Kremlin-backed trolls are sowing disinformation and have adapted their game for the 2020 campaign. U.S. Attorney General Bill Barr has been vague about the probity of accepting foreign assistance, and the U.S. Office of the Director of National Intelligence recently announced that it would no longer provide in-person briefings to Congress on foreign electoral interference. To boot, Facebook recently removed pages linked to Russian intelligence in hopes of heading off a disinformation campaign closer to election day. Four weeks before one of the most important presidential elections in its history, the United States remains vulnerable to foreign influence online — a vulnerability that stems as much from U.S. politics and media as from Russian bots and hackers.
Understanding the precise nature of foreign electoral interference online is important for two reasons. Underplaying the risks leaves U.S. elections vulnerable to manipulation by adversaries. Overplaying the risks may undermine the legitimacy of elections by allowing the losing side to claim that it lost due to foreign interference. Inflating the risks also allows rival countries who interfere in elections online to claim credit and power beyond their true capabilities. The power of foreign electoral interference lies as much in perceptions as in reality.
Grasping the problem of foreign electoral interference online is made more complicated because it can take many forms, from disinformation campaigns targeting voters, to “ “hack and leak” operations that release sensitive information to the public, to cyber attacks on election infrastructure. These activities pose different threats for U.S. elections. To understand the problem of foreign electoral interference online, we need to identify what types of threats are most dangerous and why.
Popular discussion tends to root the problem of foreign electoral interference online in the “tsunamis of falsehood” spread via disinformation campaigns on social media, but election experts note that foreign troll activity on social media is typically too small-scale and inept to have much of an impact on elections. The far greater dangers come from hack-and-dump operations that leak stolen information online and the threat of cyber attacks on voter registration databases or voting machines. Rebuffing these efforts requires dealing with foreign internet trolls and hackers, but successful foreign interference in elections is usually an inside job. Media outlets that uncritically report material from hack-and-dump operations and political parties who benefit from foreign interference are keys to the success of any foreign meddling. Combatting foreign interference online is as much a political as a technical challenge.
Online Influence in the 2016 U.S. Election
Given the outsize attention it has received, one might think that Russian social media operations in the 2016 election were novel, touched many voters, and swung the election to President Donald Trump. Each of these claims merits scrutiny. To be sure, the internet and social media supercharged Moscow’s influence operation in 2016, but foreign electoral interference is far from new. Recent books by Thomas Rid, David Shimer, and Dov Levin recount many episodes of the United States and the Soviet Union intervening in foreign elections with varying degrees of success, via cash transfers to political parties and movements, disinformation spread to journalists and activists, and various forms of dirty tricks over the last 100 years.
Rid’s Active Measures: The Secret History of Disinformation and Political Warfare is especially good at putting recent disinformation campaigns against the United States in the broader context of political competition among great powers. Shimer’s Rigged: Russia, America, and One Hundred Years of Covert Electoral Interference covers a similar period, but devotes more attention to the contemporary period and captures the infighting within the U.S. government about how to respond to Russian electoral interference in 2016 with great skill. Levin’s Meddling in the Ballot Box: The Causes and Consequences of Partisan Electoral Interventions explores whether foreign intervention by the great powers in the post-war era actually influenced electoral outcomes.
Popular wisdom not only overstates the novelty of foreign interference in elections, it also exaggerates the reach of foreign actors on social media during elections. Observers frequently note that accounts linked to the Kremlin were responsible for more than 126 million shares on Facebook, 1,108 videos on YouTube, and roughly 2.1 million tweets in the 2016 election. Shimer, for instance, chides the Obama administration for paying insufficient attention to Russian online influence and notes that “two-thirds of Americans get at least some of their news from social media.”
These large numbers, however, overstate the case. Only one in five Americans is on Twitter, few are active users, and fewer still have large followings. Rid points out that the oft-cited figure of 126 million Facebook shares from Russian accounts was roughly one-one hundredth of one percent of Facebook shares related to the election. Twitter reported that Russian automated accounts known as “bots” tweeted 2.1 million times before the 2016 election, but this represented just .05 percent of tweets viewed related to the election. The infamous Internet Research Agency troll factory in St. Petersburg devoted around $1.25 million per month to election interference, but the Clinton and Trump campaigns and related PACs raised $1.2 billion and $617 million respectively during the campaign. Assessing the scale of foreign influence attempts online suffers from a “denominator problem.” Without representing these numbers as a fraction of online activity, it is easy to misstate their impact.
Fake news of the kinds associated with Russian disinformation campaigns attracts less attention than does other sources of news. Economists Hunt Allcott and Matt Gentzkow tracked social media consumption in 2016 and found that “even the most widely circulated fake news stories were seen by only a small fraction of Americans” and estimated that the average adult saw and remembered about one fake news story during the campaign. Another study demonstrated that fake news is overwhelmingly shared by a small number of dedicated viewers, particularly on the extreme political right. Studies with more expansive definitions of fake news find a larger footprint, but the scale of such operations is still a fraction of more legitimate news.
Moreover, we should not assume that seeing a tweet or Facebook share affects voters. In Not Born Yesterday: The Science of Who We Are and What We Believe, social psychologist Hugo Mercier reminds us how hard it is to move opinions via social media. In an interview, he notes that a large body of research from political science, history, marketing, and psychology finds that “people aren’t gullible; they have all the cognitive mechanisms required to carefully evaluate communication, and mass persuasion attempts — from propaganda to advertising — nearly always fall on deaf ears.” Mercier adds that improvements in technology, from the radio to television to social media, have always “been decried as terrifying tools of mind control” but in the end,
getting people to accept uncongenial views, or to engage in costly behaviors, will remain tremendously difficult. Cambridge Analytica and the Russian bots fit this pattern: a lot of panicky talk, but no evidence that they had any significant impact on the election.
Further, political scientists find that even campaign advertising by political parties on a much greater scale than the operations conducted by foreign trolls in 2016 usually has little to no effect on voting. A rigorous study of 49 television commercials made by professional advertising firms on behalf of candidates in the 2016 U.S. election cycle found that the effects of the advertisements on vote choice and candidate favorability were small. Researchers found that the “small effects of political advertising are small regardless of context, message, sender or receiver.” It would be surprising if Russian trolls on social media were more successful.
And it is not easy for foreign trolls to capture the subtleties needed to navigate American politics. For example, Internet Research Agency efforts to mimic local news sources in 2016 largely fell flat. The Social Media and Political Participation lab at New York University analyzed Internet Research Agency tweets in which the sender tried to pose as local news sources and found these efforts to be small and ineffective. Just 27 accounts at the Internet Research Agency produced 80 percent of this material, and the average tweet from these accounts received fewer than 50 re-tweets or likes. Moreover, these Internet Research Agency tweets were not targeted at local news sources in swing states and were most active in Kansas, California, and Missouri.
To demonstrate the Kremlin’s influence online, many point to Russian-led efforts on social media to pit Black Lives Matter activists or groups supporting tolerance toward Islam against white nationalists to foment politically divisive protests. Such efforts occurred in Florida, Alabama, and Texas, but found few takers. One oft-cited attempt in Houston that saw Russian-based hackers create an online white nationalist organization called “The Heart of Texas” and an organization called “Save Islamic Knowledge” drew a grand total of 60 protestors in a city of more than 2.3 million.
While popular wisdom views Russian social media efforts as tipping the election in Trump’s favor, many election experts are more skeptical. Nate Silver notes that: “If you wrote out a list of the most important factors in the 2016 election, I’m not sure that Russian social media memes would be among the top 100. The scale was quite small and there’s not much evidence that they were effective.” Political scientists John Sides, Michael Tesler, and Lynn Vavreck, in their in-depth study of the 2016 campaign, note that: “although Russian interference was and is deeply concerning, there are many reasons to doubt that it changed the outcome of the election.” We can never know the impact of Russia’s efforts in 2016, but it seems unlikely that Moscow-backed trolls on social media won the election for Trump.
Operations by foreign actors to influence politics in the United States are objectionable regardless of their effectiveness. Any attempts by foreign actors to influence elections in the United States on social media should be condemned as violations of national sovereignty and investigated. The U.S. government, social media companies, and the mass public should remain vigilant, particularly since social media is constantly evolving. But it is also important to keep foreign influence efforts on social media in perspective. A large recent study found that from 2015 to 2018, far more disinformation about elections online came from U.S. than from Russian sources. And there is a danger in overstating Moscow’s influence online. Doing so merely reinforces the narrative of an “omnipotent” Kremlin that is so near and dear to Russian President Vladimir Putin’s heart.
The Media Problem
Kathleen Hall Jamieson, one of the most prominent scholars in political communication, offers a different take. Rather than focusing on the impact of foreign trolls on social media, she explores the influence of the Russian-WikiLeaks “hack-and-dump” operation against the Democratic National Committee in 2016. In Cyberwar: How Russian Hackers and Trolls Helped Elect a President: What We Don’t, Can’t, and Do Know, she argues that it was not so much the direct effect of Russian efforts online that helped candidate Trump in 2016, but the U.S. media’s treatment of Russian meddling.
She contends that the release of material stolen from the Democratic National Committee by some combination of Russian agents and WikiLeaks changed the media coverage of the race in a way that amplified Hillary Clinton’s perceived weaknesses and depressed her support. By uncritically passing stories from stolen documents as “leaks” and “news,” the media kept stories damaging to Clinton in the public eye. As the election approached, stories from the WikiLeaks material played on voters’ perceptions of Clinton’s duplicity — a theme emphasized by the Trump campaign. They reminded readers of her implausible explanations for using a private email server while secretary of state, the generous speaking fees she accepted while championing economic equality, and discrepancies between her public remarks and her speech to Goldman Sachs executives. For Hall Jamieson, foreign interference likely changed the election outcome in 2016 because of the failure of the American media to put purloined information in its proper context and to remind readers of the dubious and illegal nature of the source material.
Hall Jamieson admits that the 2016 election was a special case. It was a very close election, and Clinton may have been especially vulnerable to this kind of campaign. While there is no “smoking gun,” Jamieson makes the strongest argument for how Russian interference in the election might have affected a large number of voters and tipped the balance. Moreover, one need not believe that the Russian hack-and-dump operations abetted by the U.S. media swung the election to Trump to recognize the media’s uncritical reporting of hacked information as a great problem for U.S. democracy.
Attacking Election Infrastructure
Disinformation campaigns and hack-and-dump operations are not the only forms of electoral interference. Cyber attacks to disable political party websites, change voter rolls, and alter vote tallies before, on, or after election day are all on the menu. This threat is far from theoretical. In the 2014 Ukrainian presidential election, Kremlin-allied hackers fed false electoral results to Ukrainian media outlets. These results were quickly corrected, but caused temporary panic in Kyiv. Prior to the French presidential election in 2017, Russian hackers dumped stolen emails that sought to embarrass Emmanuel Macron, the candidate most skeptical of Moscow. In a brilliant counter-move, Macron’s team anticipated that Moscow might try a hack-and-dump operation and mixed false documents with true material that was likely to be the target of a hack. Once the hackers revealed the stolen information, the French government discredited the operation by pointing to the fraudulent documents it contained.
In the United States in 2016, hackers believed to be working for Russia scanned at least 21 state-level databases of election rolls and thereby let the Obama White House know that they were in position to cause chaos on election day or to delegitimize a Clinton victory by claiming to have altered the results. As investigative journalist Kim Zetter notes, election infrastructure in the United States remains vulnerable today. Indeed, ransomware attacks on voter registration data seem to be the preferred mode of assault in 2020.
Few observers believe that hackers working on behalf of Moscow tampered with election results or disabled voting machines in 2016. The Kremlin may have feared the U.S. reaction to such an escalation. But even without tampering with voting machines, Shimer finds that Russian efforts staggered the U.S. government. They tied the Obama administration in knots for months before and after the election. In vivid detail based on extensive interviews with relevant officials, Shimer depicts the frustration of officials in the Obama government. Some called for punitive actions to deter Russian efforts to interfere: Only by inflicting costs on the Kremlin, perhaps by revealing damaging information about Putin or engaging in a cyber attack of its own, could the U.S. government compel Moscow to desist. Others feared that a more forceful response would provoke even more aggressive measures from the Kremlin.
Beyond the White House, Russian electoral interference deepened partisan divides and immobilized the government. Shimer recounts how the Obama White House, prominent congressional Democrats, and House Speaker Paul Ryan agreed to issue a statement alerting the public to the scale of interference by the Russians but were rebuffed by Senate Majority Leader Mitch McConnell, who refused to sign the document. The Senate majority leader threatened to accuse the president of meddling in the election if the Obama administration went public with its own statement. After weeks of delay, on Sept. 28, the White House and congressional leaders (including McConnell) issued a watered-down statement that encouraged states to shore up their electoral defenses, but did not mention Russia, and went largely unnoticed.
In the end, Shimer reports that Obama told Putin in person to end the operation and Russian influence efforts on social media subsided somewhat, but many in the Obama White House would have preferred a more forceful response in retrospect. That White House and congressional officials were so occupied with whether and how to respond in the months before and after the election reveals the power of foreign electoral interference.
Much popular discussion emphasizes that foreign electoral interference is a technical challenge and focuses on how to combat the spread of viral disinformation and prevent hack-and-dump operations, but the problem is ultimately political. Dov Levin’s Meddling in the Ballot Box makes the case quite well. Based on original data from more than 900 elections between 1946 and 2000, Levin finds that one of the two superpowers intervened in some way in 117 elections — that is, roughly one in nine. Washington was much more likely to intervene than was Moscow (81 versus 36), and foreign interventions occurred in 60 countries across six continents. Most foreign interventions (64 percent) were covert, but these were generally less successful than more overt forms of foreign influence. By cataloging episodes of partisan electoral interference by Washington and Moscow over a long period of time in such a systematic fashion, Levin has done a great service.
Levin presents six case studies of U.S. attempts to influence elections abroad during the Cold War, ranging from Argentina in 1946 to Greece in 1967, and a statistical analysis of electoral interference from 1946 to 2000. He presents two findings. Foreign interference is most likely to occur when the intervener sees the potential winning side as an implacable foe, which is how the U.S. viewed leftist regimes in Europe or Latin America during the Cold War.
More important, successful foreign electoral interventions are “inside jobs” that require a willing partner in the target country. He recounts how in the 1953 West German parliamentary elections the Christian Democrats worked closely with the Eisenhower administration to stave off the Social Democratic Party in a close election. In coordination with the Christian Democrats, the U.S. government provided food aid, revived a friendship treaty with West Germany, and agreed to take part in Four-Power negotiations on the status of Berlin and the future of Germany with the United Kingdom, the Soviet Union and France. Levin points out that two days prior to the election, U.S. Secretary of State John Foster Dulles publicly threatened “disastrous effects” for Germany if the Christian Democrats lost.
Levin could not resist the temptation to extend his findings to the 2016 U.S. presidential elections, but this goes too far. Because the cases in his sample are largely drawn from the Cold War, overwhelmingly initiated by the United States against weak democracies, and predate the development of the internet, it is hard to draw clear parallels to contemporary U.S. politics. But Levin’s broader point of the need for willing partners in the target country is critical to understanding the problem of foreign electoral interference.
After all, Rid, Levin, and Shimer remind us that the Kremlin has tried to influence elections in the United States before. Moscow reached out to Adlai Stevenson, John F. Kennedy, and Hubert Humphrey to aid their presidential campaigns, but were rebuffed. In 1992, when three top Republican operatives suggested that the George H. W. Bush campaign ask the Russians to dig up dirt on candidate Bill Clinton, White House Chief of Staff James Baker shut down the proposal.
The United States could do far more to protect its elections from foreign influence and interference. If any country has the technical capacity to defend itself against attacks on election infrastructure using both offensive and defensive techniques, it is the United States. But the problem is ultimately a political one rooted in U.S. media and political parties who benefit from these operations.
Some media outlets are changing how they cover leaked material. The Washington Post recently issued new guidelines for handling sensitive materials obtained via hacking, including increased editorial oversight and an explicit discussion of the source of the material and the likely aims of the leaker. But Shimer reports in Rigged that some journalists from Time, the Wall Street Journal and the Washington Post saw little wrong in how they handled the emails hacked from the Democratic National Committee in 2016 and would do the same in 2020 given the chance. Shimer quotes Philip Rucker from the Washington Post: “If we did it all over again, we would still cover the emails. The coverage would have been a little less breathless, but we would not have kept information from the public if it were newsworthy.”
In addition, the norm against political parties accepting foreign aid has weakened as Trump has courted foreign assistance in 2016 and 2020 with little criticism from leading Republicans. Until the media changes its coverage of foreign influence attempts online and political parties pay a cost for accepting assistance from abroad, the United States will remain vulnerable to foreign interference in its elections.
Timothy Frye is the Marshall D. Shulman Professor of Post-Soviet Foreign Policy at Columbia University, the co-director of the International Center for the Study of Institutions and Development research lab at the Higher School of Economics in Moscow, and the author of Weak Strongman: The Limits of Power in Putin’s Russia (Princeton University Press, April 2021).
Image: Joe Shlabotnik