Thursday, November 30, 2006

The Sheep Market: Two Cents Worth

Aaron Koblin has published his thesis paper on the rather intriguing Sheep Market project.

To be honest, it is a little over my head, but still a good read. He brings up the very valid point about how, things like his Sheep Market, are clearly not worth the wage provided. Yet Turkers complete the tasks. The only thing truely comparable to his project is the YRUHRN book. Another half-experiment/half project that used the Mechanical Turk.

So just what does motivate the Turkers? What motivates someone playing a game? Seriously, the video game industry is huge and the sole purpose of most of the games is simply to get to the next level. Now, some do have intriguing storylines and other stuff of that sort. But if gamers simply wanted that, they could go watch a movie.

The high point of my Turking, and all of the old Turkers was the Image Adjustments. Semi-mindless tasks that we did over and over for 3 cents (in November of 2005). We did however make decent wages (6-12 bucks depending on your setup) But what was the point? Was there any pride in doing it? From my standpoint, no, it was just about the money.

Now, you could put HITs on mTurk into a two groups.

Group A: Mindless tasks that one couldn't take pride in, essentially tasks with right and wrong answers. Like "Is there a person in this picture? Yes | No"

Group B: Mindful tasks, like drawing a sheep, transcribing a podcast, giving advice, Ranking Your Top 3 of something. These are basically HITs with an very large or infinite number of right answers. In fact in reviewing of HIT results you see the personality of Turkers with these sorts of tasks. To the point that you sometimes recognize who did some work without checking the Worker ID.

Now Group B tasks degrade into Group A tasks when they become repetitive. For instance, if all of the sudden, I am drawing 100 sheep a day, 7 days a week, then sheep drawing has become a mindless task for me.

And so, I think this is a solution that Automated Webshoring can give us. In life there will always be repetitive tasks. Most lower paying jobs (and many higher paying jobs) become just a sea of mindless work for employees. Now you learn to live with it and technology has ridden us of some of it (copy machines!). But what if we "Websource" unskilled labor, by breaking down the tasks we give workers the ability to choose what they do, on a momentary basis. So one minute I'm a Fast Food worker taking an order, the next I'm helping a child with there homework, then I'm typing a license plate number from a picture taken by a red light camera, and then I'm typing out a small chunk of a podcast.

Now granted, some of this is down the road....some might not even be on the road. But to me, it seems like a bright future. It seems like it's something that could change the world, perhaps for the better, perhaps for worst. But one thing is for sure, our systems will keep getting smarter, faster, and more efficient by whatever path companies find.

via AWS Blog (Yes I know it took me a week to finish writing the post)


Post a Comment

<< Home