Visualization tool prototyped by Stanford humanities scholars aids the investigation of ‘Panama Papers’

The fallout from the world’s biggest data-based journalistic investigation known as the Panama Papers continues to reverberate across the world. Millions of documents have revealed in great detail how political leaders, celebrities and others used offshore bank accounts to hide their wealth and avoid paying taxes.

Nicole Coleman and Dan Edelstein (Credit: Pui Shiau, Office of International Affairs)

Few may realize that the revelations were made possible thanks in part to a visualization tool prototyped by humanities scholars at Stanford’s Center for Spatial and Textual Analysis (CESTA).

The International Consortium of Investigative Journalists (ICIJ) used a commercialized version of the software, Linkurious, to analyze 11.5 million financial and legal records from Panama-based Mossack Fonseca, the world’s fourth-largest offshore law firm, that were anonymously leaked to a German newspaper and ultimately shared with the investigative journalists.

Researchers at CESTA originally developed Knot, the graph visualization software, to enhance searches in “Mapping the Republic of Letters,” a Stanford project launched in 2009 to explore connections across intellectual communities in Europe and North America during the Enlightenment. The project took the metadata of 50,000 letters written by 7,000 individuals to create “big picture” maps that illustrated who communicated with whom and about what.

In 2012, French computer scientists SEBASTIEN HEYMANN and ROMAIN YON, and DensityDesign Research Lab members at the Politecnico di Milano, participated in a Stanford workshop, “Early Modern Time and Networks,” that led to Knot’s development. When the project ended, Heymann used the software’s open source code to launch Linkurious, said Academic Technology Specialist NICOLE COLEMAN, research director of the Humanities+Design lab in CESTA.

“The fact that this tool has broader use than the ‘Republic of Letters’ demonstrates that [a humanist’s] way of thinking about information has a really important part to play in the world of big data and how we analyze it,” she said.

Stanford French Professor DAN EDELSTEIN said humanities scholars work with incomplete and fragmentary data and need computer models that explore connections rather than impose a predetermined model on a network.

“The orientation of computer scientists is that they will build a script to solve a problem. But we didn’t have a problem. We wanted to explore. It’s more about using data to reveal and make connections,” said Edelstein, the William H. Bonsall Professor of French.

Researchers involved in the “Republic of Letters” needed a visualization tool that allowed them to make judgments and connections based on surviving information. Imposing metrics on incomplete facts “would be worse than senseless, it would be misleading,” Edelstein said. Instead, Knot allowed users to develop open-ended visual network graphs that led to the discovery of direct and indirect relationships among people, places and events.

Heymann, who had already helped develop a visualization software called Gephi, took this approach in developing Linkurious to make sense of highly complex, fragmented data. In 2015, the investigative journalists’ consortium partnered with Linkurious to uncover a worldwide tax fraud scheme that became known as Swiss Leaks.

This year, the journalists used Linkurious to analyze the Panama Papers, the biggest data-based journalistic investigation in history.

According to Coleman, the success of Linkurious shows that interpretive humanistic inquiry is relevant when developing technology that seeks to piece together a narrative based on incomplete evidence. Journalists used the tool to uncover connections between people, accounts, shell companies and assets that reveal how how rich people exploit tax havens to hide their wealth.

Despite Linkurious’ success, Edelstein said, society often ranks quantitative analysis over qualitative because there is an assumption that a quantitative approach provides better answers.

“The truth is, very few real-world problems, even scientific problems like climate change, are going to be solved purely with quantitative answers,” he said.

Humanistic thinking is essential for real-world decision-making because it involves weighing different plausible outcomes, Edelstein said. “I think tools like [Knot] reward that kind of approach because they don’t seek to directly answer a question. They seek to guide you through the steps of finding the answer yourself,” he said.

Edelstein and Coleman would like digital tools, grounded in a humanistic approach, to become a standard part of the historian’s toolkit. Currently, they are working on a new open-source project called Fibra funded by the American Council of Learned Societies that offers more flexible social network visualizations.

“Scholars already use databases,” Edelstein said. “Our goal is for these tools to become an extension of that and allow scholars to explore and make sense of the larger pools of data we’re all swimming in.”