Content Sharing within the Alternative Media Echo-System: The Case of the White Helmets

In June 2017 our lab began a research project looking at online conversations about the Syria Civil Defence (aka the “White Helmets”). Over the last 8–9 months, we have spent hundreds of hours conducting analysis on the tweets, accounts, articles, and websites involved in that discourse. Our first peer-reviewed paper was recently accepted to an upcoming conference (ICWSM-18). That paper focuses on a small piece of the structure and dynamics of this conversation, specifically looking at content sharing across websites. Here, I describe that research and highlight a few of the findings.

One important insight from our broader work, which has included deep immersion in the online content about the White Helmets, is that this conversation can be extremely disorienting. Aligned with previous work examining claims of “crisis actors” staging mass shooting events, the online discussions about the White Helmets include a large number of confusing, conflicting, and contradictory claims. In this article, I do not attempt to describe all the narratives or unwind all the claims, but instead focus on revealing some of the dynamics that support these conversations.

Background and Context

The civil war in Syria began in 2011 during the “Arab Spring,” after anti-government protests demanding the resignation of President Bashar al-Assad escalated into violence between the government and multiple opposition groups. That violence eventually evolved into a multi-sided, international conflict — with Russian and Syria providing support to Assad, the Islamic State (IS) capturing territory in the north and east, and the United States and Gulf League states forming a coalition in support of Syrian opposition groups, and Kurdish forces fighting IS.

The White Helmets are a volunteer humanitarian organization than works in rebel-held areas of Syria. The group has been trained to provide assistance to those affected by the violence, including conducting search and rescue, evacuating buildings, fighting fires, and providing medical aid. They also record their activities using video and photos to document the human toll of the violence, specifically highlighting impacts from attacks perpetrated by the Syrian government (and their Russian and Iranian allies). This includes documentation of chemical weapons attacks. In 2016, their work was featured in a documentary film that served to draw attention — from people all around the world — to the plight of Syrian people suffering at the hands of their government. Western, “mainstream” media helped to facilitate the spread of this messaging, carrying content that was supportive of the White Helmets and sympathetic to the people featured in their photos and videos. The White Helmets therefore served as a conduit for sympathy between western audiences and people in rebel-held areas of Syria, who were portrayed as victims of Assad’s brutality.

Perhaps not surprisingly, supporters of Assad’s regime saw the White Helmets differently, as functioning to support the rebellion both directly (through their on-the-ground response efforts) and indirectly (by generating attention to and fostering solidarity with Syrian people in rebel-held areas of the country). In response, a group of voices — including “independent” journalists, activists, and Syrian government officials — began to actively resist the sympathetic framing of the White Helmets. They countered with various narratives, claiming that the White Helmets were a “propaganda construct” of western media and US/NATO, that they were “crisis actors” who staged events (such as chemical attacks) to advance the interests of their US/NATO allies, and that they worked with or were themselves terrorists.

Research on the Media Ecosystem surrounding White Helmets conversations on Twitter (Summer 2017)

Our lab began to investigate the contested framing of the White Helmets in June of 2017. This work is ongoing and includes multiple aspects of this online discourse. Using methods that we developed for studying online rumoring (e.g. see Rumoring Paper 1, Rumoring Paper 2) and then evolved for studying framing contests and disinformation campaigns that rely on “crisis actors” claims, in this study we examined Twitter conversations about the White Helmets in the summer of 2017. In particular, we were looking at the surrounding media ecosystem — the online websites that provide the content that feeds and shapes the conversations we see on social media. And, even more specifically, we wanted to better understand the content-sharing practices across these different websites. In previous work, we had seen that people encouraged to “do their own research” within an alternative media ecosystem might repeatedly encounter the same content, spread across multiple different websites. We hypothesized that this could result in “false triangulation,” where people think they are getting independent confirmation from different sources, but in reality this content resolves to the same source. In this study, we attempted to better understand these content-sharing practices.

We used Twitter data to “seed” this analysis. We began with a dataset of tweets, collected between May 27, 2017 and September 9, 2017, that had either “white helmets” or “whitehelmets” in the text. We then used links embedded in those tweets to identify the different articles that were cited in the conversation. And finally, we conducted an analysis to identify where the same article (in terms of content) appeared in all or large part across two or more different domains. In other words, we identified instances where two different websites contained an exact or very close copy of the same article. [Our methods are explained in more detail in the paper.]

The Content Sharing Domain Network Graph

Using this analysis of similar articles, we created a network graph (Figures 1 & 2) to visualize how the same articles appeared across multiple domains. This graph features every website or “web domain” that was 1) cited in a tweet in our “white helmets” dataset; and 2) contained an article that also appeared (with a high amount of word similarity) in another domain in the graph. Each of these websites is a node (circle) in the graph. We have sized these nodes by the number of tweets in our collection that link to an article on this website. Nodes are connected by an edge (line) if the same article appeared on both websites. The thickness or weight of that edge is proportional to the number of different articles that appeared on both websites. The thinnest edges represent a single article shared across both sites. The thickest represent seven articles.

Figure 1: The Complete Content-Sharing Domain Network Graph for “White Helmets” Tweets, Summer 2017

Figure 1 shows the entire graph for this conversation. The graph has two large clusters (a pink one in the lower left, and a blue one in the upper right) with some connective tissue in between. There are also a number of small clusters constituted by fewer than 10 domains each.

The pink cluster was formed through the distributed sharing of two Associated Press (AP) articles. The AP model facilitates content sharing across thousands of news outlets around the world. In this data set, there were two articles from the AP — both sympathetic to the White Helmets, reporting on the murder of seven White Helmets volunteers by members of IS — that were shared by dozens of local and regional media websites. The edges in this cluster are thin, demonstrating widely distributed, but not persistent, content sharing by these sites in relation to the White Helmets conversation. Additionally, the volume of tweets pointing to each of these websites is relatively small.

The connective tissue (websites in between the two large clusters) consist of a few mainstream media web domains (e.g. theguardian.com, independent.co.uk) that are pulled together by a small number of news aggregator sites that republish content from these websites. Most of the articles from the websites in this area were supportive of the White Helmets. [Please see the paper for more details.]

The biggest graph component in terms of both tweet volume and number of websites (blue cluster in the upper right) is constituted by a group of diverse websites that are held together (in the graph) through repeated content-sharing across those websites. Figure 2 is a close-up view of the blue and yellow cluster. We provide more discussion of this cluster, which we refer to as an Alternative Media Ecosystem, below.

Figure 2: A Closer Look at the Alternative Media Ecosystem, Nodes Colored Darker (Added Red) by Degree Degree = Number of Connections

Most Tweeted Web Domains

We also include here a list of the top 10 websites, in terms of tweet volume, for “White Helmets” tweet conversations in Summer 2017 (Table 1). Eight of these ten most-tweeted websites are deeply integrated into the upper-right cluster. The articles posted on these websites are highly critical of the White Helmets, echoing various narratives — accusing them of being terrorists and/or propaganda tools of US/NATO. Two other top websites (Newsweek and Alternet) are more peripheral nodes. Newsweek appears here due to an article that was critical of the White Helmets, highlighting an instance of two volunteers being photographed while handling mutilated bodies of Syrian soldiers, and explaining how they had been fired for those acts. [Please see the paper for more details.]

Table 1. Top 10 Most-Tweeted Web Domains
╔══════════════════════╦════════════╦════════╗
║ Domain ║ # Tweets ║ Degree ║
╠══════════════════════╬════════════╬════════╣
║ 21stcenturywire.com ║ 3119 ║ 30 ║
║ clarityofsignal.com ║ 2391 ║ 1 ║
║ mintpressnews.com ║ 1630 ║ 22 ║
║ alternet.org ║ 1219 ║ 6 ║
║ sputniknews.com ║ 1110 ║ 16 ║
║ newsweek.com ║ 1046 ║ 2 ║
║ rt.com ║ 879 ║ 17 ║
║ globalresearch.ca ║ 707 ║ 33 ║
║ theantimedia.org ║ 682 ║ 30 ║
║ unz.com ║ 512 ║ 22 ║ ╚══════════════════════╩════════════╩════════╝

Main Findings

The Twitter conversations about the White Helmets during the summer of 2017 were dominated by anti-White Helmets (anti-WH) voices. Our analysis reveals that anti-WH content was much more highly cited than pro-WH content during that time period. There were more than four times as many tweets linking to web domains in the alternative media cluster in the upper right than there were to the lower left cluster and its nearby connective tissue (the “mainstream” media).

An Alternative Media Ecosystem. Our analysis demonstrates that a small number of websites play an outsized role in shaping the conversation around the White Helmets. In particular, 21stCenturyWire, MintPressNews, GlobalResearch, SputnikNews, RT, TheAntiMedia, and UNZ are highly cited and central to this conversation. Though not as highly cited here, the ActivistPost also plays a significant role as the original source for content that gets shared elsewhere in graph. These websites are situated somewhat differently in terms of their political leanings, but have some interesting commonalities. They consistently present themselves as “independent” and “alternative” media, and as offering a contrary perspective to mainstream news and views (which they claim purposefully hide the truth from people). They describe their audiences as members of the “waking generation” who want to “question more” and therefore have access to this alternative reality. We characterize this cluster of websites as an Alternative Media Ecosystem.

Content sharing is a significant dynamic of the Alternative Media Ecosystem. One of the most interesting dynamics we found in this ecosystem — which became the focus of this first study — was content sharing across the different websites. The same article might appear on several and sometimes dozens of different websites. In fact, much of the content shared across this network originated in just a few places — a few prominent journalists and bloggers who publish their work on websites like 21stCenturyWire, GlobalResearch, MintPressNews, and ActivistPost. This content then gets picked up by other, smaller and more niche sites — from partisan news sites (of varying political perspectives that seemingly bridge “left” and “right”) to clickbait sites to conspiracy theory sites to a subcluster of sites that are specifically focused on Russia-related news.

Content-sharing also occurs in the other side of the graph, among the “mainstream” media. In particular, there is a cluster of websites that republished articles from the AP. However, the structure of content sharing in that cluster is distinct from that of the alternative media ecosystem, where the overall volume (tweets linking to these websites) is much higher and channels of content sharing are both deeper and more varied . In other words, whereas in the AP cluster a large number of websites engaged in the conversation by carrying a single AP article, in the alternative media cluster a small number of websites are repeatedly producing content about the White Helmets that is consistently republished elsewhere in the cluster.

Government Funded Media. Within the alternative media ecosystem, there are also social media accounts and websites associated with geopolitical entities — including the Syrian, Iranian, and Russian governments as well as Hezbollah. In particular, Russian government-funded media outlets (i.e. SputnikNews and RT) play a prominent and multi-faceted role within this ecosystem. Some articles that appear on RT also appear on other web domains in the graph. The most prominent authors in this space have published articles on and/or been interviewed by RT. Additionally, RT and/or Sputnik content (videos) appear embedded in articles that are hosted elsewhere in the graph. This activity by Russian government-funded media is perhaps not surprising considering that Russia is an ally of Syria in the conflict, and the White Helmets can be viewed — both directly through their response activities and indirectly through their work to document the impacts of military actions on civilian populations — as a threat to their mission.

Shared content appears across ideologically diverse sites. We conceptualize this as an “echo-system” where seemingly disparate websites are often sharing the same content, but enclosed in very different wrappers, so that the content appeals to different identities (politically right or left or libertarian, U.S. military veteran, pro-Russian, anti-Semitic, anti-imperialist, pro-Muslim, anti-Muslim, etc.)

One way to look at this is that there are efforts by geopolitical actors to cultivate different online communities to spread their messages. We’ve documented this previously in the U.S.-focused discourse related to Black Lives Matter activism. In the case of anti-White Helmets discourse, we see this cultivation/activation occurring within alt-right and conspiratorial libertarian online communities as well as left-leaning activist communities. These efforts have “organic” components — in other words, the communities of participants in this online discourse are not made up entirely of “bots” or active disinformation agents, but instead consist of diverse individuals and organizations who are driven by a variety of different motivations (including political, financial, and ideological). This underlying structure — a seemingly diverse, but interconnected set of websites (and social media actors who promote these websites) — is effective in several ways. First, it can create a sense of false triangulation for those who use this ecosystem to “do their own research” on the topic. Second, it functions to draw people with different commitments and values into a shared narrative. And third, it effectively distributes responsibility for the political messaging across the network, making the effort somewhat resistant to claims of orchestration — at least from a surface perspective that fails to recognize the underlying mechanisms of information shaping.

It is important to highlight that many of the people who participate in these conversations, including some of the journalists involved, are sincere believers of the ideologies that they are accessing and spreading through these online channels. Our broader research suggests that this is not a simple system with a small set of state-sponsored actors orchestrating the spread of anti-White Helmets messages, but a complex system where state-sponsored media shape, amplify, and resonate with voices of political activists and organizations, journalists and bloggers, and “independent” media. The graph we present here, however, demonstrates that though the media websites involved in spreading these anti-White Helmets messages position themselves as independent, their content-sharing practices are interconnected with political messaging from Syrian, Russian, and Iranian government-funded media.

Like what you read? Give Kate Starbird a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.