What big data fails to tell us, but design can help us understand
By Allie Rocovich
Dietmar Offenhuber, head of the Information Design and Visualization graduate program at Northeastern, spoke about what we can learn from big data at Superstition vs. Simplification Visual Data Mining & It’s Pitfalls, on March 3. For those of you that missed it, here are a couple takeaways about how our brains work to process cause and effect, how computers process data, and how information design can help us realize what it all means in the most measured age in history.
Reasoning is often thought of as putting together clues and coming up with a causal explanation for what has happened. People do this all the time. If you see someone on line to buy a sandwich, it is safe to assume this person is hungry. Our brains are constantly noticing things and assigning reason to them, however our brains’ reasoning isn’t always true.
Similarly, it’s easy to see patterns in things that don’t actually exist. The brain is a self-organizing hierarchical system of pattern recognizers, but patterns can be very misleading. This is a problem with our brains, and as Dietmar Offenhuber also pointed out, it’s a problem with big data.
For example, we see faces everywhere. And you could say that this is something that is a human quality, this is the way our brains are built — to see faces everywhere. Computers have the same problem. We always run into this pattern of finding patterns that lead us to the wrong direction.
Alexander Riegler said this is a problem with artificial intelligence. We are too much focused on pattern recognition. And recognizing patterns is not the same as understanding why something is happening. He said, “There is a close relationship between pattern discovery and superstition since humans and animals alike excel at finding structures where there are none.”
In 2008, Chris Anderson suggested that we have reached the end of theory, in a controversial article posted on Wired.com. He argues that in the most measured age in history, we no longer need to understand why things are happening; we can succeed in our endeavors by recognizing patterns alone. Anderson writes, “[Having so much data] forces us to view data mathematically first and establish a context for it later.” He uses Google as an example:
“For instance, Google conquered the advertising world with nothing more than applied mathematics. It didn’t pretend to know anything about the culture and conventions of advertising — it just assumed that better data, with better analytical tools, would win the day. And Google was right.”
Psychologist Burrhus Frederic Skinner’s “Superstition of the Pigeon Experiment” introduces “Superstition,” a concept that unveils an issue with Anderson’s suggestion that we can use data alone to reach our goals. Superstition is a nonexistent relationship that is perceived to exist between two or more points of data.
In Skinner’s Superstition of the Pigeon experiment, pigeons were given food on a random basis. Their brains were constantly trying to find a cause and effect relationship between something they were doing and the act of being fed. One pigeon associated spinning counterclockwise in the cage twice with being fed. The next day, the pigeon spun twice in hopes of receiving food, and did not receive food. This confused him: the cause and effect relationship he perceived proved to be incorrect. This is superstition. The pigeons were unable to find a relationship because there was none — the food distribution was random.
“This is where visualization comes in,” said Offenhuber. “It is very easy to jump to conclusions with data, but it is problematic to do so. Visualization is trying to rearrange something in a way that makes sense.” Visualization can provide simplification in a world of abundant, complex data.
Offenhuber worked on a project that aimed to visually display complaints that people were making out their city in Linz, Austria. He created an LED façade, where he displayed complaints received by the city on the Ars Electronica Center. People could visualize in the public sphere what people were complaining about. On his website, Offenhuber writes, “I find this fascinating because there are now many different systems that publish such complaints on a website. But I think it’s something completely different when they’re displayed right in the cityscape.” By creating a visual representation for the experiences people were having in Linz, Offenhuber created a way to analyze the dynamic between individual citizens and the municipal administration. The moving lights represent that fact that the city is always in motion and never in a perfect state of equilibrium. There’s never a time when everything has been cleaned up, maintained, or repaired, and the visual demonstrates that this may never be the case.
Offenhuber said, “In the US, people just casually point out things that should be fixed. In Austria, people would write page long letters with sarcasm. [In a] public space, we can’t choose who we run into. This is what I tried to convey here. People who submit a complaint are made cognizant of the fact that it’s then immediately visible by anyone in the cityscape. If people knew their complaints were being broadcasted in public, how would they react?”
Click here to learn more about Offenhuber’s work with the Ars Electronica Center.
Dietmar Offenhuber has worked on information design projects around the world. Many of his projects focus on the relationship between visual representations and urban governance. To learn about other influential speakers coming to campus this semester, check out our Events page.
