Featured Image: The Washington Post
by Ryan P. Burge, Eastern Illinois University
The 2016 election will be remembered for a lot of reasons, not the least of which is that it was the first election in American history in which the electoral process was likely deliberately manipulated by a foreign power. Even GOP members of Congress have publicly stated that the Russians were involved in an assault on American democracy. Oklahoma senator James Lankford said, “I don’t have any doubt that Russia’s been engaged in trying to interfere with our elections, just as they’ve done with other countries all around the world, all around their region. They’ve been very engaged in this.” One of the most effective vectors for manipulation was social media, specifically Twitter.
In 2013, the Kremlin created an outfit called the Internet Research Association, whose primary mission was to influence foreign elections. While largely dormant in its beginning it began ramping up its number of Twitter accounts and volume of tweets in the run up to the 2016 presidential election. This is the outfit whose 13 members were indicted on Friday by a grand jury in Washington for these activities.
During the electoral season, Twitter was (apparently) completely unaware that Russians were using its platform to tilt public opinion away from Hillary Clinton. However in January of 2018, after a lengthy investigation, Twitter emailed 700,000 of its users to notify them that they had interacted with a suspected Russian troll account. Two weeks later, that number doubled to nearly 1.5 million. In its entirety, Twitter handed to Congress the full account details of 3,814 users that were likely connected to the Internet Research Association. Then it scrubbed that data off its public servers.
However, NBC News managed to collect a vast amount of information about these 3,814 accounts, including a full tweet list which is over 200,000 entries in total. The dates of the tweets were from July of 2014 to September of 2017. They were kind enough to make that publicly available to researchers. I wanted to see if those trolls used religion as part of their propaganda campaigns or did they stick strictly to politics?
The collection of words searched for is in no way exhaustive, but covers a lot of terms that have been used for religious traditions and beliefs in the last few years. It’s important to note that the total number of religious words used in all these tweets is fairly small. In fact, just 1.6% of the words in the entire dataset were religious words. However, there were some interesting patterns that emerged. The most frequently used stem, by almost a factor of two, was “Islam.” In fact it was used more than the next two most frequent words (“pray” and “Christ”) put together. Also, note that “Muslim” was used the fourth most frequently. In contrast, words that relate to specific policy issues like “gay” or “abortion” are basically non-existent. It’s clear that these trolls were focused less on the Christian faith and more on Islam given how divisive the treatment of Muslims was in 2016 and since.
It’s also clear from this data that the trolls used these term in an event driven way. They were not trying to generate buzz out of nothing, but instead wanted to ride the wave of current events and put their own spin on the news of the day. The biggest spike was on March 22, 2016, which was when a terrorist attack occurred in the Brussels airport, killing at least thirty people and wounding hundreds more. The second most popular date, December 3, 2015 also marked a terrorist attack, this time on U.S. soil in San Bernardino, California. This attack was carried outby individuals who had been radicalized by watching online propaganda videos from ISIS and other terror groups. It’s clear that Islamic terrorism was a primary vector for these Twitter trolls and they used any negative events to push their message.
How influential were these Twitter accounts? The answer varies dramatically. The one account that is getting the most media attention is one from @ten_gop. It’s name on Twitter was “Tennessee” and many people assumed that it was being run by state party officials, however it was actually property of the Kremlin. According to published reports, that ten_gop account had 136,000 followers before it was shut down by Twitter in August of 2017. An archive of that account’s tweets can be found here. Looked at generally, the most successful accounts used religious imagery between 2-4% of the time, therefore it doesn’t seem like it was a primary motivator for success on social media.
This finding is affirmed by a scatterplot of about 130 Twitter accounts included in the data dump. The x-axis represents the percent of tweets that included a religious reference, while the y-axis is that account’s followership (in thousands). The blue line tries to identify a trend between these two factors. As is evident here, there is not really a relationship between using more religious terminology and having more followers. It seems that what motivated follower counts was something else entirely. I looked at the 25 tweets that received the highest number of retweets and none of them contained any religious language at all. Many of them were trying to discredit the mainstream media or Hillary Clinton’s campaign. In essence, they were purely political.
I took a few minutes to read through a handful of the tweets from a variety of accounts, and the one impression I got was clear: these were not written by bots. I have done a little work with artificial intelligence and the technology has a long way to go when it comes to natural language processing. There is no way an algorithm could write text that seems so natural and has the nuance of the English language. The other thing is a lot of these tweets are extremely benign. One of the most popular is: “Muhammad Ali, the only person whose Hollywood Walk of Fame Star is hanging on a wall, not for anyone to step on.” That’s not overtly religious or political, but it was retweeted 11,616 times. It’s extremely likely that it was read by over 100,000 people. It would surely build a following, which opens up a huge audience to subtle propaganda.
We don’t yet know if these foreign Twitter trolls are being employed to attack other institutions. It doesn’t take too much imagination to foresee an effort by a foreign government to try and discredit the Catholic Church, or a particularly influential religious leader in a specific region. After doing this analysis, I can say that I have no idea how even a perceptive individual is going to be able to tell when they are just reading innocuous social media banter or a targeted and highly orchestrated attack on the “powers that be” in the United States and across the world.
Overall, religion was not a central theme of the 2016 Presidential campaign. While there was discussion about Trump’s lack of religious bona fides, and Clinton’s background in the United Methodist church, those were used more to push voters away from the opposing candidate instead of toward themselves. In sum, it seems that religion was just not that salient, despite the fact that voter coalitions lined up that way.
Full coding syntax for this analysis is available on my Github.