Circulation of Russian State Media Narratives in the U.S. Media Ecosystem

• In Event: Mechanisms of Domination?: State Sponsored Information and Media Ecosystems

Abstract

We investigate the diffusion of Russian state media narratives and information in the US media ecosystem and how these patterns shifted following the full-scale Russian invasion of Ukraine on February 24th, 2022. There is growing recognition of the infiltration of Russian talking points in US domestic media. While there has been much emphasis on right-wing amplification of Russian state media narratives, in the immediate aftermath of the full-scale invasion we also observed instances when even mainstream US outlets cited Russia Today (RT) stories. Until they were exposed by a Politico story, Reuters had a partnership with Russian state media TASS and syndicated TASS content. We explore and explain these patterns in our paper. Drawing on new methods for multilingual narrative clustering and a database of 1.8 million news website articles we collected with real time web scraping from January 1, 2022 to August 31, 2022, we estimate which United States media outlets published information and narratives from Russian state media and whether these patterns shifted with the onset of the conflict.

We develop and test two theories to explain patterns of Russian state media influence. One possibility is that Russian state media may have benefitted from informational asymmetries which increased after the start of war. By increasing demand in the US for information on Russian and Ukraine and restricting access to key regions associated with the conflict, the crisis may have created a “news void” that Russian state media was uniquely positioned to fill. This perspective would predict that the diffusion of Russian state media increased following the war, especially for topics and events which Russia was uniquely positioned to report on.

A second possibility is that US media outlets strategically choose depending on audience partisanship whether to include content from Russian state media in their news content. Following the start of the war, US media outlets, especially left-of-center partisan media outlets, may have faced a reputational cost for continuing their partnerships and role in diffusing Russian state media narratives. For example, after Politico exposed the Reuters-TASS relationship, Reuters publicly announced the discontinuation of the partnership. From this perspective we expect that any decline in amplification of Russian state media narratives occurred especially among center and left wing media outlets and that right wing media outlets were more immune from reputation costs.

We draw on a new data set of 1.8 million articles collected from 58 news websites from January 1, 2022 to August 31, 2022. This database includes mainstream US news sources, low quality US news sources, Russian state media sources, and Ukrainian news sources. For the Russian and Ukranian sources we collected all language versions available for a total of 85 total sources. We collected the Ukrainian data sources in order to disentangle strategic amplification of Russian state media narratives from supply side mechanisms. For example, the news void theory would predict that Russian and Ukranian sources are substitutes and whether or not a US media source amplifies one or the other depends on Russian and Ukrainian news sources' respective access to information about the conflict. We collected these articles through daily web scraping using news websites’ RSS feeds and website archives and conducted extensive validations of our collection procedure using Internet Archive snapshots.

Methodologically, this project requires us to extract narratives from a large corpus of news articles about the Russia-Ukraine War across contexts (Russia, Ukraine, United States) languages (Russian, Ukrainian, English), and news sources (e.g. The New York Times, The Washington Post, Russia Today, Sputnik News, Natural News). This presents a challenge for classic NLP algorithms which are not typically multi-lingual, or may pick up artifacts from articles which identify the source (e.g. The New York Times style guide compared to the AP style guide). We employ and test a method to identify narratives with novel multi-lingual graph-based methods.

Our approach to narrative extraction allows us to identify not only which news websites included narratives which emerged first in Russian state media ecosystem, but also how these news websites carried this information: whether they directly syndicated content from a Russian state media outlet (such as Reuters in their previous contract with TASS) or carried the narratives but transformed them into the same copy, with or without attribution. The patterns we observe have the potential to open up new lines of inquiry regarding the structure of state propaganda diffusion in the United States and elsewhere.

Authors

• Hannah Waight, New York University

• Kevin Aslett, University of Central Florida

• Megan Brown, University of Michigan

• Jason Greenfield, New York University

• Anton Shirikov, University of Kansas

• Jonathan Nagler, New York University

• Margaret E Roberts, University of California, San Diego

• Joshua A. Tucker, New York University

• Solomon Messing, New York University