Google Image Labeler: Friend or Foe?
Google has made everyone a little less productive with their launch of the mind-numbing Google Image Labeler. During the “game”, you and another bored Web user are paired up to view and
tag label images. Once you come up with matching tag labels you are allowed to move on to the next image, thus helping Google index its entire image collection. Fun, right?
But what’s the real reason behind the new Google game? Is it simply a fun way to help Google catalog its collection of images? Or is the game “pure evil“, as Kevin Burton seems to think?
I’m not sure about Kevin’s theory, but the folks at BuyGoogle seem to be on to something interesting. The BG gang notes that the game was based off Luis Von Ahn’s Google-licensed ESP Game, which Ahn talked about during July’s Google TechTalks (video here). Based on that video, BG things the video is less about labeling images and more about machine learning while they prepare to make good on the recent Neven Vision acquisition. Hmm…
“Once Google has a set of images that are labeled with a high signal-to-noise ratio, this would be an ideal corpus of material for their newly acquired “machine vision” technology to train on. And once Google’s computers are trained to interpret images and video, then Google’s algorithms can do for images and video what they’ve done for text — organize it and make it universally accessible and useful.”
If that’s that real reason behind Google’s game, then I’ll be happy to sacrifice an hour of
my work day personal at home time to help Google catalog its images. And though I can see Kevin’s point about Google using us as their little worker bees, if the end product is something that everyone will benefit from and users don’t mind pitching in, hand me my Google uniform.
As the BG blog noted, Ahn actually put the worker bee concern into perspective during his speech in July.
“One slide says that if people diverted the time they play solitaire to construction projects, they could build 1285 Empire State Buildings, or 450 Panama Canals every year. Why not have them label the Web?’
Seriously. People waste a lot of time performing mindless activities on the Web, why not encourage them to at least do something useful with their time?
I don’t think the recent Google Image Labeler is anything for users to get up in arms about. I think you’d be crazy to view this as Google on its path to world domination. Google is just doing what they’ve always set out to do – organize the world’s information.
With that in mind, in order to “best organize”, Google should find a way to allow users to view the entire image (not just the thumbnail) to better distinguish the image and differentiate it from other like images. What good is it to have a thousand images labeled “dog”, when with a little more detail “dog” could become “cocker spaniel” or “border collie”. Or maybe it’s not even a dog at all, maybe it’s really a luck dragon. The clearer the image, the more accurate the label will be.
[Side note: Wouldn’t it be fun if Google did this for websites too? They could show users a series of screenshots and ask them to identify the keywords that go along with the page. It could be like Name That Site.]
Regardless, good for Google for having the foresight to tap into the community while providing safeguards (users tags have to match) to make sure ill-intentioned users don’t label a monkey a giraffe or a Microsoft icon a Google icon (the horror!).
I’m all for channeling my boredom-induced energy into helping Google label the Web, especially if it gets us one step closer to making images and video as interpretable as text. I mean, you can only play so much Solitaire.