In this age of analytics and big data we focus on the speed of the computer and the size of the attached storage but we’re overlooking the most critical element. Yes computers manipulate and store huge quantities of data, but the main processing is still done by the human mind – the human mind is what turns data into actionable information. So, speed up the link between the computer and the human mind and you’ll get much better results.
“We work with the military, big corporations and people frustrated by trying to analyze large volumes of data,” said Creve Maples, Ph.D. and CEO of Event Horizon. He explained that when people come to him they have often tried all sorts of analytic techniques but haven’t gotten many useful insights into what the data really means. He described a different approach that his company uses to analyze big data. “What we’ve learned to do is present data in formats that can be quickly taken in and understood by people.” That means they use formats that go beyond traditional line graphs and scatter plots and bar charts.
[ I do lively presentations on this and related topics – www.MichaelHugos.com ]
Feed Us the Data in a Format We can Quickly Understand
According to Creve, the human brain is normally processing about 20 Gigabytes of data per second; that’s the input from the five senses and the brain handles this data in real-time. Research at Event Horizon suggests people can track up to 27 different variables as they process their 20 GB of real-time input. Consider the example of driving a car. The driver can watch the road, operate the car, talk to a passenger, adjust the car heater and radio and subconsciously be listening to the sound of the engine and feeling the vibrations of the car as it moves. The driver’s mind is taking this all in and handling it as a normal matter of course.
But then as soon as there is a slight change in the engine noise or the vibrations of the car the driver’s senses are alerted, and the driver is all over the situation trying to find out what happened to cause this sudden change.
Several years ago Creve worked with engineers at Penske Racing; Penske was loosing races but they couldn’t figure out why. They put real-time sensors on their race cars and those sensors streamed data on about 22 independent variables. They collected this data during races and then afterwards did all sorts of analysis on it. After 2 years of effort they couldn’t find answers.
They came to Event Horizon as a last attempt. Creve said he took the raw data but told the Penske engineers he didn’t want to see the analysis that had already been done. Creve and his staff wanted to approach the problem from a fresh perspective.
“We displayed the different data variables as cartoon elements. For instance, as tires heated up we showed them getting bigger, as other variables changed we changed the shape and color of the cars. We took vehicle data and rendered it visually without showing numbers.”
Once they had created these visualizations and ran them as animated sequences they found the problem on the first day. It was that the response time of the race cars' steering systems was too slow. With moving arrows they showed visually the direction of the tires and the rotation of the car steering wheel. Animated sequences clearly showed there was a small lag time between when drivers turned the steering wheel and when the car wheels actually turned. So drivers were constantly making many small adjustments and it slowed them down just enough to lose races.
Having Fun with Big Data
Imagine analyzing performance data from a new product by shrinking down and traveling freely inside the device such as this complex electro-mechanical gear assembly shown below. What better way is there to see how the parts interact and see where the friction builds up?
(visualization image courtesy of Event Horizon)
Creve described another visualization they created when working with a computer chip designer. The designer came to them and said a new chip was overheating and nobody could figure out why; the new design should have been able to dissipate the heat. So they created a 3D visualization of the chip and let people fly through the circuits of the chip. The circuits were rendered as tubes you could fly through and the walls of the tubes were colored using a range of blue for cool to red for hot with color gradations in between.
Then they brought in the wife of one of the Event Horizon scientists to try out the visualization. She didn’t know anything about chip design but she located the point of thermal failure very quickly. The area of the chip that got hottest most quickly was easy to spot as she flew through the chip. And when she examined that area closely she saw there was no way for the heat there to be drained off. It turned out the chip designer had left out a heat-sink connection in that part of the chip.
This is an example of the discoveries and delights that happen when people get a spontaneous human-machine interaction made possible by good user interface design that effectively engages our senses. Creve coined a term for this; it’s called “anthropo – cyber – synchronicity”.
We would never say “we had fun” as a reason for using a data analytics application. But fun is literally what happens when an analytics system connects with the power of our senses and allows us to suddenly understand and interact with data in a novel way.
Does this seem a bit like a video game? Could game technology help us do analytics?
[ I’m building a game to design and visualize supply chains – it’s called SCM Globe – you can try it for free; I’d love to get your feedback. ]