on March 19, clicking on the fatal link and giving the Russians access to 50,000 emails. Former Trump foreign policy adviser George Papadopoulos said that he was told on April 26 by an academic closely connected to the Kremlin that the Russian government had obtained compromising information about Clinton. “They have dirt on her,” Papadopoulos was told. “They have thousands of emails.” On June 12 WikiLeaks founder Julian Assange told a British TV show that emails related to Clinton were “pending publication.” On June 15 the supposedly Romanian (but in reality Russian) Guccifer 2.0 claimed to have hacked the DNC and began directing reporters to the newly launched DCLeaks site. WikiLeaks joined in on July 22. Guccifer 2.0, WikiLeaks and DCLeaks ultimately published more than 150,000 emails stolen from more than a dozen Democrats. Long-established newspapers did as much as network platforms to disseminate the leaked content. Hillary Clinton herself attaches more blame for her defeat to FBI James Comey’s intervention on October 28—when he told Congress that he was reopening an investigation into her private email server—than to Russian disinformation.
Another point often overlooked is that not all online manipulation during the campaign was calculated to hurt Clinton. True, YouTube tilted its users towards Trump. Guillaume Chaslot discovered that, regardless of whether users started with a pro‑Clinton or a pro‑Trump video, they were many times more likely to have a pro‑Trump video recommended by YouTube. Twitter also helped Trump as most Twitter bots (not only the Russian ones) favored him.54 However, there is striking evidence that the Search Engine Suggestion Effect was exploited by Google in Clinton’s favor. Search suggestions on Google, Yahoo and Bing on August 3, 2018, differed strikingly. The user who consulted Yahoo was promoted to search for “Hillary Clinton is a liar” or “Hillary Clinton is a criminal.” On Bing the top two suggestions were “Hillary Clinton is a filthy liar” and “Hillary Clinton is a murderess.” But the user who typed “Hillary Clinton is” into Google’s search box was prompted to search for “Hillary Clinton is winning” or “Hillary Clinton is awesome.” 55 Robert Epstein argues that Google search rankings favored Clinton over most of the six-month period that he and his research associates monitored. Between October 15 and Election Day, “search rankings favored Mrs. Clinton in all 10 of the search positions on the first page of search results.”56
Yet the truly decisive factor in the 2016 was probably none of the above. It was the different ways the Trump and Clinton campaigns themselves used the network platforms, particularly Facebook. According to the New Yorker, Facebook “offered to ‘embed’ employees, for free, in presidential campaign offices to help them use the platform effectively. Clinton’s campaign said no. Trump’s said yes.” Trump used Facebook to raise $280 million. More than a third of that money—vastly more than the Russians spent—went to pay for targeted Facebook ads. These included a voter-suppression drive in the days before the election, targeting “idealistic white liberals, young women, and African Americans.” Theresa Hong, the Trump campaign’s digital-content director, later told an interviewer, “Without Facebook we wouldn’t have won.” 57 Her verdict has been echoed by Gary Coby, the director of advertising at the Republican National Committee and director of digital advertising and fundraising for Trump’s campaign, and by Brad Parscale in an interview with Lesley Stahl for 60 Minutes and in a conversation with Michael Isikoff in Lisbon last November. Ali-Jae Henke, the head of elections at Google, explained how his company also provided help to both campaigns:
They say like “look, we really want to get attention and we want to reach as many people as possible and these are kind of the areas politically where we might have challenges or the different types of voting blocs we need to reach” … and so then I am able to in that advisory capacity be like, “well this is what moms look like online, this is how we find them ...”
Parscale explained the role of Cambridge Analytica in this process:
[They] didn’t play a role in crafting ads [but] helped with a research strategy to help us raise money. We needed to build an infrastructure. [Cambridge Analytica] provided staff, resources, because we had to grow a large organization, fast. They did a lot of polling, and they did a lot of building some directional arrows for us [regarding] where to place the money, being able to provide reporting back that says, “Here are trends that are happening,” so I could move the budget around in a way and I could make recommendations to [then candidate Trump] and to leadership, saying, “Here’s an opportunity. We should go into this part of Michigan. We should go into this part of Wisconsin.” [Cambridge was] able to drive that kind of information … and in a simple consumption model, daily.58
Also important was the way Facebook directed users to content on the Breitbart website, run since March 2012 by Steve Bannon. On August 17, 2016, Bannon was appointed chief executive of Trump’s presidential campaign. “I wouldn’t have come aboard, even for Trump,” he later said, “if I hadn’t known they were building this massive Facebook and data engine. Facebook is what propelled Breitbart to a massive audience. We know its power.”
This was the crucial difference between 2012 and 2016. By the time of the later election, Facebook had acquired a database of American voters far superior to anything either party could possibly have built on its own. It had done so not only by persuading a majority of Americans to join Facebook, but also by (among other things) logging the phone call and messaging histories of Android smartphone users who installed Messenger or Facebook Lite and then synced their phone contacts with the app.59 Facebook had also paid third-party websites and apps to let it place cookies, invisible pixels, “like” and “share” buttons on them, thereby acquiring data on people who are not Facebook users.60 Facebook’s approach to data-gathering was even more ruthless than Google’s.61 Its attitude towards how that data got used by third parties was, at best, cavalier. And only one campaign made full use of Facebook’s data. We may ask counterfactual questions to our heart’s content about how the election would have turned out if the Russians had played no role. Yet the crucial point is that Russian meddling was a subplot in a much bigger crisis of the American political system produced by the unregulated and generally reckless operation of the network platforms. Without the Russians, Trump might not have won. It would still have been close. Without Facebook, he would have stood no chance.
That there would be a backlash against the network platforms after their role in the 2016 election was easily predictable.62 A few writers—such as Jonathan Zittrain and Tim Wu—had been warning about their growing power for some time. Established content publishers such as Rupert Murdoch and Michael Bloomberg had obvious commercial reasons for going on the offensive. For Wu, Facebook was like a television network but with no “sense of responsibility. No constraints. No regulation. No oversight. Nothing.” Robert Thomson of News Corp talked about “tech tapeworms in the intestines of the internet.” Speaking at Davos in January 2017, George Soros warned of “a web of totalitarian control the likes of which not even Aldous Huxley or George Orwell could have imagined,” and called on European Union Competition Commissioner Margrethe Vestager to be the “nemesis” of the network platforms.63
The most credible critics have been the insiders—former Facebook employees such as Antonio Garcia Martinez, author of Chaos Monkeys, or Sandy Parakilas, a former operations manager, who publicly criticized the company’s handling of privacy issues in 2017, warning that “The company won’t protect us by itself, and nothing less than our democracy is at stake.”64 Former vice president for user growth, Chamath Palihapitiya, told an audience at Stanford’s Graduate School of Business: “I think we have created tools that are ripping apart the social fabric of how society works. … The short-term, dopamine-driven feedback loops that we have created are destroying how society works. No civil discourse, no co-operation: misinformation, mistrust.” He felt “tremendous guilt” about his own part in this because, deep down, he and his former colleagues “kind of knew something bad would happen.” In a similar vein, Facebook’s first president Sean Parker admitted that the platform was consciously designed to take advantage of “a vulnerability in human psychology” by delivering “a little dopamine hit every once in a while.” Parakilas joined forces with Dave Morin, Justin Rosenstein and Roger McNameee—all former Facebook employees or early investors—as well as Tristan Harris and Lynn Fox (both ex-Google) to establish the Center for Human Technology and launch a campaign with the title: “The Truth about Tech.”
Negative publicity in the wake of the election has led to a decline in public trust in Facebook and Twitter.65 Facebook also now lags quite far behind Amazon, Google, Apple and Microsoft in the Small Business Trust Index. A Gallup Knight survey published in January 2018 showed that 57 per cent of Americans regarded the way tech companies chose which stories to show to users as “a major problem” for democracy, while 73 per cent said the same about the spread of inaccurate information on the Internet.66 At that time only 49 per cent favored regulation of how websites provide news. By February 2018, however, the proportion “concerned that government would do too little” to address the problem had risen from 40 to 55 per cent.67 Young Americans in particular have lost trust in Facebook (though they trust it more than Washington and Wall Street).68 The Economist summed up the prevailing mood: the big tech companies were “too big, anti-competitive, addictive and destructive to democracy”—BAADD. 69
At Facebook’s second annual Social Good Forum in December 2017, Mark Zuckerberg described how his company uses artificial intelligence to identify users who might be contemplating self-harm or suicide. He did not discuss the possibility that Facebook might itself be driving people to self-harm or suicide. As Deborah M. Gordon has suggested, online social networks replicate on a vast scale many of the more insidious features of friendship circles amongst girls in a middle school.70 Using data from 5,208 adults over two years from a national longitudinal panel, Holly Shakya and Nicholas Christakis argue that “the more you use Facebook, the worse you feel.”71 They found that “most measures of Facebook use in one year predicted a decrease in mental health in a later year … [B]oth liking others’ content and clicking links significantly predicted a subsequent reduction in self-reported physical health, mental health, and life satisfaction.” The authors suggest that use of social media gives the impression of “meaningful social interaction” but is in fact no substitute for the real thing and therefore undermines wellbeing and health. Even Facebook’s own research comes to similar conclusions about the effects of overuse of social media by students.72
Especially troubling is the effect of social media on children. As James Bridle has shown, Kid’s YouTube seems designed to lure young users towards disturbing videos with titles like: “Surprise Play Doh Eggs Peppa Pig Stamper Cars Pocoyo Minecraft Smurfs Kinder Play Doh Sparkle Brilho” or “BURIED ALIVE Outdoor Playground Finger Family Song Nursery Rhymes Animation Education Learning Video.” A search for “Peppa Pig dentist” leads to a video on which Peppa Pig is “tortured, before turning into a series of Iron Man robots and performing the Learn Colours dance.” Others feature Peppa eating her father or drinking bleach.73 Facebook Messenger Kids, launched in 2017, seems calculated to introduce children under 13 to the app in order to get them hooked as early as possible. The analogies with cigarettes and corn syrup in the twentieth century and opioids in the twenty-first are not fanciful, as Marc Benioff of Salesforce and Aza Raskin of Mozilla and Firefox have acknowledged.74 There is some evidence that younger users of Facebook have been kicking the habit, but only in favor of Instagram (which Facebook owns) and Snapchat (which it would have liked to own).75
The revulsion against the power of the network platforms has not been confined to the United States. In Britain, Facebook has suffered a reputational hit, especially with older people. Although the evidence is less compelling that the company played a malign and decisive role in the Brexit referendum, journalists such as Carole Cadwalladr have done their utmost to make that case.76 There was controversy over the role of social media in the 2018 Irish referendum on abortion, too, as Facebook and Google restricted advertisements in moves widely interpreted to be helpful to the proponents of constitutional change. Did Facebook interfere in the 2017 election in Iceland by selectively displaying its “I Voted” button on some voters’ pages but not others?77 In India, fake news stories on WhatsApp service have triggered riots, lynchings, and fatal beatings. In Sri Lanka, after a Buddhist mob attacked Muslims over a false rumor, a presidential adviser put it nicely: “The germs are ours, but Facebook is the wind.” In Myanmar, too, violence against the Rohingya minority has been fueled, in part, by disinformation and incendiary content systematically spread on Facebook by the military. The United Nations investigator in charge of examining the persecution of the Rohingya, told the New Yorker, “I’m afraid that Facebook has now turned into a beast, and not what it was originally intended.”78 A Burmese legislator called the company “dangerous and harmful for our democratic transition.”79 It would be easy to give other examples. In Egypt, the Arab Spring was supposed to be a democratic revolution propelled by social media. Eight years later, Wael Ghonim—a leading figure in the Tahrir Square protests of 2010—is pessimistic. “We wanted democracy,” he has said, “but got mobocracy.” It seems unlikely that Kenyan democracy benefited from Cambridge Analytica’s work for Uhuru Kenyatta in the 2013 and 2017 Kenyan elections. According to Freedom House, online manipulation and disinformation tactics played an important role in elections in 18 countries in 2016. In undemocratic regimes, too, the manipulation of social media is now standard practice, from China to Saudi Arabia.
Yet the analogies offered by the critics of the network platforms are not consistent. As one journalist complained with respect to Facebook:
I’ve heard government metaphors (a state, the E.U., the Catholic Church, Star Trek’s United Federation of Planets) and business ones (a railroad company, a mall); physical metaphors (a town square, an interstate highway, an electrical grid) and economic ones (a Special Economic Zone, Gosplan). For every direct comparison, there was an equally elaborate one: a faceless Elder God. A conquering alien fleet. … Maybe Facebook is a church and Zuckerberg is offering his benedictions. Maybe Facebook is a state within a state and Zuckerberg is inspecting its boundaries. Maybe Facebook is an emerging political community and Zuckerberg is cultivating his constituents. Maybe Facebook is a surveillance state and Zuckerberg a dictator undertaking a propaganda tour. Maybe Facebook is a dual power—a network overlaid across the United States, parallel to and in competition with the government to fulfill civic functions—and Zuckerberg is securing his command. Maybe Facebook is border control between the analog and the digital and Zuckerberg is inspecting one side for holes. Maybe Facebook is a fleet of alien spaceships that have colonized the globe and Zuckerberg is the viceroy trying to win over his new subjects.80
This kind of muddle helps explain the inconsistencies in the global debate on regulation.
The European Commission, as George Soros foresaw, has taken the lead in seeking to regulate the U.S.-based network platforms. In June 2017, the Commission’s antitrust division fined Google $2.7 billion for “anticompetitive practices” related to Google Shopping, the company’s product comparison tool (specifically, for favoring the company’s own site over competitors). The European Union’s General Data Protection Regulation (GDPR), which came into force in May 2018, requires online services to make it easier for customers to transfer their information to other providers and even competitors, as well as strengthening people’s control over their data. At the same time, the European authorities have also been active in regulating online “hate speech.” In May 2016, Facebook, Microsoft, Twitter, and YouTube signed an agreement with the European Commission to “prohibit the promotion of incitement to violence and hateful conduct” by removing content inciting violence or hatred against protected groups within 24 hours of its being posted. On December 5, 2016, the companies announced plans for an industry database of “hashes”—unique digital signatures—of all extremist material banned on their platforms. National governments have added to this pressure. In 2017 Germany passed a law threatening Facebook, Twitter and other social media companies with fines of $50 million if they failed to give users the option to complain about hate speech and fake news or refused to remove illegal content within 24 hours. After a series of terrorist attacks in London in 2017, British Prime Minister Theresa May and French President Emmanuel Macron threatened to impose steep fines on companies that failed to remove extremist propaganda from online platforms. Shortly thereafter, Google announced a four-part plan to address terrorist propaganda that included the increased use of technology to identify terrorist-related videos, the hiring of additional content moderators, the removal of advertising on objectionable videos, and the directing of potential terrorist recruits to counter-radicalization videos.81
These regulations are seen by some as a model for the United States. Another way of looking at them is as a halfway house between the American laissez faire regime and the Chinese system of much stricter state control. An important difference between Europe and China is that, unlike the Europeans, the Chinese have succeeded in building their own technology giants: the online retail site Alibaba, the search engine Baidu and the Internet conglomerate Tencent. That these companies are subordinate to the Chinese state is clear. They are obliged to share their data with central authorities such as the People’s Bank of China (PBoC), which therefore has access to users’ payment history, creditworthiness and contacts. This universal “back door” into the data represents the first step towards a comprehensive system of “social credit.”82 On June 1, 2017, a new cybersecurity law came into effect in China that requires technology companies to help the authorities remove content that “endangers national security, national honor and interests.” The Chinese police are rapidly expanding their network of surveillance cameras and facial-recognition technology. “The political and legal system of the future is inseparable from the internet, inseparable from big data,” Alibaba’s Jack Ma told a Communist Party commission overseeing law enforcement in 2017. In future, he said, “Bad guys won’t even be able to walk into the square.”83 In the context of a one-party state, the combination of ubiquitous smartphones, network platforms, big data and artificial intelligence makes possible a precision-targeted totalitarianism beyond the dystopian visions of Orwell and Huxley. At the same time, the immense attractiveness of the rapidly growing Chinese market to Western tech companies makes them susceptible to pressure from the Chinese government. For example, Twitter joined Facebook and YouTube in restricting social media accounts of popular dissident Chinese businessman Guo Wengui, in response to pressure from Beijing.
Regulation of the Internet is on the increase. Nearly half of the 65 countries assessed in Freedom on the Net 2017 experienced declines in online freedom last year, while just 13 made gains, most of them minor. Less than one quarter of users reside in countries where the internet is designated “free.” Not only China, but also Venezuela, the Philippines, and Turkey were among 30 countries where governments were found to employ armies of “opinion shapers” to spread government views.84 Yet it is not self-evidently obvious which is more dangerous: a regulated Internet, in which governments exercise at least some control over network platforms, or an unregulated one, in which private companies continue to gather and exploit the personal data of citizens for profit and without scruple.
Facebook’s response to the criticism directed against it over the past two years has been unconvincing. “Personally,” declared Zuckerberg two days after the 2016 election, “I think the idea that fake news on Facebook, which is a very small amount of the content, influenced the election in any way is a pretty crazy idea.” His February 2017 manifesto, entitled “Building Global Community,” was long on aspirations, short on specifics. Four months later, Facebook unveiled a new mission statement to “give people the power to build community, to bring the world closer together.” The assertion that Facebook was engaged in community-building was disingenuous. The implication was that Facebook should not be held responsible, like a media company, for the content that appeared on its platform. “Things happened on our platform that shouldn’t have happened,” Sheryl Sandberg conceded. But “at our heart we’re a tech company. We hire engineers. We don’t hire reporters. No one is a journalist. We don’t cover the news.”85 A former senior employee explained:
The view at Facebook is that “we show people what they want to see and we do that based on what they tell us they want to see, and we judge that with data like time on the platform, how they click on links, what they like.” And they believe that to the extent that something flourishes or goes viral on Facebook—it’s not a reflection of the company’s role, but a reflection of what people want. And that deeply rational engineer’s view tends to absolve them of some of the responsibility, probably.86
This defense had lost credibility, however. The insistence that the content on Facebook was somehow not Facebook’s problem effectively collapsed in the wake of the revelations about 2016.
In the course of 2017, a new strategy evolved, which might be characterized preemptive self-regulation. On September 21, Zuckerberg pledged to increase the resources of Facebook’s security and election-integrity teams in order to work “proactively to strengthen the democratic process.” Facebook would henceforth require that all political ads disclose which Facebook page paid for them and ensure that every ad a given advertiser ran was accessible to anyone on the buyer’s page. Facebook would double the number of employees and contractors working on user safety and security issues to 20,000 by the end of 2018. It would also build new artificial-intelligence systems to detect what Zuckerberg described as “bad content and bad actors.” This could mean a significant increase in operating costs. But Zuckerberg told investors: “I am dead serious about this. I’ve directed our teams to invest so much in security on top of the other investments we’re making that it will significantly impact our profitability going forward.” These pledges amounted to an admission of responsibility for content.
Yet the practical consequences were confusing. In October 2017, Facebook introduced “Explore Feed,” which required media companies to pay for inclusion in the News Feed.87 Further changes were introduced in January 2018, when Zuckerberg announced that News Feed would prioritize “meaningful interaction” over “passive consumption of low-quality content,” demoting “things like clickbait headlines and false news, even though people often click on those links at a high rate.” He added: “We want to make sure that our products are not just fun, but are good for people … good for the world.” At the same time, Facebook would start to boost certain publishers whose content was “trustworthy, informative, and local,” according to reader surveys. In the course of 2018, Zuckerberg made numerous such announcements:
- In January Facebook hired Nathaniel Gleicher, the former director for cybersecurity policy on President Obama’s National Security Council, to counter “information operations.”
- In July, it removed thirty-two accounts running disinformation campaigns that were traced to Russia. A few weeks later, it removed more than six hundred and fifty accounts, groups, and pages with links to Russia or Iran.
- In March, Zuckerberg pledged “dramatically [to reduce] the amount of data that developers have access to, so that apps and developers can’t do what Kogan did” in providing data to Cambridge Analytica.88
- That same month, he said that “People should know who is buying the ads that they see on Facebook, and you should be able to go on any page and see all the ads that people are running to different audiences.” He said that this would be in place by the November 2018 midterms.89
- Also in March, Facebook announced that it would “try to make privacy settings clearer by creating a central hub where users can examine the data they are sharing” with third-party developers.”90
- In September he announced a “three-year project” to “rebuild all of our content enforcement systems to proactively find harmful content rather than wait for people to flag issues.” He added: “It is our responsibility to amplify the good and mitigate the bad.” 91
Similar measures of self-regulation have been announced by the other network platforms over the past eighteen months.
- In April 2017, Google announced the release of “Fact Check”: “For the first time, when you conduct a search on Google that returns an authoritative result containing fact checks for one or more public claims, you will see that information clearly on the search results page.” This was another sop to established publishers whose revenue Google and Facebook had been devouring.
- In October 2017 Google ended its “first click free” policy, which required publishers to give away some stories in order to appear high in its search rankings. Google chief executive Pichai spoke of “a flight to quality.”92
- In November 2017, Google announced the “Trust Project,” hosted by Santa Clara University and developed in conjunction with more than 75 news organizations worldwide. The goal of the project would be to introduce eight “trust indicators,” such as “author expertise,” “citations and references,” and “diverse voices,” into the search ranking algorithm.
- In December 2017, Google announced its intention to increase to 10,000 the number of employees tasked with removing extremist content from YouTube.
In short, the network platforms’ response to the crisis of 2016 has been a barrage of promises to regulate themselves. “I actually am not sure we shouldn’t be regulated,” Zuckerberg said in a CNN interview in March. “I actually think the question is more, what is the right regulation rather than ‘Yes or no, should it be regulated?’”93 He told Ezra Klein that he could imagine “some sort of structure, almost like a Supreme Court, that is made up of independent folks who don’t work for Facebook, who ultimately make the final judgment call on what should be acceptable speech in a community that reflects the social norms and values of people all around the world.”94 This was an unusual concession, as Zuckerberg—like his Roman role model Augustus—rarely offers to limit his own power. “One of the things that I feel really lucky we have,” he also told Klein, “is this company structure where, at the end of the day, it’s a controlled company. We are not at the whims of short-term shareholders. We can really design these products and decisions with what is going to be in the best interest of the community over time.” This was an allusion to the preference shares that give Zuckerberg control over Facebook. Advertising might be Facebook’s principal source of revenue, he conceded, but that was only so that his platform could be free to users. “I think probably to the dissatisfaction of our sales team here, I make all of our decisions based on what’s going to matter to our community and focus much less on the advertising side of the business.”
In addition to numerous interviews, the leaders of the big technology companies have set out the case for self-regulation in a succession of hearings: on October 31-November 1, 2017, in April 2018—when Zuckerberg himself testified before two congressional committees—and again in September 2018. “Which are you?” Senator Dan Sullivan (R-Alaska) asked Zuckerberg in April. “Are you a tech company, or are you the world’s largest publisher?”
Zuckerberg: I view us as a tech company, because the primary thing that we do is build technology and products.
Sullivan: You said you were responsible for your content, which makes you kind of a publisher, right?
Zuckerberg: I agree that we are responsible for the content, but we don’t produce the content. I think that when people ask us if we are a media company or a publisher, my understanding of what the heart of what they are really getting at, is “Do we feel responsibility for the content on our platform?” And the answer to that, I think, is clearly “Yes.”
This recognition of responsibility for content was an important moment in the April hearings. Also important, however, was Zuckerberg’s assertion that Facebook is “a system of different things: we compete with Twitter as a broadcast medium; we compete with Snapchat as a broadcast medium; we do messaging, and iMessage is default-installed on every iPhone.” This was part of a wider argument he sought to make against any attempt to apply antitrust law to Facebook. “I think that anything that we’re doing to constrain [the big tech companies] will, first, have an impact on how successful we can be in other places,” Zuckerberg also said. “I wouldn’t worry in the near term about Chinese companies or anyone else winning in the United States, for the most part. But there are all these places where there are day-to-day more competitive situations—in Southeast Asia, across Europe, Latin America, lots of different places.”95 In other words, what is good for Facebook is good for America, and vice versa—as General Motors CEO Charles Wilson famously said of GM in 1953. Regulate big tech at your peril, for you may unintentionally help the Chinese competition.
The scale of the self-defense effort by Silicon Valley since 2016 cannot be understate. Facebook has followed Google’s lead by investing heavily in Washington lobbying, hiring Joel Kaplan, a former policy adviser to President George W. Bush, and Sandy Luff, a former aide to Attorney General Jeff Sessions. According to data from the Center for Responsive Politics, Facebook had contributed a total of $641,685 since 2014 to the members of Congress that Zuckerberg faced during his visit to Capitol Hill. The top recipients of that money included Senators Cory Booker and Kamala Harris.
However, these efforts could not prevent a flurry of activity by legislators and regulators. In August 2017, an unusual bipartisan alliance of 27 Democratic and Republican senators introduced the Stop Enabling Sex Traffickers Act (SESTA), despite opposition from the Internet Association and Google. It was passed, along with the Allow States and Victims to Fight Online Sex Trafficking Act (FOSTA), in April 2018. In October Senators Amy Klobuchar, Mark Warner and John McCain introduced the Honest Ads Act (S. 1989), which would require the network platforms to reveal the buyers and content of all campaign-related ads as well as to maintain a public list of all political advertisers spending $500 or more. Other legislative initiatives of the past year include Keith Ellison’s 21st Century Competition Commission Act, an antitrust bill; Senator Edward J. Markey’s bill to commission research on the impact of technology on children’s health, and California measure to prohibit the use of bots without identification (the so-called “Blade Runner” law SB-1001). Senator Amy Klobuchar has also proposed two bills to change the criteria for mergers. At the same time, the Federal Trade Commission (FTC) is conducting an “open non-public investigation” into whether or not Facebook violated the terms of the 2011 consent decree when users’ personal data were made available to Kogan and Cambridge Analytica. In 2012 the FTC fined Google $22.5 million for violating the terms of another consent decree. It remains to be seen if Facebook will be penalized more or less harshly. Facebook is also under investigation by the Federal Bureau of Investigation, the Securities and Exchange Commission and the Department of Justice. Also investigating Facebook are the attorney generals of New York, Massachusetts and Connecticut.
Not to be outdone, the president himself has begun to use Twitter to attack the other big tech companies. “Facebook was always anti-Trump,” he tweeted in October 2017. “The Networks were always anti-Trump hence,Fake News, @nytimes(apologized) & @WaPo were anti-Trump. Collusion?” (“That’s what running a platform for all ideas looks like,” Zuckerberg responded in a Facebook post.) In 2018, Trump attacked Amazon for tax evasion and taking advantage of the U.S. Postal Service. On April 5, Trump told reporters that “Amazon is just not on an even playing field. You know, they have a tremendous lobbying effort, in addition to having The Washington Post [which Jeff Bezos acquired in 2013], which is, as far as I’m concerned, another lobbyist.” He was, he said, “going to take a pretty serious look” at Amazon because “the playing field has to be leveled.” According to an unnamed source who had spoken to Trump, the president had “wondered aloud if there may be any way to go after Amazon with antitrust or competition law.”96 Even Trump’s preferred platform, Twitter was not spared. On July 26, 2018, he accused Twitter of “shadow banning” Republicans and said he would look into what he called a “discriminatory and illegal practice.” The big tech companies “better be careful because you can’t do that to people,” Trump said in August, following the expulsion from Facebook of far-right conspiracy theorist Alex Jones. “I think that Google, and Twitter and Facebook, they are really treading on very, very troubled territory and they have to be careful. It is not fair to large portions of the population.” Google, too, came under attack. In August 28, 2018, Larry Kudlow, Trump’s economic adviser, said that the administration was “taking a look” at whether or not Google and its search engine should be regulated by the government. Trump complained that “Fake CNN” was “prominent” in search results for him. “They have it RIGGED, for me & others,” he complained, “so that almost all stories & news is BAD.” Conservative media were being “shut out.” Was this “illegal?” he asked, accusing Google of “controlling what we can & cannot see.” In a tweet on August 18, Trump accused social media of “totally discriminating against Republican/Conservative voices. … They are closing down the opinions of many people on the RIGHT, while at the same time doing nothing to others.”
The only surprising thing is that this kind of complaint was not made sooner. FEC disclosures from last year’s presidential campaign showed that 95 per cent of tech employee donations went to Hillary Clinton, and only 4 per cent to Donald Trump. The liberal politics of Silicon Valley was regularly on display in 2017, beginning as early as January, when thousands of Google employees walked out of the office for a rally sanctioned by the company—and indeed addressed by chief executive Sundar Pichai—to protest against Trump’s executive order banning travel from seven predominantly Muslim countries. Google’s firing of James Damore, for writing an in-house essay that questioned the desirability of hiring and promoting more female engineers, seemed to furnish fresh evidence of a corporate culture skewed to the left. In March 2018 YouTube imposed “strikes” on the videos of some prominent far-right actors and conspiracy theorists, including Mike Cernovich, Infowars (Alex Jones), Atomwaffen and Sargon of Akkad (Carl Benjamin). The company maintained that its “reviewers remove content according to our policies, not according to politics or ideology.”97
In a similar way, employees at Facebook have been dismissed for apparently political reasons. Benjamin Fearnow was fired from Trending Topics for leaking Zuckerberg’s condemnation of an “All Lives Matter” sign at Facebook. In the words of Brian Amerige, a senior Facebook engineer, and founder of “FB’ers for Political Diversity,” the company has “a political monoculture that’s intolerant of different views. We claim to welcome all perspectives, but are quick to attack—often in mobs—anyone who presents a view that appears to be in opposition to left‑leaning ideology.” When Facebook imposed an outright ban on the anti-immigration, anti-Islam group “Britain First,” it was explained that the group had used language “designed to stir up hatred against groups in our society.” On July 27, after a direct appeal from the parents of a child killed at Sandy Hook, Facebook took down four Infowars videos and suspended Alex Jones for a month. On August 5 Apple stopped distributing five podcasts associated with Jones on the ground that they purveyed “hate speech.” Facebook also shut down four of Jones’s pages for “repeatedly” violating rules against hate speech and online bullying. Zuckerberg’s attempt to explain his reluctance to ban Jones backfired when he explained to Kara Swisher of Recode:
The principles that we have on what we remove from the service are: If it’s going to result in real harm, real physical harm, or if you’re attacking individuals, then that content shouldn’t be on the platform. [But] … The approach that we’ve taken to false news is not to say: You can’t say something wrong on the internet. I think that that would be too extreme. Everyone gets things wrong, and if we were taking down people’s accounts when they got a few things wrong, then that would be a hard world for giving people a voice and saying that you care about that. … I’m Jewish, and there’s a set of people who deny that the Holocaust happened. I find that deeply offensive. But at the end of the day, I don’t believe that our platform should take that down because I think there are things that different people get wrong. 98
The resulting storm of criticism illustrated the shift in attitudes in Silicon Valley. The libertarian instincts of an earlier generation of Silicon Valley entrepreneurs are being forced to yield to the more censorious attitudes of more recently hired employees who have been schooled in the modern campus culture of “no platforming” any ideas deemed to be “unsafe.”
For conservatives, as well as for right-wing populists, the alarm bells can no longer be ignored. Alex Marlow, editor-in-chief of Breitbart News, and film-makers Peter Schweizer and James O’Keefe are among those to add their voices to the growing chorus of complaint about Silicon Valley’s bias. Of considerable importance is Prager University v. Google, a lawsuit filed by Dennis Prager that accuses YouTube of violating his first amendment rights by “regulat[ing] and censor[ing] speech as if the laws governing free speech and commerce do not apply to it.” Facebook was recently forced to apologize to Prager for removing videos with the titles “Where Are the Moderate Muslims?” and “Make Men Masculine Again.” At the time of writing, allegations of anti-conservative bias on network platforms are being made with increasing frequency. Renee DiResta’s insists that the platforms “need to be able to take down users and sites that fail the tests of authenticity, organic distribution and integrity reputation.” But it seems inevitable that many of those who fall foul of “viewpoint agnostic moderation” will complain of politically-motivated censorship.99
So What Is to De Done?
American lawmakers must realize that the status quo is indefensible. The network platforms currently enjoy unprecedented power over the public sphere not only in the United States but around the world. Yet they have shown themselves to be very poor custodians of their users’ personal data. They have shown themselves to be vulnerable to abuse by foreign and domestic actors. They can no longer plausibly claim not to be publishers or media companies, as they are increasingly under pressure to curate, sort and otherwise manage the content that they host, and to do so in ways that have significant political implications. Minor modifications of the law, such as SESTA, do not address the fundamental question of how to limit the power and capacity for harm of the big tech companies.
There are five options that seem worthy of consideration:
1. Scrap “net neutrality” in order to empower the Internet service providers (ISPs) relative to the network platforms;
2. Update antitrust doctrine and law so that the network platforms can be broken up;
3. Increase the regulation of the network platforms by either the FTC or the FCC, acknowledging that they are now public utilities;
4. Repeal Section 230 altogether, thereby making the network platforms legally liable for the content they host, and leave the rest to the courts; and
5. Impose First Amendment obligations on the network platforms, recognizing that they are too important a part of the public sphere to be able to regulate access to it on the basis of their own “community standards.”
The first of these options, the ending of so-called net neutrality, is the only one to date that has been acted upon. Unfortunately, it is also the least likely to be effective. FCC chairman Ajit Pai has repealed Obama-era rules that were intended to ensure equal access to the Internet by preventing ISPs from charging users more to see certain content and to curb access to some websites. Under Pai’s new regime, ISPs will be able to block access, slow down or speed up service. For AT&T, Comcast and Verizon this has obvious appeal; for the network platforms it is much less attractive. The two kinds of company have adopted predictably opposing stances. From the point of view of the citizen-consumer, however, the net benefits of nixing net neutrality are not obvious. The most likely outcome would seem to be that Internet users will end up either paying more for certain kinds of content and enduring slower speeds for other kinds content. That does not address any of the fundamental problems described above. It merely shifts power from the network platforms to the ISPs. There is no reason to believe they are superior actors from the point of view of the public interest in a functioning market for ideas.
Of considerably more importance is the revival of interest in antitrust as a tool. This began on the left, with the so-called “antitrust hipsters” around Barry Lynn of the Open Markets Institute, who was effectively ousted from the New America Foundation last year because of Silicon Valley donors’ opposition to his work. In the past year, however, interest has grown in the idea that the big tech companies are simply too big, even amongst conservatives. The analogy with Standard Oil, which was broken up more than a century ago, is drawn with increasing frequency. In Washington, proponents of antitrust refer to Amazon, Facebook and Google as “Standard Commerce, Standard Social and Standard Data.” Tim Wu of Columbia Law School argues that Facebook should relinquish Instagram, Messenger and WhatsApp; Google should give up YouTube and DoubleClick; Amazon should spin off Amazon Web Services. Scott Galloway takes a similar view and includes Apple on his list.100 Such arguments have ceased to be the preserve of progressives.101
Antitrust law was developed in the United States as a response to the rise of trusts and combinations—that is, cross-ownership and management structures facilitating collusion. The focus of early antitrust law was (and in large part continues to be) enterprise size, market share, and strategic market positioning. At first, John D. Rockefeller was praised for bringing down the price of kerosene from 26 cents in 1870, when his market share was 4 per cent, to 7 cents in 1890, when his market share reached 90 per cent. Standard Oil was feted for bringing efficiency to the oil industry, which in turn fueled the development of steel, railroads, and the technological industries associated with them. However, in the early 1900s, journalists such as Ida Tarbell helped turn public opinion against Rockefeller (the “King of Combinations”) and his allegedly anti-competitive tactics. Trusts became a political lightning rod. Supreme Court Justice Louis D. Brandeis coined the phrase “the curse of bigness,” arguing that firms could be too big to treat employees on equal terms, to be efficient, and to treat rivals fairly. This laid the foundation for the 1911 landmark judgment in Standard Oil Co. of New Jersey v. United States, which broke Standard Oil up into 34 separate firms.
For half a century, Brandeis’s approach was dominant in the courts. In 1956, the biggest platform of that era, AT&T, settled an antitrust case by signing a consent decree in which it made two important concessions: it agreed not to expand its business into new markets (such as computers) and it made all of its patents available to others at no charge. The consent decree did not break up AT&T—that did not happen until 1984—but it was an essential first step, and one with great consequences as one of the many patents in AT&T’s portfolio was the transistor. In 1968 the Department of Justice’s merger guidelines were that any acquisition of a company with a market share of more than 3 per cent by one with a share higher than 15 per cent should be challenged. However, this view of antitrust law was challenged by Robert H. Bork and Ward S. Bowman in their seminal 1965 article “The Crisis in Antitrust,” which asserted that the intention of antitrust law was simply the protection of “consumer welfare.” Bork and Bowman argued that antitrust law could in fact harm consumer welfare by punishing aggressive pricing and preventing mergers that would reduce costs and therefore prices. “Consumer welfare” came to be equated with economic efficiency, notably in the Reagan administration’s 1982 merger guidelines, which aimed to proscribe the “ability of one or more firms profitably to maintain prices above competitive levels”—a radical departure from the previous 1968 guidelines, which had aimed “to preserve and promote market structures conducive to competition.” Another change was the narrowing of the definition of “barriers to entry” to exclude incumbent advantages from economies of scale and capital requirements. The argument was that all firms were subject to the threat of potential competition, so market power was always fleeting, and antitrust enforcement rarely needed.
This helps explain why neither IBM in the 1960s nor Microsoft in the 1990s was split up, despite their dominance of, respectively, computer hardware and software. The most that happened was that IBM had to open its platform to independent software developers and Microsoft was obliged to disclose details about the workings of its Windows operating system to rivals. The failure of the antitrust action against Microsoft looks, with hindsight, like a major turning point. Microsoft had sought to bundle its Windows operating system with its own web browser, Internet Explorer, to the disadvantage of its rival, Netscape. On April 3, 2000, Judge Thomas Penfield Jackson ruled that Microsoft had committed monopolization, attempted monopolization and tying, in violation of the Sherman Antitrust Act and on June 7, 2000, ordered a breakup of Microsoft. However, Jackson was removed from the case after he talked to reporters in an off-the-record discussion before his final decision. The DC Circuit Court of Appeals then overturned Jackson’s rulings and in 2002 the company reached a settlement with the Department of Justice that left it intact. Microsoft survived, though it has been argued that the warning shot fired by Judge Jackson had a significant effect on the conduct of Bill Gates and his colleagues. Plans to program Internet Explorer so that it would redirect users away from Google to MSN Search, or simply warn them against Google, were quietly shelved.
The interpretation and enforcement of today’s U.S. competition laws follow two main principles: the per se concept, whereby behaviors with no judicially redeeming characteristics are illegal per se, and the rule of reason, where the illegality of behaviors rests on their probable negative effect on market competition or in creating “restraints of trade.” Antitrust laws and market regulations are enforced in three ways: criminal and civil enforcement actions brought by the Department of Justice’s Antitrust Division, civil enforcement actions brought by the FTC, and lawsuits brought by private parties asserting damage claims. Yet in practice the law did almost nothing to stand in the way of the emergence of the network platforms. How could “consumer welfare” be harmed by “free” services such as Facebook. Where were barriers to entry to the Internet? The absence of good answers to such questions encouraged an increasingly overt defense of monopoly in Silicon Valley. During a 2011 Senate Judiciary Committee antitrust hearing, Alphabet Executive Chairman Eric Schmidt observed: “It’s also possible to not use Google search.” If competition was “one click away,” how was Google in breach of antitrust law? In his book Zero to One, Paypal founder and early Facebook investor Peter Thiel argued that competition was for losers, as it eroded profits; only companies with a shot at establishing a monopoly were worth investing in. At the start-up incubator Y-combinator, according to its president Sam Altman, “We … ask how the company will one day be a monopoly … We’re looking for businesses that get more powerful with scale and that are difficult to copy.”
In reaction to such boasts, some attempt to revitalize antitrust law was highly likely. In “Amazon’s Antitrust Paradox,” Lina M. Khan argues that the big tech companies have created conflicts of interest by using their dominant platforms to promote their own services against those of their competitors. This is in line with the reasoning advanced by the European Commission when it fined Google for giving more prominence to its own shopping services in search results. Khan has also revived the presumption of predation in cases of below-cost pricing, which others have argued creates pernicious long-term effects on productive investment, worker wages, product quality, and consumer choice.102 Roger McNamee, one of Facebook’s earliest investors, has denounced “the U.S. government’s laissez-faire approach to regulation” and called for (among other things) a ban on further acquisitions by the network platforms, an insistence on more equitable end-user license agreements with meaningful opt-out clauses, a return of data to the ownership of consumers, and—last but not least—a revival of “the country’s traditional approach to monopoly.”103
There are two problems with the attempt to revive pre-Bork and Bowman antitrust, aside from the extreme difficulty of achieving a rapid paradigm shift in the minds of judges. First, the software industry is prone to natural monopoly or oligopoly. Software reduces friction along existing value chains, often reconfiguring industries into two-sided markets, with platforms intermediating between owners of assets or providers of services and consumers. As demonstrated by the work of Nobel laureate Jean Tirole, two-sided markets exhibit powerful network effects and therefore produce just a few dominant firms, an effect compounded by the aggressive use of patents in software markets and a deliberately engineered lack of inter-operability. Preferential attachment models, such as those developed by the physicist Albert-László Barabási, describe markets where strength begets strength. For example, in social media networks, the users with the most friends or followers are the most likely to get any additional nodes added to the network. This produces a power-law-like distribution that makes it all but impossible for latecomers to succeed. These insights from network science challenge the Chicago School’s belief that high capital requirements or regulation are the most important barriers to market entry; network effects create a new kind of barrier. Breaking up network platforms would reduce these effects because the whole network is genuinely more valuable than the sum of its parts (Metcalfe’s law).
In any case, the historian is bound to point out that even supposedly successful antitrust actions in reality achieved much less than was intended. The outcome of the breakup of Standard Oil was in fact to make Rockefeller richer. The 34 “Baby Standards” were worth a combined $600 million in 1911, swelling to $2.9 billion in value and paying out $920 million in dividends by 1921, making Rockefeller one of the richest men in history. The interoperability forced on Microsoft, which allowed rivals to make their products more compatible with Windows, only made Microsoft more central to the software ecosystem. After the original 2001 judgment ordering the breakup of Microsoft, many predicted the company’s decline. But Microsoft retained more than 90 per cent of the operating system market until last year and Apple did not surpass it in market capitalization until 2010.
If breaking up big tech is either impossible or pointless, the obvious answer is to regulate the network platforms as utilities. There are ample precedents. In the 1876 case of Munn vs. Illinois, the Supreme Court upheld the power of the government to regulate private industries such as the railroads, though the railroads defied any attempts to regulate prices and the Munn decision was eventually reversed. The creation in 1886 of the Interstate Commerce Commission gave the government real power to intervene, but enforcement was left to the courts, which usually sided with the railroads. It was not until the twentieth century that the railroads came to be regulated—in other words, after the monopolized market had leveled off in innovation, growth and profitability. History shows that, in the United States, government intervention often serves to cement the dominance of large players for the foreseeable future, and that equity holders generally fare quite well. As Gabriel Kolko demonstrated in two seminal books—Railroads and Regulation and The Triumph of Conservatism—late nineteenth century regulation was often designed to stabilize profitability in oligopolistic industries that had reached maturity. This is a judgment supported by Carlotta Perez’s more recent work on technological development, Carl H. Fulda’s study of inland water-carriers, aircraft, and motor carriers, and conservative economist and jurist Richard A. Posner’s work on natural monopoly regulation. It is therefore hard to summon up much enthusiasm for the prospect of a more assertive FTC—or for that matter FCC—working in tandem with the biggest companies of Silicon Valley to create a regulatory framework very likely to entrench their market dominance.
Senator Mark Warner’s October 2018 paper on “Potential Policy Proposals” illustrates what a more heavily regulated tech sector might have to contend with.104 Out of twenty proposals, two have already made it into law:
1. Duty to label bots (“Blade Runner” law).
2. Disclosure requirements for online political advertisements (Honest Ads Act).
Four of Warner’s proposals envisage new legislation:
3. Comprehensive (GDPR-like) data protection legislation and establish an enforcement agency
4. Legislation banning so-called “dark patterns” such as Facebook nudges to induce users to upload their contacts.
5. A Data Transparency Bill that would “require companies to more granularly (and continuously) alert consumers to the ways in which their data was being used, counterparties it was being used with, and … what each user’s data was worth to the platform.”
6. A Data Portability Bill: “predicated on a legal recognition that data supplied by … users (or user activity) is the users’—not the service provider’s.”
Only one Warner proposal envisages reducing the scope of legislation, namely removing Section 230 immunity for state-level “dignitary torts” such as defamation, false light, public disclosure of private facts.
Four other proposals would, implicitly or explicitly, increase the power of existing federal agencies or create a new agency:
7. Restore the FTC’s rulemaking authority with respect to privacy.
8. Create an interagency task force for countering asymmetric threats to democratic institutions
9. Establish a public initiative for media literacy
10. Increase deterrence against foreign manipulation
A total of eight would impose new duties or responsibilities on network platforms, which presumably one or more of these regulatory agencies would enforce:
11. Duty to determine [the] origin of posts and / or accounts
12. Duty to identify inauthentic accounts
13. Information fiduciary duty, stipulating “not only that providers had to zealously protect user data, but also [that they had to] pledge not to utilize or manipulate the data for the benefit of the platform or third parties.”
14. Obligation to make anonymized activity data available to “independent, public interest researchers” (Public Interest Data Access Bill)
15. Requirement to get first-party consent for data collection
16. Requirement to provide consumers with the sources of data used to make algorithmic determinations or classifications.
17. Requirement for platforms to make their services interoperable with other platforms.
18. Obligation to make “essential facilities” (e.g., Google Maps) publicly available a “fair, reasonable, and non-discriminatory” (FRAND) terms.
Only one (the requirement to open federal datasets to university researchers and qualified small businesses) asks anything from the government.
This is a list fraught with technical difficulties. For example, it is often impossible to origins of posts and accounts. Insisting on identification could have unintended consequences for users living in authoritarian states and even for the civil rights of those living in democracies. Is it entirely clear to whom data belongs, if it is data based on the platform’s observation of user behavior? Would not increased portability of data go hand in hand with increased insecurity? But the biggest objection to Warner’s approach is that, as in the past, increasing the powers of federal agencies would incentivize the already establish network platforms to collude with those agencies to raise entry-barriers. For example, data portability might entrench incumbents by obliging any new entrants to give them access to their data.
A fourth approach would be to go beyond Warner’s more limited proposal on Section 230 and repeal it altogether, ending the exemption of network platforms from liability for the content they host. That exemption made sense when these companies were fledglings or did not yet exist. Today, it gives the biggest companies in the world both power and influence without responsibility or accountability. One appealing feature of getting rid of Section 230 is that it would be left to the courts to bring the network platforms to heel when plaintiffs could show that a harm had arisen from, say, a fake news story disseminated by Facebook’s News Feed. Already the courts have established that the network platforms are not exempt from liability to warn and product liability, because they have a duty of care to warn users of potential dangers. Indeed, given social media’s uniquely deep and wide knowledge of users’ interactions and relationships, they have unprecedented abilities to foresee potential harms. In one important case, a model (“Jane Doe”) had been lured by scammers on ModelMayhem, a social network for models and photographers, who then drugged and raped her, filming the incident for a pornographic video. Internet Brands, the owner of ModelMayhem, was aware of this rape ring, but did not warn any of its users. In Jane Doe No. 14 v. Internet Brands, Inc., the 9th circuit court ruled that “Doe’s negligent failure to warn claim did not seek to hold Internet Brands liable as the ‘publisher or speaker of any information provided by another …’ and therefore the Communications Decency Act did not bar the claim.” This case has established an important limit on Section 230 immunity, but it also exposes the anachronistic nature of Section 230 itself. If Internet Brands should have warned Jane Doe of the dangers of using ModelMayhem, why should not Facebook have warned all its users of the dangers of Russian-generated fake news targeted at them through the News Feed?105
There is, however, an important corollary. If we are to end the fiction that network platforms are not, in some respects, media companies or publishers, then we must at the same time end the equally dangerous fiction that they are not in many respects the modern public sphere. A first step has already been taken in this direction. In Packingham v. North Carolina
the Supreme Court last year overturned a state law that banned sex offenders from using social media. In the opinion, Justice Anthony Kennedy likened internet platforms to “the modern public square,” arguing that it was therefore unconstitutional to prevent sex offenders from accessing, and expressing opinions, on social network platforms. In other words, despite being private companies, the big tech companies have, in some cases, a public function. “While in the past there may have been difficulty in identifying the most important places (in a spatial sense) for the exchange of views,” Justice Kennedy wrote, “today the answer is clear. It is cyberspace—the vast democratic forums of the Internet in general, and social media in particular.” In May this year the Southern District of New York gave a similar ruling in Knight First Amendment Institute v. Donald J. Trump, Hope Hicks, Sarah Huckabee Sanders and Daniel Scavino:
We hold that portions of the @realDonaldTrump account—the “interactive space” where Twitter users may directly engage with the content of the President’s tweets—are properly analyzed under the ”public forum” doctrines set forth by the Supreme Court, that such space is a designated public forum, and that the blocking of the plaintiffs based on their political speech constitutes viewpoint discrimination that violates the First Amendment.106
As president, Donald Trump could not therefore block Twitter users from seeing his tweets.
If the network platforms are the new public sphere, then it cannot be their responsibility to remove “hateful content,” as 19 prominent civil right groups demanded of Facebook in October 2017, because hateful content—unless it explicitly instigates violence against a specific person—is protected by the First Amendment. To be sure, Kate Klonick has argued that tech companies should not “be held to a First Amendment standard,” because that would mean “porn stays up, spam stays up, everything stays up.” But this is not convincing. It is better that porn and spam “stay up” than that our freedom of speech be circumscribed by the community standards of unaccountable private companies, run by men who imagine themselves to be emperors.
In a series of brilliant essays, the former Facebook engineer Sam Lessin has challenged much received wisdom about the Internet. As he puts it, the Internet “increasingly represents a strange hybrid of the public and private spheres.” The problem revealed in 2016, he argues, was not fake news or feed ranking but the fact that “a public candidate can for the first time effectively talk to each individual voter privately in their own home and tell them exactly what they want to hear.” Moreover, the disappearing message feature of Snapchat and Instagram makes tracking of all advertisements harder. “Lots of data, and systems which can react properly to the interests, beliefs, and feelings of different people lead to a world where technology and brands tell us exactly what we want to hear in a way that can’t be tracked or audited.” Far from worrying about “echo chambers,” we need to worry about “personal, private and disappearing messaging that can be powerful but can’t be broadly traced or audited.”107
For Lessin, the Internet is no more likely to produce two opposing ideological camps than it was to produce a global community. Polarization is just a phase on the way to a much more complete atomization as “the pressure built into the internet’s DNA, accelerated by things like artificial intelligence, … threatens to undermine our ability to understand one another [and] see reality the same way.” Lessin poses a series of difficult questions:
Is Anonymity a Feature to Be Protected, or a Bug to Be Quashed? (Looks like the latter.)
Should Anyone Be Able to Reach Everyone? (Not clear. Who gets through the email gateways?)
Is Money a Form of Speech? (If extreme views are stickier than moderate ones, they are cheaper to disseminate online.)
Who Decides What Algorithms and Human Policies Control Our View of Reality? (Answer: network platforms, i.e. private corporations in conjunction with ~200 governments. Means rising costs.)
Will We Tolerate Unregulated Escape Hatches for Free Speech? (Answer: if not, the global village blows up.)108
His worry is that the Internet is a kind of fission bomb. First it shrank the world into a global village; now it threatens to start “a chain reaction that, just like a nuclear weapon, will cause the world to violently explode.” The problem is that, with the consolidation of the public sphere in the network platforms and the rapid collapse of the private sphere, we are losing the ability to let off steam because nothing is any longer “off the record.”109
Nothing is off the record, but no record is wholly reliable. As technological advances make it easier to create believable fake photos, video and audio recording, we may soon find ourselves unmoored from the modes of rational verification we have evolved since the Scientific Revolution. The network platforms flourished in an anarchic Internet because they appeared to create trust. But our trust in them was misplaced, because their business models incentivized them to send us fake news from fake accounts if it was sufficiently engaging. Our biggest vulnerability turns out to be that while our appetite for entertainment—branded as fiction—is finite, our appetite for “edutainment”—entertaining content branded as educational or informative—seems almost infinite. As a result, when we go online, we find ourselves a “world where we are connected to less trustworthy people and organizations than we ever would have been in the physical world. The value of the human network has almost reversed itself—from being an incredibly good way to ‘clean’ content and refine information to a system that packages together good and bad information and leaves the two indistinguishable.”110 Worse, we are no longer able to distinguish when we are connected to another human being or to a bot or other device. Those who talk of the “Internet of Things” seem to believe we shall be able to tell the 8 billion people from the 30 billion “things” that will be connected to the Internet by 2020. That will be hard when many of the things will appear to be people.111
To some commentators, the future looks exceedingly bleak. The advance of artificial intelligence, they argue, dooms mankind to a new totalitarianism, rendering liberal democracy and free-market economics “obsolete.” According to Yuval Noah Harari, “once we begin to count on AI to decide what to study, where to work, and whom to date or even marry, human life will cease to be a drama of decision making. … We are now creating tame humans who produce enormous amounts of data and function as efficient chips in a huge data-processing mechanism.” We shall soon be to data what cows are to milk.112 Lessin is only slightly more optimistic. He envisages two viable futures (and one untenable one that tries to lie between the two extremes). In his authoritarian future, national governments end up regulating all private and public speech, and the Internet effectively breaks up. In such a scenario, all countries end up going down the Chinese road. However, there is an alternative world—of “technologically guaranteed free speech,” based on blockchain technology. In this future, each person would have a public and private key-pair, as with today’s cryptocurrencies such as Bitcoin. But this would be a blockchain-based system of communication and record-keeping:
Everyone could be identified by their public key. Anyone could write public or private “claims” about any other key (signing the information with their key and the keys of intended recipients). And, critically, all the “speech” shared over time could be stored in an append-only distributed database … that would make it impossible for anyone to rewrite history or limit speech. … anyone could say anything to anyone else. Anyone could remember anything they wanted. No one’s identity could be blocked or eliminated. We would have true freedom of speech, memory and identity.
Yet this would be no libertarian paradise because “people could have as many different identities as they want … fraudulent identities spewing falsehood would abound. … there would be a huge amount of junk, lies and attempted manipulation … the most abhorrent speech [would be] unstoppable. Bullying would abound.”113 Subversive interventions by foreign governments would presumably abound, too.
None of these futures is very appealing, but each becomes more plausible the longer we leave the Internet in its present state, with generally incompetent or repressive regulation by governments or bogus self-regulation by the network platforms. The events of 2016 should have made it clear that the status quo is indefensible. The network platforms are too powerful and too reckless, too addicted to making us addicted, to be left to regulate themselves. We do not wish to follow China down the road to the surveillance state. It is doubtful that we should passively follow Europe as it seeks, at one and the same time, to live off the network platforms, by taxing and fining them, and to delegate the power of public censorship to them. It would be far better if the United States, before succumbing to the terrors of science fictional futures, sought to learn from history.
The network platforms are new, but they have recognizable antecedents. They combine the pricing power of the big railroad companies with the monopolistic tendencies of Standard Oil; the political influence of William Randolph Hearst with the technological leadership of AT&T and IBM; the convenience of Microsoft with the addictiveness and harmfulness of cigarettes. That we should trust them not to “be evil” is out of the question. We should always have expected them both to move fast and to break things that we did not want to have broken. It will not be enough to tilt the balance of power a little in the favor of the ISPs. Antitrust alone is not likely to be a swift enough remedy, and it would be a chronic optimist who believed a federal government agency would solve all the problems described above. (The recent record of financial regulators offers almost no encouragement.) Far better, surely, to oblige the network platforms to deal directly with citizens and competitors through the lawcourts, by removing the privilege on which they have grown so fat—the exemption from liability for content, established by Section 230—and at the same time to impose on them the obligation that is appropriate for entities that now so dominate the public sphere: the duty to uphold the First Amendment. This combination of remedies seems consonant with the traditions of American government. It recognizes the difficulty of reviving antitrust law in the age of network economics. It avoids the pitfall of entrusting too much power to bureaucrats. Yet by simultaneously increasing the network platforms’ costs and reducing their power to distort the democratic process, the measures proposed here would, if implemented, stand a reasonable chance of reining in the new Rockefellers—not to mention the Caesars—of Silicon Valley.
Niall Ferguson is the Milbank Family Senior Fellow at the Hoover Institution and a senior fellow of the Center for European Studies at Harvard University, where he served for twelve years as the Laurence A. Tisch Professor of History.
1 Joann S. Lublin, “David Rockefeller’s Rolodex Was the Stuff of Legend. Here’s a First Peek,” Wall Street Journal, December 5, 2017.
2 Alexis C. Madrigal, “The Education of Mark Zuckerberg,” Atlantic, November 20, 2017.
3 Scott Galloway, “Silicon Valley’s Tax-Avoiding, Job-Killing, Soul-Sucking Machine,” Esquire, February 8, 2018
4 Evan Osnos, “Can Mark Zuckerberg Fix Facebook Before It Breaks Democracy?” New Yorker, September 10, 2018.
5 Adam J. White, “Google.gov,” New Atlantis, Spring 2018.
7 The allusion is to Eli Pariser, The Filter Bubble (2011).
8 White, “Google.gov.”
9 Tim Wu, The Attention Merchants: The Epic Scramble to Get Inside Our Heads (Knopf, 2016)..
10 eMarketer, September 2017.
11 According to Verto Analytics, Facebook users in the United States spent 18 hours, 24 minutes on Facebook in Oct. 2017, compared with 32 hours, 43 minutes, recorded by Verto in Oct. 2016.
12 Adam D. I. Kramer and Cindy K. Chung, “Dimensions of Self-Expression in Facebook Status Updates,” Proceedings of the Fifth International AAAI Conference on Weblogs and Social Media (2011).
13 Elisa Shearer and Jeffrey Gottfried, “News Across Social Media Platforms 2017,” Pew Research Center, September 6, 2017.
14 Jack Nikas, “How YouTube Drives People to the Internet’s Darkest Corners,” Wall Street Journal, February 7, 2018.
15 Title V is also known as the Communications and Decency Act (CDA), for anti-obscenity provisions that have since been struck down by the Supreme Court.
16 Ashley Gold and Joanna Plucinska, “U.S., Europe Threaten Tech Industry’s Cherished Legal Shield,” Politico, October 8, 2018.
17 James Pethokoukis, “Should Big Tech be Held More Liable for the Content on their Platforms? An AEIdeas Online Symposium,” March 20, 2018.
18 Kate Klonick, “The New Governors: The People, Rules, and Processes Governing Online Speech,” Harvard Law Review (2018).
19 Klonick, “New Governors.”
20 Osnos, “Can Mark Zuckerberg Fix Facebook?”
21 Gary King, Benjamin Schneer and Ariel White, “How the News Media Activate Public Expression and Influence National Agendas,” Science, 258 (2017), 776-780.
22 Osnos, “Can Mark Zuckerberg Fix Facebook?”
23 Soroush Vosoughi, Deb Roy and Sinan Aral, “The Spread of True and False News Online,” Science, 359, 6380 (2018), 1146-1151. See also David M. J. Lazer, Matthew A. Baum, Yochai Benkler, Adam J. Berinsky, Kelly M. Greenhill, Filippo Menczer, Miriam J. Metzger, Brendan Nyhan, Gordon Pennycook, David Rothschild, Michael Schudson, Steven A. Sloman, Cass R. Sunstein, Emily A. Thorson, Duncan J. Watts, Jonathan L. Zittrain, “The Science of Fake News: Addressing Fake News Requires a Multidisciplinary Effort,” Science, 359, 6380 (March 2018), 1094-1096.
24 Renee DiResta, “There are Bots. Look Around,” Ribbon Farm, May 23, 2017.
25 Matthew Hindman and Vlad Barash, “Disinformation, and Influence Campaigns on Twitter,” Knight Foundation (October 2018).
26 Caleb Jones, “Visualizing Polarization in Political Blogs,” AllThingsGraphed, October 9,2014.
27 Pascal Wallisch, “Illumination Assumptions Account for Individual Differences in the Perceptual Interpretation of a Profoundly Ambiguous Stimulus in The Color Domain: ‘The Dress’,” Journal of Vision, 17 (4), 5 (2017), 1-14.
28 M. D. Conover, J. Ratkiewicz, M. Francisco, B. Gonc¸alves, A. Flammini, F. Menczer, , “Political Polarization on Twitter,” Proceedings of the Fifth International AAAI Conference on Weblogs and Social Media (2011).
29 William J. Brady, Julian A. Wills, John T. Jost, Joshua A. Tucker and Jay J. Van Bavel, “Emotion Shapes the Diffusion of Moralized Content in Social Networks,” www.pnas.org/cgi/doi/10.1073/pnas.1618923114 (2017),
30 Daniel Preotiuc-Pietro, Ye Liu, Daniel Hopkins, and Lyle Ungar, “Beyond Binary Labels: Political Ideology Prediction of Twitter Users,” 10.18653/v1/P17-1068 (2017), 729-740.
31 Adam Hughes and Onyi Lam, “Highly Ideological Members of Congress Have More Facebook Followers Than Moderates Do,” Pew Research Center, August 21, 2017.
32 Levi Boxell, Matthew Gentzkow and Jesse M. Shapiro, “Is the Internet Causing Political Polarization? Evidence from Demographics,” working paper (March 2017).
33 Zeynep Tufekci, “YouTube, the Great Radicalizer,” New York Times, March 10, 2018.
34 Paul P. Murphy, Kaya Yurieff and Gianluca Mezzofiore, “Exclusive: YouTube Ran Ads from Hundreds [of] Brands on Extremist Channels, CNN, April 20, 2018.
35 Marlon Mooijman, Joe Hoover, Ying Lin, Heng Ji and Morteza Dehghani, “Moralization in Social Networks and The Emergence of Violence During Protests,” Nature Human Behaviour, 2 (June 2018), 389-396.
36 Stefano DellaVigna and Ethan Kaplan, “The Fox News Effect: Media Bias and Voting,” Quarterly Journal of Economics, 122, 3 (2007), 1187-1234.
37 Sasha Issenberg, “How Obama’s Team Used Big Data to Rally Voters,” MIT Technology Review, December 19, 2012: https://www.technologyreview.com/s/509026/how-obamas-team-used-big-data-to-rally-voters/.
38 Robert Epstein and Ronald E. Robertson, “The Search Engine Manipulation Effect (SEME) and its Possible Impact on the Outcomes of Elections,” PNAS, August 4, 2015: www.pnas.org/cgi/doi/10.1073/pnas.141982811.
39 Robert Epstein, “Taming Big Tech: The Case for Monitoring,” Hacker Noon, May 17, 2018.
40 Robert Epstein, Roger Mohr, Jr. and Jeremy Martinez, “The Search Suggestion Effect (SSE): How Search Suggestions Can Be Used to Shift Opinions and Voting Preferences Dramatically and Without People’s Awareness,” Paper presented at the 98th annual meeting of the Western Psychological Association, Portland, OR, April 26, 2018.
41 Matthew Rosenberg, “Professor Apologizes for Helping Cambridge Analytica Harvest Facebook Data,” New York Times, April 22, 2018.
42 Craig Timberberg, Tony Romm and Elizabeth Dwoskin, “Facebook Said the Personal Data of Most of its 2 Billion Users Has Been Collected and Shared with Outsiders,” Washington Post, April 4, 2018.
43 Matthew Rosenberg, Nicholas Confessore and Carole Cadwalladr, “How Trump Consultants Exploited the Facebook Data of Millions,” New York Times, March 17, 2018.
44 Yuriy Gorodnichenko, Tho Pham and Oleksandr Talavera, “Social Media, Sentiment and Public Opinions: Evidence from #Brexit and #USElection,” NBER Working Paper No. 24631 (May 2018).
45 Vijaya Gadde and Yoel Roth, “Enabling Further Research of Information Operations on Twitter,” October 17, 2018.
46 Hindman and Barash, “Disinformation”
47 Jane Mayer, “How Russia Helped Swing the Election for Trump,” New Yorker, October 1, 2018.
48 Kate Starbird, “The Trolls Within: How Russian Information Operations Infiltrated Online Communities in 2016,” Medium, October 20, 2018.
49 Hindman and Barash, “Disinformation”
50 Gadde and Roth, “Enabling Further Research.”
51 Lazer et al., “Science of Fake News,” 1095. See also https://www.nytimes.com/2018/01/02/health/fakenewsconservativeliberal.html.
52 Brendan Nyhan, Andrew Guess and Jason Reifler, “Selective Exposure to Misinformation:
Evidence from the Consumption of Fake News During the 2016 U.S. Presidential Campaign,” working paper, January 9, 2018.
53 Jorg L. Spenkuch and David Toniatti, “Political Advertising and Election Results,” working paper (March 2018): https://www.kellogg.northwestern.edu/faculty/spenkuch/research/advertising.pdf.
54 Yuriy Gorodnichenko, Tho Pham and Oleksandr Talavera, “Social Media, Sentiment and Public Opinions: Evidence From #Brexit and #Uselection,” NBER Working Paper 24631 (May 2018).
55 Robert Epstein, Roger Mohr, Jr. and Jeremy Martinez, “The Search Suggestion Effect (SSE): How Search Suggestions Can Be Used to Shift Opinions and Voting Preferences Dramatically and Without People’s Awareness,” Paper presented at the 98th annual meeting of the Western Psychological Association, Portland, OR, April 26, 2018. Google denies the charge: See Gillian Tett, “Facebook or Google—Which Should Worry Us More?”, Financial Times, May 2, 2018.
56 Robert Epstein, “Taming Big Tech: The Case for Monitoring,” Medium, May 17, 2018.
57 Osnos, “Can Mark Zuckerberg Fix Facebook?”
58 Connie Loizof, “‘When You Spend $100 Million on Social Media,’ It Comes with Help, Says Trump Strategist,” TechCrunch, November 2017.
59 Robert McMillan, “Facebook Logs Text, Call Histories for Some Android Users,” Wall Street Journal, March 26, 2018.
60 Frank Bajak, “AP FACT CHECK: Does Facebook Collect Data on Non-members?” Washington Post, April 11, 2018.
61 Brian X. Chen, “Google’s File on Me Was Huge. Here’s Why It Wasn’t as Creepy as My Facebook Data,” New York Times, May 16, 2018.
62 Niall Ferguson, The Square and the Tower: Networks and Power from the Freemasons to Facebook (New York: Penguin, 2018).
63 George Soros, speech at the Davos World Economic Forum, January 25, 2017.
64 Sandy Parakilas, “Facebook Won’t Protect Your Privacy,” New York Times, November 19, 2017.
65 “Facebook’s Trust Problem,” eMarketer, January 11, 2018.
66 Knight Foundation/Gallup, “American Views: Trust, Media and Democracy” (2017), 10.
67 Kim Hart, “Exclusive: Public Wants Big Tech Regulated,” Axios, February 2018.
68 Harvard Kennedy School Institute of Politics, Survey of Young Americans’ Attitudes toward Politics and Public Service, 35th Edition: March 8-March 25, 2018.
69 Silicon Valley, We Have a Problem,” Economist, January 18, 2018.
70 Deborah M. Gordon, “Local Links Run the World,” Aeon, February 1, 2018.
71 Holly B. Shakya and Nicholas A. Christakis, “The More You Use Facebook, the Worse You Feel,” Harvard Business Review, April 10, 2017.
72 David Ginsberg and Moira Burke, “Hard Questions: Is Spending Time on Social Media Bad for Us?” Facebook, December 15, 2017.
73 James Bridle, Something is Wrong on the Internet,” Medium, November 6, 2017.
74 Franklin Foer, World Without Mind: The Existential Threat of Big Tech.
75 Jennifer King, “Are Young Adults Growing Tired of Constant Social Connectivity?” eMarketer, March 17, 2018.
76 For a critique, see Dominic Cummings, “On Referendum #24L: Fake News from the Fake News Committee, Carole, and a Rematch Against the Public,” July 27, 2018.
77 Hannes Grassegger, “The Button: Did Facebook’s “Voter Button” Influence Elections in Iceland?” Das Magazin (2018).
78 Osnos, “Can Mark Zuckerberg Fix Facebook?”
79 Paul Mozur, “Myanmar’s Military Said to Be Behind Facebook Campaign That Fueled Genocide,” New York Times, October 15, 2018.
80 Max Read, Does Even Mark Zuckerberg Know What Facebook Is?” New York Magazine, October 2, 2017.
81 Based on Danielle Keats Citron, “Extremist Speech, Compelled Conformity, and Censorship Creep,” forthcoming in the Notre Dame Law Review.
82 John Gapper, “Alibaba’s Social Credit Rating is a Risky Game,” Financial Times, February 21, 2018.
83 Lisa Linn and Josh Chin, “China’s Tech Giants Have a Second Job: Helping Beijing Spy on Its People,” Wall Street Journal, November 30, 2017.
84 Freedom House, Freedom on the Net 2017: Manipulating Social Media to Undermine Democracy (Washington, DC: Freedom House, 2018).
85 Charlie Warzel, “How People Inside Facebook Are Reacting To The Company’s Election Crisis,” BuzzFeed, October 20, 2017.
87 Alexis C. Madrigal, “When the Facebook Traffic Goes Away,” The Atlantic, October 24, 2017.
88 Kevin Roose and Sheera Frenkel, “Mark Zuckerberg’s Reckoning: ‘This Is a Major Trust Issue,’” New York Times, March 21, 2018.
89 “Mark Zuckerberg in His Own Words,” CNN, March 21, 2018.
90 Hannah Kuchler, “Facebook Moves to Make Privacy Policies More Transparent,” Financial Times, March 29, 2018.
91 Facebook post, September 8, 2018.
92 White, “Google.gov.”
93 “Mark Zuckerberg in His Own Words,” CNN, March 21, 2018.
94 Ezra Klein, “Mark Zuckerberg on Facebook’s Hardest Year, and What Comes Next,” Bloomberg, April 2, 2018.
95 Osnos, “Can Mark Zuckerberg Fix Facebook?”
96 Jonathan Swan, “Trump Hates Amazon, Not Facebook,” Axios, March 28, 2018.
97 Jonah Engel Bromwich, “YouTube Cracks Down on Far-RightVideos as Conspiracy Theories Spread,” New York Times, March 3, 2018.
98 Ezra Klein, “The Controversy Over Mark Zuckerberg’s Comments on Holocaust Denial, Explained,” Vox, July 20, 2018.
99 Renee DiResta, “Free Speech in the Age of Algorithmic Megaphones,” Wired, October 12, 2018.
100 Scott Galloway, The Four: The Hidden DNA of Amazon, Apple, Facebook and Google (New York: Portfolio/Penguin, 2017)
101 See e.g., Jonathan Tepper with Denise Hearn, The Myth of Capitalism: Monopolies and the Death of Competition (forthcoming).
102 Lina M. Khan, “Amazon’s Antitrust Paradox,” Yale Law Review, 126, 710 (2017), 710-805.
103 Roger McNamee, “How to Fix Facebook—Before It Fixes Us,” Washington Monthly, (January-March 2018).
104 Mark Warner, “Potential Policy Proposals for Regulation of Social media and Technology Firms,” Draft White paper (2018).
105 In practice, Section 230 would need to be phased out. One way to do this would be to make clear that the Section 230 safe harbor does not apply to the platform’s contributory or vicarious liability for third-party content. Vicarious liability (which would arise when the platform has the right, duty, or ability to control conduct and benefits from the conduct) would be especially important for shaping the behavior of platforms that benefit from third-party content (by selling advertising tied to the content, for example). A phased-in duty to control would give big tech companies a very strong incentive to address tortious material on their platforms, much as YouTube was induced to deal with rampant copyright infringement on its platform by a long and expensive round of litigation.
106 U.S. District Court, Southern District of New York, ruling in May 23, 2018.
107 Sam Lessin, Free Speech and Democracy in the Age of Micro-Targeting,” The Information, December 5, 2016.
108 Sam Lessin, “The Tower of Babel: Five Challenges of the Modern Internet,” The Information, November 6, 2017.
110 Sam Lessin, “How the Internet Broke and What to Do About It,” The Information, September 4, 2018.
111 Philip Howard, Pax Technica How the Internet of Things May Set Us Free Or Lock Us Up (New Haven / London: Yale University Press, 2015).
112 Yuval Noah Harari, “Why Technology Favors Tyranny,” The Atlantic, October 2018.
113 Lessin, “How the Internet Broke.”