Software developed at Stanford helps uncover connections in the Panama Papers
Earlier this month news agencies around the world began releasing stories based on the largest leak of documents ever, the Panama Papers (https://panamapapers.icij.org/). The data visualization tool that journalists used to uncover connections between people, accounts, shell companies, and assets in this massive data set originated at the Humanities + Design (http://hdlab.stanford.edu) research lab in Stanford’s Center for Spatial and Textual Analysis (http://cesta.stanford.edu) — a product of humanities thinking applied to network analysis.
The visualization tool developed at Stanford, Knot (paper: http://dl.acm.org/citation.cfm?doid=2499149.2499174 demo application:http://knot-dev.herokuapp.com/investigate.html), was the result of a collaboration between humanities researchers at Stanford, the SUL/CIDR Academic Technology Specialist for CESTA, designers from DensityDesign Research Lab at the Politechnical University of Milan (http://www.densitydesign.org), and two French students, Sebastian Heymann and Romain Yon. Heymann and Yon commercialized the software in late 2012 as Linkurious (http://linkurio.us), the tool used by members of the International Consortium of Investigative Journalists on the Panama Papers.
Knot is one of several data visualization tools (http://hdlab.stanford.edu/) developed within "Mapping the Republic of Letters" (http://republicofletters.stanford.edu), a study of the intellectual communities of 17th and 18th century Europe and early America based on metadata from correspondence, travel, and publications. There were five Stanford graduate students and three humanities faculty involved in the August 2012 workshop, “Early Modern Time and Networks” (http://athanasius.stanford.edu/) that provided the content and research questions for Knot.
Stanford researchers working with these large complex and fragmentary historical data need tools that support a human-scale investigation of the people, places and things in these networks. Knot, for example, eschews the predictive network analysis method more common in the social sciences, in favor of a ‘close-reading’ method where the network is explored one node at a time or in small filtered groups. The network exploration begins with a keyword-in-context query of the data. For 18th century correspondence, a search for "Paris" for example, offers Paris as a birthplace, a death place, the source of a letter, or its destination. Selecting "Paris as birthplace" would result in all of the letter authors and recipients born in Paris. Additional filtering choices can be made to expand or limit the scope of inquiry. The design of the research experience with Knot is based on the concept of following a train of thought from one node to another or, thinking through data.
What does 18th century correspondence have to do with 21st century journalism? Historians and journalists both work with incomplete, heterogeneous data sets and need tools to help them piece together a narrative based on fragmentary evidence. The questions both groups need to ask of data often require additional research and a fine-grained exploration. Knot is an example of injecting humanistic method into technology to help make sense of complex relationships. In this case, the people and organizations behind more than 11.5 million financial and legal records.
Many ‘big data’ challenges require collaborative 'small data' investigations. The success of Linkurious is evidence that humanistic inquiry, grounded in interpretation, has much to contribute to the development of technologies if they are to help us reveal ambiguity and paradox, allowing human-scale exploration of complex systems. Work in the vein of Knot will continue with the latest Humanities + Design project, Fibra, funded by the American Council of Learned Societies (http://www.acls.org/). To learn more about Fibra, visit hdlab.stanford.edu.