Current Issue
This Month's Print Issue

Follow Fast Company

We’ll come to you.

2 minute read

Innovation By Design

We're Pushing Nature's Network Architecture To A Catastrophic Crash

Nature can compensate for failure. Until one too many things go wrong.

Scientists tell us that the world is ending, but when you look around, things don’t seem so bad! There may have been wildfires and droughts in California, sure, but Whole Foods still has my avocados, orange juice, and almonds. Maybe life, to paraphrase Jeff Goldblum, will just "find a way" no matter what we do to it.

It’s a nice thought, until you watch Network Earth—and realize that in nature, things tend to get only a little bit bad before, suddenly, the whole system collapses.

Based upon an esoterically titled research paper Universal resilience patterns in complex networks, Network Earth is a five-minute data visualization by Mauro Martino, head of the Cognitive Visualization Lab for IBM Watson, and Jianxi Gao, a researcher at Northeastern University.

It presents its staggering thesis visually: that network relationships connect all species on Earth. Remove one or two of those species, and the network compensates as if nothing has happened. Remove enough, and an entire ecosystem dies.

The project is what Martino calls a "data-film," which is able to visualize complicated scientific and statistical ideas through animation.

"From my ‘director’ point of view, the paper is the script from which to develop a film," Martino says. The ensuing challenges range from writing the voice over to coding the visualizations themselves. While Martino can produce impressive 2-D charts in a few days or even hours, as a director of the fully animated video he worked an undisclosed number of hours, eventually cutting half of all the content he generated.

"One of the biggest problems is always simplification," he says, referring to both the script itself and the graphs he included, working in tandem to make the data approachable, but not crop the numbers out of the piece. So while, at 2:00 in, any data scientist will notice the bipartite network—a complicated graph showing the interrelationship of two groups—the narrative doesn’t attempt to label or classify the graph itself.

Instead, the narrator is able to explain what's happening in the graph (in this case, showing ant species in a network) as it breaks down on the screen, saying, "as we remove ant species at random, one by one, the connections in the network grow weaker and start to disappear." As you watch, the connections break off—and it just sort of innately all makes sense. The graphs almost feel like a citation to the story.

These visuals are purposefully complicated, and purposefully full of hard numbers. Martino calls this data "the language of science." And he doesn’t want to hide or mask it behind cutesy clipart.

"I do not like the point of view of a fluid narrative," he says. "I would like to force the viewers to take on more critical roles." Martino points to 2:46, as the camera sits in front of four lines for 17 seconds—an eternity in video—as proof of this tendency. Each line represents the point of utter collapse in an environmental network, a critical moment where every node in the network dies. It's a simplistic image, but it's horrifying in the context of the story.

"The image of four lines is not as seductive as a dynamic network or a 3-D planet full of links," he says, "But I tried to force people to look at the reality, that the beauty of a discovery can be found in predicting when a simple line breaks, that the beauty is in understanding what's behind that line." Science, math, and life itself.