Flaws in the Algo: How Social Media Fuel Political Extremism
New research documents how social media algorithms are driving polarization.
Posted Feb 09, 2018
“Social media algorithms can be purposefully used to distribute polarizing political content and misinformation,” note the authors of a recent and timely study on fake news by Oxford University’s Computational Propaganda Project.
In “Polarization, Partisanship and Junk News Consumption over Social Media in the U.S.,” the culmination of a three-month study supported by the National Science Foundation and published earlier this week, Oxford researcher Vidya Narayanan and colleagues found that Facebook pages tied to the “extreme hard right” are responsible for circulating “more junk news than all the other [social media] audiences put together.” On Twitter, meanwhile, “a network of Trump supporters shares the widest range of known junk news sources and circulates more junk news than all the other groups put together.”
Though fake news content “is unevenly spread across the ideological spectrum,” the authors conclude, it is weighted heavily toward far-right extremism, with Facebook by far the preferred platform, though by no means the only one. (See for example “Russian Trolls Ran Wild on Tumblr and the Company Refuses to Say Anything About It” and “YouTube’s Latest Counter-Propaganda Effort Lumps PBS with State Media from China and Russia,” from just the last few days alone.)
By “junk news,” the authors refer to sources that “consistently publish political news and information that is extremist, sensationalist, conspiratorial, masked,” or otherwise falsified. The description casts a deliberately wide net among social media platforms that have no editorial departments judging what is reliable, much less fact-checking departments verifying asserted claims as true or false. Still, with more than two billion monthly active users and (according to Reuters Digital News Report in 2017) a remarkable 48% of U.S. respondents that year using it as a source of news, the influence of platforms like Facebook is considerable, not least in driving stories and reports at the height of political campaigns, when fact-checking is paramount. Facebook has since revealed that content from the Kremlin-tied Russian Internet Research Agency “reached 126 million U.S. citizens before the 2016 presidential election,” and in its own review of the same election “Twitter found that more than 50,000 automated accounts were linked to Russia,” with the number of users who interacted with these accounts reaching 1.4 million.
The Oxford authors studied the distribution of posts and comments on public pages that contain links to junk news sources in the U.S. across the political spectrum. They then mapped the influence of central sources of junk political news and information that “regularly publish content on hot button issues.” From visualizations of these clusters, they could track the paths of association online and socially while tracing junk narratives to their source. They also tracked how the URLs of extremist and conspiratorial groups were shared over Twitter and Facebook, in ways confirming that junk news frequently metastasizes unchecked on social media, reinforcing bias and polarizing groups through a self-selecting tribalism.
Coincidentally, the same day the Oxford study released, the New York Observer ran the crucial, well-sourced article “Social Media Is Helping Putin Kill Our Democracy” by John R. Schindler, a security and counterintelligence expert and former National Security Agency analyst. Drawing on a significant amount of evidence, including a detailed BBC interview with Robert Hannigan, a former British spy boss discussing the role of foreign actors in stoking disinformation and leaking via cyberespionage, Schindler outlines what he calls “the malignant influence of social media, particularly when it’s linked to authoritarian regimes seeking to undermine the West.” He joins sociologist and tech scholar Zeynep Tufekci warning in the latest issue of Wired that we're currently living through "the (democracy-poisoning) golden age of free speech" and Australian journalist Chris Zappone detailing in The Age that Silicon Valley's "techno-libertarians" are "a weak link in democracy's defense against authoritarians."
“Social media made Moscow’s clandestine work much easier and more profitable,” Schindler argues with ample proof, with planned and executed reforms by Big Tech still far from adequate as safeguards and corrective measures. “Although the lies currently emanating from the Kremlin resemble Cold War Active Measures in overall form and content, they are now disseminated so quickly, and through so many fronts, trolls, and bots, that Western governments are severely challenged to even keep up with these weaponized lies, much less push back. For this, we have the Internet to thank. While none can deny the countless benefits of the online age, this is one of its most pernicious side effects.”
With U.S. voter systems vulnerable to hacking by foreign states and actors, and social media remaining the preferred platform for Russian influence operations, Schindler is both correct and justified in warning, “It’s time the West seriously addressed the problem, and quickly, since this Kremlin spy game isn’t going away. Left unchecked, this is the ‘new normal’ that will gradually erode Western democracy itself.”
Just yesterday, NBC News independently confirmed that “Russians penetrated U.S. voter systems” during the 2016 election. According to Jeanette Manfra, head of cybersecurity at the Department of Homeland Security, “We saw a targeting of 21 states and an exceptionally small number of them were actually successfully penetrated.” “Without doubt,” she added, “we were able to determine that the scanning and probing of voter registration databases was coming from the Russian government.”
Highly critical of how Silicon Valley's "techno-libertarian ideology" aids Russia and China, Schindler is (with a growing number of tech specialists and commentators) “skeptical that social media giants can reform themselves.” They are, he says, “too rich plus too vain and contemptuous of opinions other than their own. Their self-righteous pontifications would make Gilded Age robber barons blush. As an ardent champion of free speech, I don’t want the government to step in, but there may be no choice.”
Still, with the White House refusing to confirm with seventeen U.S. intelligence agencies that Russia meddled in the 2016 election, while actively blocking Congressional action on sanctions as a reprisal for such meddling, it is hard to see how such corrective action could advance, at least at the federal level. Others have argued that “the savior of democracy might well be a combination of the U.S. city with the U.S. court system,” because “federal law does not force online ad sellers such as Facebook or Alphabet Inc’s Google and YouTube to disclose the identity of the buyers,” but state laws can and do.
That may be the only way to stanch the flow of dark posts and unregulated junk news, with its endless conspiracy-mongering and stoking of disinformation, all designed to confuse and polarize voters. As Tufekci is quoted as saying of online platforms such as YouTube, “Its search and recommender algorithms are misinformation engines…. [Its] recommendation algorithm is not optimising for what is truthful, or balanced, or healthy for democracy.”
V. Narayanan, V. Barash, J. Kelly, B. Kollanyi, L.-M. Neudert, and P. N. Howard (2018). “Polarization, Partisanship and Junk News Consumption over Social Media in the US.” Data Memo 1. Oxford, UK: Project on Computational Propaganda. [Link]