Google Image Labeler
My newest obsession has me working for Google... for free. But I'm utterly hooked on Google Image Labeler. It's a fascinating system for crowdsourcing metadata to figure out which search terms people are most likely to apply to a particular image.
When you go there, you are paired with an anonymous, unseen partner for two minutes at a stretch. An image appears on the screen, and you keep typing keywords until you and your partner hit a match. Then another image appears, and another, until the clock runs out. If you are truly stumped, you can pass, but you have to wait for your partner to pass as well. When the time is up, you see an overview of the images, the common word you both found, and the keywords your partner typed. Also, while you are looking at the images, there's a list of words that are "off limits." These are keywords that have already been identified often enough that they have been retired.
Mostly, I just think it's fun. It's like a game; you do get points. The more precise a word is, the more points it gets. Terms like "man" or "woman" and descriptions of colors are low on the scale, usually 50 points. But you can get up to 150 for something like the precise species of an animal. Of course you also get more points if you move quickly and do more pictures. But your partner is someone you don't know and can't see, so you also have to try to guess what they are going to say.
But I'm really fascinated to see what keywords other people choose. Anything from before about 1970, be it a self portrait of Leonardo da Vinci or a fifties housewife, gets labeled "old." Any woman between the ages of about 15 to 40 ends up with "hot," unless she's obviously a mother or a worker. Or unless she is more than 20 pounds overweight, in which case she will receive "fat" and "gross." But any man who styles his hair at all is consistently identified as "gay." Sometimes this happens often enough that "gay" actually ends up in the permanent "off limits" list. There is also a lot of use of very slangy terms I'd never use in a search, such as "guy" and "dude."
Another interesting thing to see is the distinction between a strict description of the image as it is seen and the metaknowledge of what it actually means. On a photo of Eva Peron, I was trying all different permutations of "Eva," "Evita," "Peron," "Eva Peron," "Argentina," etc. We finally got a match when I typed "blonde." And seeing my partner's guesses later, they were all simple descriptors like "woman," "lady," "photo," and "portrait." Though I'm in the same boat when it comes to identifying video games or anime characters.
Try it out and tell me what you think. Who knows? We may even end up being partners without knowing it.
When you go there, you are paired with an anonymous, unseen partner for two minutes at a stretch. An image appears on the screen, and you keep typing keywords until you and your partner hit a match. Then another image appears, and another, until the clock runs out. If you are truly stumped, you can pass, but you have to wait for your partner to pass as well. When the time is up, you see an overview of the images, the common word you both found, and the keywords your partner typed. Also, while you are looking at the images, there's a list of words that are "off limits." These are keywords that have already been identified often enough that they have been retired.
Mostly, I just think it's fun. It's like a game; you do get points. The more precise a word is, the more points it gets. Terms like "man" or "woman" and descriptions of colors are low on the scale, usually 50 points. But you can get up to 150 for something like the precise species of an animal. Of course you also get more points if you move quickly and do more pictures. But your partner is someone you don't know and can't see, so you also have to try to guess what they are going to say.
But I'm really fascinated to see what keywords other people choose. Anything from before about 1970, be it a self portrait of Leonardo da Vinci or a fifties housewife, gets labeled "old." Any woman between the ages of about 15 to 40 ends up with "hot," unless she's obviously a mother or a worker. Or unless she is more than 20 pounds overweight, in which case she will receive "fat" and "gross." But any man who styles his hair at all is consistently identified as "gay." Sometimes this happens often enough that "gay" actually ends up in the permanent "off limits" list. There is also a lot of use of very slangy terms I'd never use in a search, such as "guy" and "dude."
Another interesting thing to see is the distinction between a strict description of the image as it is seen and the metaknowledge of what it actually means. On a photo of Eva Peron, I was trying all different permutations of "Eva," "Evita," "Peron," "Eva Peron," "Argentina," etc. We finally got a match when I typed "blonde." And seeing my partner's guesses later, they were all simple descriptors like "woman," "lady," "photo," and "portrait." Though I'm in the same boat when it comes to identifying video games or anime characters.
Try it out and tell me what you think. Who knows? We may even end up being partners without knowing it.