Amazon.com has out-googled Google with its creepily brilliant Mechanical Turk service, a means of embedding human beings in software code. If you’re writing a program that requires a task that people can do better than computers (identifying buildings in a photograph, say), you can write a few lines of code to tap into the required human intelligence through Mechanical Turk. The request automatically gets posted on the Turk site, and people carry out the Human Intelligence Task, or HIT, for a fee set by the programmer, with Amazon taking a commission.
As Amazon explains, this turns the usual computer-human interface on its, uh, head:
When we think of interfaces between human beings and computers, we usually assume that the human being is the one requesting that a task be completed, and the computer is completing the task and providing the results. What if this process were reversed and a computer program could ask a human being to perform a task and return the results? What if it could coordinate many human beings to perform a task?
I have no clue how useful Mechanical Turk will prove immediately, but Philipp Lenssen (who foresaw the service in a remarkable post earlier this year) thinks the “potential is immense.” Certainly, the implications are mind-bending. In an essay I discussed last week, George Dyson described how the Internet provides a platform, or operating system, that enables computers to harness and learn from the work of people: “Operating systems make it easier for human beings to operate computers. They also make it easier for computers to operate human beings.” Google uses this capacity implicitly by basing its search engine on human actions and decisions – as we make our daily strolls through the Web, Google gets smarter. Amazon’s Mechanical Turk uses the capacity explicitly, turning people into a “human layer” in software.
But let’s not get too comfortable in our new role. No one, after all, is indispensable.