OFFF Report : Aaron Koblin On Data Visualization

OFFF Report : Aaron Koblin On Data Visualization
Piers Fawkes, PSFK
  • 14 may 2009

This work was spotted by Wired Magazine who commissioned him to research futher. He crunched data so that he could look at flight patterns within 200m x 200m space.

Koblin described the output of this work:

“Here we can see different life patterns with different airports.”

Disaster Data


When Konlin joined Yahoo! he found it difficult to get departments to access data. After he created a representative analysis of traffic accidents in major US metropolises departments were more keen. With the new data he visualized the email system where he could visualize email. Then he started to look at real time patterns with keywords in emails. He showed how terms like Hurricane Felix spread as the storm thrusted towards the US coastline.

Sensible Cities

koblin sensible cities.jpg

This work led him to AT&T where he created the project ‘Sensible Cities’ where he visualized the real time feeds of long distance and IP telephone data. He helped AT&T see how certain cities had relationships with other cities. He even suggested that you could predict an American city’s demographics dependent on the phone-relationships made between those cities and others.

aaron koblin text message.jpg

Koblin’s work on telephone use led to a project where he studied text massage usage in Amsterdam on New Years Eve and mobile phone location on Holland’s Queen’s Day.

Amsterdam SMS messages on New Years Eve from Aaron on Vimeo.

TV Allegiance


At Google Aaron Koblin has been looking at the data behind TV advertising. Because data is readily available through the market, Koblin can easily find out who wis watching what, when and for how long. A comparison of data between CNBC and Playboy shows that here was a higher allegiance to CNBC while Playboy viewers watched for a short amount of time before leaving.

He showed how he was trying to turn channel data into noise. He showed how different channels had different sound patterns depending on their viewer habits and allegiance.


10,000 Sheep


In the latter part of the lecture Koblin showed how he used Amazon’s crowd-sourcing technology Mechanical Turk to gather data. With this system people complete very simplistic tasks and Koblin wanted to visualize this data.

For his first project he asked the crowd to draw a sheep on a special web page for 1 US cent. He collected lots of sheep – 10,000 in fact. On the site SheepMarket users can view every single sheep – and even watch how they were drawn (as the computer re-renders it with the same strokes as the original submitter).


Koblin told the audience:

“I set out to see how people would draw sheep but in the end I could see people. We often see the end results of a creative talk but we rarlet see how other people draw.

A learning (that we also heard in the talk with Multi-Touch) was that people like to erase and start again. Technology should allow this. Koblin also tested Mechanical Turk by asking users to paint tiny sections of a $100 dollar bill. The project can be found here.

Radiohead’s House Of Cards Music Video

house of cards.jpg

Koblin’s next project was for a video for Radiohead and their track House of Cards. Using laser robot scanners collecting distance points and bouncing light triangles at the Tom Yorke’s face he helped create a powerful piece of moving image.

In terms of tips for would-be-data-visualizers would want to learn, he gave a few lessons he learned from the project:

* Looking at something ordinary in a new way can make it extraordinary
* Using multiple visualization techniques moves the work from tech demo to storytelling
* Think about the data, not the real world
* You don’t have to use all the data
* Let the data free

On the last point, Koblin explained how he released the data and some visualization applications to allow anyone to adapt the “beautiful data”. Several of the user-generated versions can be found on a YouTube group here.

Bicycle Made For 2000

bike for 2000.jpg

Koblin also described his Bicycle Made For 2000. Using the first song sang (Daisy) by a computer at Bell Labs (the same one that is sung when the computer xxx dies in Space Odyssey), he took the track and cut it into many pieces. Then he asked thousands of people using Amazon’s Mechanical Turk to listen to those pieces and try to sing what they heard. By averaging out the 1000s of responses, he managed to create this:

On the site here – you can drill down to hear every contributor’s note.

At the end of his talk in Lisbon, Aaron Koblin received what can only be described as a rock-star’s standing ovation by a crowd of over 3,000 attendees.


Aaron Koblin

+Aaron Koblin
+data visualization
+fitness / sport
+Market Research

Future Of Health: Rethinking Healthcare Engagement And Service Delivery

Fashion Today
Advertising Today
No search results found.