I only just heard about the Google Image Labeler via the IAI mailing list.
Here’s a description:
You’ll be randomly paired with a partner who’s online and using the feature. Over a two-minute period, you and your partner will be shown the same set of images and asked to provide as many labels as possible to describe each image you see. When your label matches your partner’s label, you’ll earn points depending on how specific your label is. You’ll be shown more images until time runs out. After time expires, you can explore the images you’ve seen and the websites where those images were found. And we’ll show you the points you’ve earned throughout the session.
So, Google didn’t just assume people would tag images for the heck of it. They build in a points system. I have no idea if the points even mean anything ouside of this context, but it’s interesting to see a game mechanic of points incentive, in a contest-like format, being used to jump-start the collective intelligence gathering.
Later in the day, I hear from James Boekbinder that this system was invented (if he has it right) by a mathematician named Louis Ahn, and Google bought it. He points to a great presentation Ahn has on Google Video about his approach.
Ahn’s description says that people sometimes play the game 40 hours a week, while I’m hearing from other sources that research showed users putting a lot of effort into it for a short time, then dropping and not coming back (possibly because there’s no persistent or tranferable value to the ‘points’ given in the game?).
2 thoughts on “Google Image Labeler, using game mechanics for swarm intel”
Boekbinder is correct, Google licensed Ahn’s ESP game. Ahn’s more recent work is Peekaboom (http://www.peekaboom.org/) which matches tags for a picture to particular areas within the image using a similar game-type method.
It would be interesting to compare drop-off rates to a version built into a WoW/Second Life-type game where the points could actually have value.
Comments are closed.