Yesterday Google launched Image Labeler, a game in which 2 random internet users are asked to tag (label) a picture. If both players give the same tag, they both get 100 points, and they will get a new picture. If the picture is too difficult or not clear, you can click on pass for another picture. Each games lasts for 90 seconds.
The game is based on the ESP game from Luis von Ahn, who licensed the game to Google. De idea behind the game is that if two people give an identical tag to a picture, the tag must be viable, so that Google can use it in its search results. Possibly they even have an algorithm that only if more than two people use the same tag it will be used. I assume that non-matching tags are also stored, and will be compared to tags that other players give to the same picture.
It seems like a smart move from Google to use the power of the internet population to do some work for them – for free. But will it really work? It of course gives Google the possibility to get all its pictures in its database tagged for free. This should give a Google image search in the future even better results. The question, however, is whether the answers will all be good enough. After playing the game several time I noticed there is a much bigger chance of getting a match when you just type in the main color of the picture. Also it does not make much sense to to use difficult words, because other players also do not do that. Simple words like ‘people’, ‘book’ or ‘blue’ have the biggest chance of earning points. That’a weak point in the system, that could be solved by giving higher points for more difficult or double words (‘space shuttle’ instead of ‘space’ or ‘rocket’)
Another weakness is that a lot depends on your partner. Many game players seem to be not too smart, they give a pass at simple pictures, or only type ‘car’ when they see a picture of a car after 10 seconds. This is a bit irritating, but sometimes you have a partner who seems to think the same as you do, and suddenly you get 10 matches within a minute. An algorithm that puts stronger players together (in which a stronger player is one who scores higher than average), should make the game more interesting, and therefore better for Google.
This algorithm would also ensure that people would play the game for a longer period of time. Now there is no real incentive to play more often or to get a high score. The only thing you can earn are points, and if you are really good (or are really lucky) you might even reach the top-5. But most scores there are almost impossible to reach, and for the overall high scores you need to beat people who seem to play this game full-time.
But I have to admit that the game is addictive. It’s a bit like a casual game, that you play every now and then to do something different for a few minutes. But small prizes, such as a Google Image Labeler t-shirt or extra Gmail storage (I would play the game 10 hours straight for that), could make the game more succesful.
I look forward to finding out how popular Google Image Labeler will become. Will it appear on popular casual gaming sites as well?
(also posted in Dutch at Marketingfacts.nl)