Every year illicit activities around the world generate trillions of U.S. dollars in estimated profits. When sums are large enough, the person or groups involved must find ways to control those funds without attracting attention. The cash must be laundered.
Methods of doing so vary. But common among them is the need to filter dirty money through legal business structures to obscure its true source, mixing it up, or "washing" it, with enough clean tender in the process.
This project looks at a network of companies and assets allegedly used to do just that. It is part of an ongoing investigation into foreign organized crime and money laundering activities in the U.S.
The contours of this network are detailed in thousands of pages of corporate filings, federal records and court documents.
The network graphs below visualize the connections revealed in those records. They measure the links between individuals and companies over time, helping map fiduciary and other business relationships between them and various firms, properties and other assets.
In general, such graphs allow users to spot connections and trends among large datasets that they might otherwise fail to notice. A network graph is thus most useful when casting a wide net.
But as with all methods of data analysis, network graphs are rarely conclusive in and of themselves. They are means of focusing investigative resources and testing hypotheses.
In the case of the network depicted here, for example, connections between convicted money-launderers and certain individuals or companies are not necessarily evidence that those individuals and companies were involved in laundering illicit proceeds. They are merely evidence of the potential for involvement.
Accordingly, all names associated with this investigation will remain anonymous until it is complete.
Using the network visualization software Gephi, one can run algorithms to shape networks by various attributes. Time intervals can also provide views of the network as it develops — in this case, mapping the assorted businesses, individuals and properties that show up within it over time.
Those entities appear as circles, or "nodes," the connections, or lines, between which are called "edges."
Below is one key portion of the network in the years 2005 to 2007, before many of its business connections flourished.
Over time the network matures and expands as more connections come than go. This next snapshot, covering 2012 through 2016, shows the extent to which that same core network sector has grown.
The sizes of nodes have increased proportional to their roles connecting other people, companies and properties within the network. This measure of a node's ostensible prominence is called its "betweenness centrality." It's measured by the number of shortest paths from all sides of the network to all others that must pass through that node.
Explore the full network yourself using Sigma, a JavaScript library for drawing network graphs:
All data was manually retrieved from public records. I then parsed and formatted it using R and mapped the network connections in Gephi, rendering the interactive version using the Sigma.js plugin. From there I exported snapshots of the network as SVGs and isolated them in Adobe Illustrator. I built this page with Bootstrap.
Special thanks to instructors Peter Aldhous and Jeremy Rue, UC Berkeley Graduate School of Journalism, the creators of Sigma, @jacomyal and @Yomguithereal, and the creator of the Sigma.js exporter plugin for Gephi, Scott Hale of the Oxford Internet Institute.
Christian Stork, 2016