A social network analysis of influence operations

Ben Sohl

Background

The world is currently undergoing a technological revolution.  The advent of the internet, like other great innovations in human history such as the plow, the printing press, and the combustible engine, is changing the nature of just about every human endeavor.  Foreign policy is no different.  There is currently a transformation in the nature of the conduct of foreign affairs happening in front of our very eyes. 

In the realm of foreign policy, the revolution in communications technologies has created new avenues of influence for competitor states.  No longer logistically difficult, foreign states can use information warfare to change and shape public opinion in states they target.  These campaigns have come to be referred to as “influence operations.”

Public opinion is the lifeblood of democracies.  Therefore, campaigns to influence public opinion using newly developed information technologies represents a significant innovation of coercive power.  This innovation has quickly emerged as critically important to a state’s coercive toolkit.  In 2016, the United Kingdom held a referendum on whether to leave the EU.  “Leave” narrowly won with 51.9% of the vote[1].  Several states, like Russia, had core security interests involved in this vote – the stability or instability of the EU.  Russia’s coercive tool of choice to influence this vote was not traditional tools of power like military threats, but rather an influence campaign[2]. 

Likewise, the outcome of the 2016 US presidential campaign would directly affect the core interests of many states, again including Russia.  This includes presidential policy on NATO, the EU, and US sanctions on Russia over the conflict in Ukraine.  Once again, the Russian state resorted to an influence campaign to shape the outcome of this event. 

Research questions

In 2017, I published an article titled “Influence Campaigns and the Future of International Competition.”[3]  The article contained two hypotheses: the first was that democratic states are uniquely vulnerable to influence campaigns and the second was that the success of Russia’s recent influence campaigns would result in the increased proliferation of these activities throughout the international system.  The goal of this project is to investigate the veracity of these two hypotheses and with it, use social network analysis to better understand the nature of influence campaigns. 

Why Social Network Analysis?

Social network analysis provides us with a unique set of tools to understand the nature of influence campaigns.  Over time, these campaigns have become more complex, with a multitude of actors taking part.  These actors have various relational ties with one another, from victims, to perpetrators, to surrogates.  Understanding not just the composition of influence campaigns, but the relationships between the actors is critically important to understanding how they function. 

Social network analysis will also provide us with analytical tools to better understand the nature of influence campaigns.  Using attribute data, we will be able to interrogate the relationship between authoritarian states and democratic states within the context of influence campaigns.  This data analysis will provide us with interesting insights into who these campaigns are targeting and why they are targets. 

Questions

In order to build on our understanding of influence campaigns, this project will look at several questions.  These questions are below.

1.      Have influence campaigns increased over time?

2.      Are there clusters within influence campaign networks? 

3.      Can we identify perceived spheres of influence based on cluster analysis and ego networks?

4.      Do betweenness measures identify states who use a high number of proxies or engage in more influence campaigns?

5.      Are influence campaign networks homophilous or heterophilous based on whether they’re authoritarian or democratic states?  And what can this analysis tell us about who is being targeted and why?  For example, do authoritarian states use influence campaigns against other authoritarian states or only against democratic states?

Data and Methodology

In order to interrogate our questions, we will need to incorporate several datasets.  The first dataset is one that contains all major influence campaigns since 2016, who engaged in the campaign, who was the target of the campaign, and whether there was a proxy relationship to another state. This dataset will be done on an annual basis to identify patterns over time and on a non-annual basis with the number of total campaigns between states to view overall trends.  The dataset will also be directional. 

Given the limited number of major influence campaigns conducted over the last four years, assembling this dataset will be viable.  To begin, we will start with the data compiled in the “Authoritarian Interference Tracker”[4] built by the Alliance for Security Democracy, housed in the German Marshall Fund.  Next, we will include the dataset found in the report “The Global Disinformation Order,”[5] as part of the Computational Propaganda Research Project by the Oxford Internet Institute at Oxford University.  From there, we will do independent research to fill any holes in the data.

Secondly, we will utilize Freedom House’s “Freedom in the World 2019” report to create attribute data based on the amount of freedom in a given state.  This attribute file will allow us to include the democratic or authoritarian nature of a state into our analysis. 

With our data in place, we will conduct a series of tests to better understand these campaigns and what they mean.  First, we will conduct a Newman Girvan analysis to identify clusters within the network.  This cluster analysis should help us draw out any small world networks of these campaigns including the primary aggressor, their proxies, and the victims.  Using cluster analysis and ego networks, we will be able to identify who the hegemons in the global system are and where their perceived spheres of influence extend to.  We will also be able to see how they are prioritizing their influence campaign assets based on the number of campaigns they’ve waged against different states.  We can use centrality measures such as betweenness to illustrate who lies at the center of these networks.  This will be helpful particularly if proxies engage in campaigns on behalf of multiple actors.  Once they’ve been identified, separating hegemons by ego networks will allow us to better understand their behavior on an individual basis.

In order to analyze our data further, we will dichotomize our attribute data file to narrow the freedom index into states that are democratic and those that are authoritarian.  We will then use an E-I index analysis to determine how heterophilous or homophilous our data is based on the style of government.  This will provide us with key information.  If the dataset is highly homophilous, that will tell us that states of a similar level of democracy or authoritarianism engage in influence campaigns against each other.  If it is heterophilous, it provides evidence towards our hypothesis that authoritarian states are more likely to engage in campaigns against democratic states.  With this information, we should be able to draw out important conclusions. 

By changing the size and color of the different nodes, each representing a country, we will be able to provide this critical analysis in an easy and digestible presentation.  Taken together, the use of social network analysis will greatly increase our understanding of the nature of influence campaigns. 

---

[1] BBC News, “EU Referendum Results.” Accessed October 19, 2019. https://www.bbc.com/news/politics/eu_referendum/results

[2] David Kirkpatrick, “Signs of Russian Meddling in Brexit Referendum.” New York Times, November 15, 2017. https://www.nytimes.com/2017/11/15/world/europe/russia-brexit-twitter-facebook.html

[3]Ben Sohl, “Influence Campaigns and the Future of International Competition.” The Strategy Bridge, September 12, 2017. https://thestrategybridge.org/the-bridge/2017/9/12/influence-campaigns-and-the-future-of-international-competition

[4] Alliance for Security Democracy, “Authoritarian Interference Tracker,” Accessed October 15, 2019. https://securingdemocracy.gmfus.org/toolbox/authoritarian-interference-tracker/

[5] Samantha Bradshaw and Philip Howard, “The Global Disinformation Order.” Computational Propaganda Research Project, Oxford Internet Institute, University of Oxford.  Accessed October 19, 2019. https://comprop.oii.ox.ac.uk/wp-content/uploads/sites/93/2019/09/CyberTroop-Report19.pdf