outsourcing your work as a captcha

I guess everyone’s seen the robot test captcha thing on Google these days. If you try to use their websearch engine too fast, then you’ll soon be proving that you’re not a script that’s running on some spammer’s computer.

I'mNotARobot

Often, though, you’re then next asked to select which squares have store fronts, or doors, or signs or food.

CaptchaSigns

And of course, since we want that content, we dutifully “prove” that we’re not a robot. But—and I realize this might sound a little cynical of me—what if we’re actually being forced into conscripted labor, as if we were Google’s robots?

What if we’re actually being forced into conscripted labor, as if we were Google’s robots?

Try to follow along…

Amazon Mechanical Turk

Amazon has a variety of services within the AWS space. The one I’m thinking about at this moment is their Amazon Mechanical Turk. If you have a computer and Internet and want to make some money doing (usually) mundane tasks, then Amazon will pay you to do so.

For instance, Amazon might pay a hundred people to look at one image after another and to indicate/highlight where in the image they see a sign or a store front or whatever it is that Amazon needs highlighted. Humans are great at this. Artificial intelligence applications are getting there, only it takes a supercomputer these days in order to do these tasks.

What if Google doesn’t want to use their supercomputers nor wants to pay anyone to do object recognition either?

Google Maps Streetview

Google’s mapping featureset with Streetview represents a way for them to make a lot of money. And their collection of project managers would love to know where storefronts are within all that captured data. (Imagine that they’ve paid drivers to drive around a car with 360° cameras.) Because behind every storefront is a business who could pay Google money for placement within Google Local.

Now, Google has datacenters with plenty of available processing power to do this. But what if… they’re using us instead.

Think about it, we’re asked to identify objects within photos (which look like they’re taken from the Streetview data) and we’re being asked to identify things (businesses) which could make Google money or things (signs) which could be used in mapping directions.

Call me cynical but Google is looking a little guilty on this one. Why aren’t we identifying the squares with puppies in them? Because puppies don’t buy listing upgrades, that’s why.