Visual Investigative Scenarios: The Map You Draw By Hand

A Drawing Tool For Investigations

In the early two-thousand-tens, the Organized Crime and Corruption Reporting Project had a specific recurring problem. The journalists in the network were good at finding connections. They were not so good at communicating those connections to readers. A story might involve a person who owns a company that funds a politician who awarded a contract to a relative who works for another company that is offshore in a jurisdiction with weak transparency rules. The story was real. Saying it as a sentence made it sound like a thriller. Showing it as evidence required something different.

The standard journalism response was to build a graphic. The graphics team at a large newspaper would take the reporter's notes and produce an illustration showing the people, the companies, and the connections. The illustration would be hand-designed by a graphic artist. It would be accurate. It would be beautiful. It would also cost a lot of money and take days to produce. A small newsroom doing investigative work on a regular basis could not afford this kind of custom illustration for every story.

[calm]

The consortium thought about this problem and came to an unusual conclusion. The tool they needed was not a more powerful version of Gephi, the algorithmic graph software. Gephi was good for understanding the data, but its outputs looked like network science papers. The tool they needed was closer to a drawing program, but specialized for the kind of drawings investigative journalists actually make. A drawing program where the elements were people and companies, the connections were specific kinds of relationships, and the output was a finished illustration suitable for publication.

They called the project Visual Investigative Scenarios, often shortened to VIS. The first version was released in two-thousand-fourteen. It was a web application, free, open source, designed specifically for the act of building investigative diagrams by hand. The tool was deliberately not algorithmic. The reporter dragged elements onto a canvas, arranged them as they saw fit, drew connections between them, labeled everything, and exported the result as a publishable image. The visual decisions were the reporter's. The tool just made them easy to execute.

This is a different model from what Gephi offers. Gephi automates the layout. VIS leaves the layout entirely to the journalist. Gephi works well when you have hundreds of entities and want the algorithm to find structure. VIS works well when you have a dozen entities and want to tell a story about them. The two tools occupy different parts of the investigative toolkit, and most working journalists end up using both, for different purposes, at different stages of an investigation.

Why Manual Beats Automatic For Storytelling

To appreciate the design choice in VIS, you need to understand a specific failure mode of algorithmic graph layouts. The algorithm finds the layout that satisfies its internal criteria. The criteria are usually some combination of node spacing and edge length. The algorithm does not know which entities are important to the story. The algorithm does not know which connections deserve emphasis. The algorithm does not know what the article is going to argue.

The result is that an algorithmic layout produces a picture that is structurally accurate but argumentatively neutral. The picture shows the data. It does not show what the data means. The reader has to figure out what the picture is trying to say. The reporter has to caption it carefully. The picture supports the story but does not tell the story.

A manually arranged picture works differently. The journalist places the protagonist at the center. The protagonist's allies cluster nearby. The protagonist's adversaries are placed across the canvas. The chain of ownership that the article is exposing runs along a clear visual line. The reader looks at the picture and immediately sees what the article is about, before reading a word.

This is craft work. It is design work. It is closer to what a graphic artist does than to what a network scientist does. The journalist is making editorial choices about emphasis, hierarchy, and clarity. The choices are the same kind of choices a writer makes when ordering paragraphs in an article. The early paragraphs introduce the topic. The middle paragraphs build the argument. The closing paragraphs deliver the conclusion. A well-designed investigative diagram works the same way, with the reader's eye guided through the same arc.

VIS provides the canvas on which these editorial choices can be made. It does not impose any algorithmic structure. It just makes it easy to add entities, draw connections, style them appropriately, and export the result. The intelligence is supplied by the journalist.

The Specific Vocabulary

The thing that makes VIS more than just a generic drawing program is its vocabulary. The tool has built-in element types specifically for investigative journalism. There are entity types for people, companies, government bodies, vessels, events, and assets. There are relationship types for ownership, employment, family relationships, financial transactions, and contractual relationships. There are styling conventions that make different kinds of entities visually distinct.

The vocabulary is loose enough that the journalist can adapt it to any specific investigation. The vocabulary is constrained enough that the resulting diagrams have a consistent visual language. A reader who has seen one VIS diagram can immediately understand a second one, because the symbols and conventions are familiar. The vocabulary becomes a shared language across the publications that use the tool.

[serious]

This is similar to what the Follow The Money schema does for data. Follow The Money provides a shared vocabulary for talking about investigative subjects. VIS provides a shared visual language for showing them. Both have the effect of making investigative journalism more legible across publications, across countries, across investigations. The reader of a story in Romania can look at the diagram and understand the structure of the corruption being exposed, even without speaking Romanian, because the visual language is consistent.

The Specific Workflow

The workflow with VIS is straightforward. You start a new investigation. You add entities one at a time. For each entity, you give it a name, a type, and any relevant identifying details. You can attach photographs, link to source documents, add notes. The entity becomes a card on the canvas.

Once you have your entities, you draw the connections between them. Each connection is also typed and labeled. Ownership is one kind of connection. Family is another. Employment is another. You can add notes to each connection, explaining the evidence behind it.

As you work, you arrange the canvas to make sense. The protagonist might go in the middle. The protagonist's supporters might cluster around them. The adversaries might be on the other side. The flow of money might be drawn in a specific direction. The visual layout is editorial. The journalist is composing.

When the investigation is complete, the diagram is exported as an image or as an interactive web embed. The image can be published in the printed version of the story. The interactive embed can be used on the website, with hover-over details for each entity and connection. The reader engages with the structure of the investigation at whatever depth they want.

The Compositional Power

VIS is most useful when combined with other tools, rather than used in isolation. The pattern that works well is to start with algorithmic exploration, then move to manual composition. You take your data. You load it into Gephi. You run a layout algorithm. You see the structure. You identify the clusters, the bridges, the outliers. You understand what the data says.

Then you switch to VIS. You start with a blank canvas. You add only the entities that matter to the story you are telling. You arrange them deliberately. You draw only the connections that the article will discuss. The Gephi exploration has shown you the truth. The VIS composition shows the reader what part of the truth matters for this article.

[calm]

This division of labor is the same division that exists in any kind of investigative work. The research phase is messy, broad, and exploratory. The publication phase is clean, focused, and decisive. The research uncovers everything. The publication shows the specific things that matter to the argument. The tools that serve the research phase are different from the tools that serve the publication phase, and a working journalist uses both.

For an investigation into mineral exploration permits in a Swedish county, the workflow might be to build the full ownership graph in Gephi, find the structure, identify the key entities, and then construct a clean VIS diagram showing just the parts the article will discuss. The Gephi work is internal to the investigation. The VIS work is what readers see. Both are journalism, but they serve different functions.

The Limitations

VIS has limitations that are worth knowing about. The tool is good for diagrams up to about thirty or forty entities. Beyond that, the manual arrangement becomes unwieldy and the resulting diagram becomes too dense to read. For larger structures, an algorithmic tool is the right choice, or the structure should be broken into multiple smaller diagrams.

The tool is also less flexible than a general-purpose drawing program. A graphic artist working in Illustrator can do things VIS cannot. The constraint is part of the value, because it keeps the diagrams consistent across investigations, but it also means that occasional needs cannot be met within the tool. For these cases, the workflow is to export from VIS and then refine the result in a more flexible drawing program.

The license is the GNU Affero General Public License, which is the same family as DocumentCloud and Datashare. The tool can be self-hosted, but the consortium also runs a public hosted version. For most working journalists, the hosted version is the right choice. The self-hosted option is available for cases where the investigation data should never leave the reporter's infrastructure.

What This Has To Do With Working Journalists

The practical use of VIS for a working reporter is to produce the publishable diagram for any story that involves networks of people and companies. The diagram is the visual centerpiece of the article. The diagram is what readers remember after they have forgotten the details of the prose. The diagram is what gets shared on social media, what appears in other publications discussing the story, what becomes the symbol of the investigation.

[serious]

Investing time in making the diagram excellent is investing time in the durability of the story. The article will be read for a few days after publication. The diagram will be referenced for years afterward. The diagram is the asset that compounds. Doing it carelessly is leaving value on the table. Doing it well is making the journalism more impactful than it would otherwise be.

For a Swedish reporter working on a story about mining permits and corporate ownership, the workflow is to do the research in whatever tools serve the research, then to compose a clean diagram in VIS showing the specific ownership structure the article is exposing. The diagram should be readable in fifteen seconds. The diagram should communicate the central claim of the article visually, before the reader has read any of the prose. The diagram should be designed, not just generated.

The Larger Argument

The argument that VIS represents is that investigative journalism is, at its best, a visual practice. The work happens in documents and data. The argument happens in prose. But the central insight is often a structural claim about who is connected to whom, and the structural claim is most clearly expressed as a picture. The picture is the journalism made visual.

This is not always how journalism is taught. The dominant model of journalism education has been about writing. Words on the page. Sentences arranged into paragraphs. The visual elements are sometimes treated as decoration, supplied by a separate graphics department or omitted entirely for budget reasons. The investigative tradition has been particularly text-heavy, perhaps because the foundational works of the field, like All The President's Men, were prose narratives.

The visual investigative tradition is a more recent development. The tools have made it accessible. The Panama Papers diagrams, the various corruption investigations of the last decade, the network analyses that accompany major stories, have all contributed to a sense that the picture is part of the journalism, not separate from it. VIS is one of the tools that has made this practice possible at scale. The picture has become legitimate.

[calm]

For a working reporter today, the implication is that visual composition is a skill worth developing. The reporter who can build a clear, readable, compelling investigative diagram is producing a different kind of journalism than the reporter who can only produce prose. The diagram is not a replacement for the prose. The diagram and the prose work together. But the reporter who can do both is doing more complete work. The tools exist. The skill takes practice. The investment pays off in every investigation that has a structural claim at its center. VIS is one of the places to start practicing.