07
Nov
2016
21:50 PM

Wikileaks and the Podesta Emails

Thank goodness for Julian Assange and Wikileaks, as well as the others who have dared fight the established political forces in this country. Thanks to their efforts, the veil has been lifted and we can all see how manipulative and crooked these folks are as they do their level best to fleece the average citizen and make themselves wealthy beyond their wildest dreams. So it is with Hillary Clinton in the 2016 campaign, as the recent hacks of the John Podesta emails have confirmed. For full details, you can start here: https://wikileaks.org/podesta-emails/.

Podesta, Hillary Clinton's campaign manager and a long-time associate of the Clintons has been exposed as a master manipulator, working with many others behind the scenes to tilt the campaign in Clinton's favor. Thanks to Wikileaks, we can see very clearly the efforts of a host of players to do everything in their power to discredit Bernie Sanders and Donald Trump in an effort to put their candidate in the White House. The individual emails lay bare the machinations of the Democratic National Committee in scurrilous detail, and make for entertaining reading. Of course, many of Hillary Clinton's supporters will dismiss any notions of wrongdoing courtesy of the rather pathetic pronouncements of FBI Director Comey, but the evidence is plentiful, regardless of the FBI's "official" position.

In this post, I'll take a network graph view of the players involved, using data from the http://gdeltproject.org. This will help shed light on the primary participants, how they interrelate, and who the "targets" of their mischief are. At some point, I'll also work up a text analysis of the email content, but that's for another post.

If you want to go directly to the network and begin interacting with it, go here: http://visualidity.com/projects/podesta-1/. In the meantime, I'll be using Gephi to filter the network so we can make some observations and draw conclusions about the participants. One of the best ways to begin examining the network is to see who the biggest players are, as measured by their connections (degrees) to others in the network. These would arise out of related mentions; in this case, I allowed GDELT to create the initial network, so I don't have full visibility into the original data. We'll have to rely on GDELT's methodology, which I'm comfortable using.

Using Gephi, we can run a variety of graph statistics to better understand the content and structure of the network. These measures will include various centrality statistics to help us understand the prominence and positioning of various network figures, plus some general network measures such as eccentricity. Eccentricity tells us how many connections it takes to traverse the entire network; in many networks we are accustomed to 4, 5, and even 6 required connections to access every other node in a network. This is the root assumption behind six degrees of separation - or six degrees of Kevin Bacon for those familiar with the Hollywood version of network connectivity. In this case, we have a very interesting finding - no one requires more than two connections to cross the network. This is due to Podesta being the fulcrum of this network, with everything flowing through his emails. We may consider this a closely connected network, but not a dense one. Density measures the proportion of nodes connected to one another; in this case it is only about 6% (.064), as most everything flows through Podesta or Hillary Clinton. In that sense, we have a rather hierarchical graph structure, with a few very large players surrounded by more than 250 smaller ones.

Here's our full network as seen on the web (after running the Force Atlas 2 layout algorithm in Gephi):

You can see the presence of a handful of very large nodes near the center of the graph; these are the primary influencers in our network. Now let's examine some of the individual nodes and their respective networks, beginning with John Podesta himself:

Note the close proximity (and very similar size) between Podesta and Hillary Clinton. This should be unsurprising, given his role as her campaign manager. Obviously, she is a central figure in most of the emails going to and from Podesta's account. We may also note the proximity of Clinton's competition in the form of Bernie Sanders and Donald Trump. Also note that while Trump has the same node coloring as Podesta and Clinton, Sanders does not. This is the result of his being in a different modularity class (cluster) based on the graph structure. In all, we have 9 different clusters in this network, although this number could just as easily been 7 or 8, depending on our sensitivity level used in Gephi.

A view of Hillary Clinton's network looks similar to Podesta's:

Note also the statistics in the left margin, which provide details about each node's influence and connections. Clinton has a degree level of 289, compared with Podesta's 287. Likewise, their centrality measures are nearly identical; Clinton's closeness and harmonic closeness are equal to 1, making her the most influential figure in the network, with Podesta close behind.

Now let's look at some secondary figures, starting with Juanita Broaddrick, one of Bill Clinton's rape accusers. Broaddrick has a degree measure of just 11, but is directly connected to both Hillary and Bill Clinton, as well as Podesta. Clearly, there was some concern over the possibility of her tainting the campaign; this might be one of the more interesting sets of emails to read.

Another interesting figure is the recently disgraced Donna Brazile, former head of the Democratic National Committee and CNN contributor. Brazile was caught feeding questions in advance to the Clinton campaign, and as such caused a bit of a stir. We know this because she has a degree level of 60, among the upper echelon of all network nodes.

Finally, let us take a look at Anthony Weiner, another notorious figure on the fringes of the Clinton campaign. Weiner, the estranged husband of Clinton confidant Huma Abedin, is best known for his phone sexting adventures, and has again appeared in the news when the FBI discovered an enormous cache of Clinton emails on his laptop while investigating Weiner on other charges. While Weiner's network is comprised of just 9 connections, they include notable figures such as President Obama, Huma Abedin, and Hillary Clinton.

These are but a few examples for how network graphs can be used to see connections within a dataset, and how they can be utilized to see otherwise hidden or obscured patterns. There is much more that could be done beyond this simple example; I invite you to play with the network and see what else is there. Thanks for reading!



You may also like