As Aaron Koblin walks me through the two floors of cereal bars, scooter
parking, and conference rooms in Google’s New York office, he
apologizes. “I wish I could give a better tour,” he says. “But it’s just
so huge.”
 |
| the sheep market |
Using the Processing programming language, developed by his UCLA
thesis adviser, Casey Reas, with Ben Fry, Koblin turns the messiest sets
into beautiful, if equally complex, images. These days he’s playing
with Mechanical Turk, an Amazon.com crowd-sourcing tool that pays tiny
sums for menial tasks.
For his project the Sheep Market, 10,000 users
drew left-facing sheep for two cents each. The human error made it
interesting. “Six hundred and sixty-two of them didn’t meet sheeplike
criteria,” Koblin says, so he cut them out.
But in another project, Ten
Thousand Cents, he left the mistakes in—one contributor wrote “$0.01!!!
Really?”—to see if the data, warts and all, could resolve into a
convincing image. It did.
 |
| ten thousand cents |
For Koblin, order hides
even in chaos. Which is why his new job as technology lead of Google’s
experimental marketing department, Creative Lab, is tinged with irony.
“The first thing you realize here,” he says, “is that you’re never going
to understand the entirety of everything.” (taken from print magazine).