This week in Web 2.0 class, we played with the
Google Image Labeler and then watched a Google Tech Talk from Luis von Ahn (July 26, 2006) about human computation.
1. The game was fascinating. I am a fan of online games, but few of them have a purpose where the output of the players is actually being used on the Internet. We had to work towards scoring over 900 points. You are matched with a strange and asked to give photos labels. When you can match a label with your partner, you score points. There is a timer present and taboo words that you cannot use to keep the game moving fast.
It is much harder than it looks. You can't communicate with your fellow labeler. Time ticks on. Every time your partner suggests a label, you can see the number of suggested labels increase. It adds another layer of pressure!
2. Lots of interesting information from this video. He begins by discussing the concepts of CAPTCHs and how this tool was used to determined if the user was a human being or a hacker. Great example of an online poll used with some university students. Within days, millions of surveys were being submitted by a computer program which made the results null and void.
His talk though was more about human computation. First, there are things that computers cannot do that only humans can do like labeling images. Second, images on the web are not accessible to the visually impaired because they lack captions. How do you get humans to label images, for free and fast? You make it a game, a game with a purpose.
ESP game - which I believe is now the Google Image Labeler, is addictive. During the presentation, von Ahn displayed some of the description of the ESP game from users. One in particular stood out to me, "anonymous intimacy". YES! I know exactly what this means!
For example, while playing this game, I was matched with a stranger that I could not see. Sometimes my partner was slow on giving labels or wanted to pass when I really liked the image. I found myself shouting at my monitor, "No", "What are you doing?, or "Come on, suggest something". Then, I was paired with someone who was in sync with me. We were matching labels within 10 seconds of seeing the image for the first time. "Yes!" "Way to go partner". "Look at our points." Even though I could not see these people, I instantly had some sort of connection with them.
Other games he introduced included Peekaboom and Verbosity. Both of these games had a purpose where players were matched with strangers to reach a common goal. ESP game is an example of a symmetric game where players are given the exact same information at the same time. Peekaboom and verbosity are asymmetric games where the players receive different information and try to get their partners to guess what they are trying to say.
How could games with a purpose be used for instruction on a massive scale? Here are some intial ideas that I had:
1. ENGLISH - Anything that gets students writing and critical thinking is a useful instructional tool. Play the game as a class at first and encourage students to play until they hit a specific score.
2. SOCIOLOGY - The words that we use to tag these images say something about us as a people. Analyze the taboo words. Look at the words that are not making the cut. What is going on?
3. eLearning - Give instructors access to learning objects and allow them to use the same concept as the ESP game to assign metadata. Repositories never list all of the data that you are really seeking.
Love this game. I can see why it's addictive. Love the concept of games with a purpose to solve a problem like labeling images in Google.