It seems my change in approach to stumble upon has paid off. Lately I’ve been getting much high quality pages in my stumbles, the culmination of which was this.
If you follow the link to this site, you’ll get an explanation for this strange video, but here’s a brief excerpt for those unwilling to click.
Tweenbots are human-dependent robots that navigate the city with the help of pedestrians they encounter. Rolling at a constant speed, in a straight line, Tweenbots have a destination displayed on a flag, and rely on people they meet to read this flag and to aim them in the right direction to reach their goal.
Theses tweenbots are the creation of Kacie Kinzer, a student at the two year ITP program at NYU. She built the robots for a graduate project meant to explore the how people in the busy New York City might respond to a robotic encounter. Would they be willing to stop and help a creature made to look vaguely human? Or would they just walk on by? Kinzer assumed that her robots would perish in the dangerous streets of New York city, unhelped and unloved.
Luckily, this was not the end result. Much to Kinzer’s surprise, people were more than willing to stop for a minute and point the tweenbot in the right direction. All of her tweenbots made it to their target destinations, as evidenced by her video.
It’s easy to dismiss this as just a funny little social experiment, but it’s worth taking note of exactly what Kinzer has done. She’s designed a robot that is capable of something that to date no other robot (that I know of) has been able to do; navigate a complex unfamiliar environment. The only requirement is that there are people willing to help it.
Of course from a practical standpoint, this isn’t all that exceptional. I don’t expect that will see an application for tweenbot style navigation anytime soon. But as a lesson on how to created problem solving systems, the tweenbot offers some very useful points. Ultimately, the tweenbot suceeds because accepts information from it’s enviroment promiscuously, without regard for accuracy or source. There is no validation step. The only thing that the tweenbot can hope for is that it’s cute appearance will make people take pity on it.
The assumption we usually make in this sort of information is that this lack of validation and this promiscuity makes it vulnerable. This is exactly what Kinzer thought originally. However, in actuallity, systems like this can be very robust. The good interactions cancel out the bad. This is why the auidence on Who Wants to be a Millionaire is usually right, and why prediction markets give accurate forecasts
In some ways, the tweenbots are a good analogy for crowdsourcing on a large scale. To reap the benefits of the crowd you have to open yourself up to large scale interaction, making you potentially vulnerable. And you have to come up with some sort of message, theme, or framework that encourages positive interactions.
And I think most importantly, the tweenbots show that the best tools recognize that they are part of a larger system that can play an active part in coming up with a solution. That the most powerful tools rely on emergent behavior that is not programmed or specified anywhere in the system, but bubbles up through the interaction of it’s members.
In short, Kinzer’s tweenbots are one of those very fortitious, concrete examples of a set of very abstract principles. They can be examined from a number of angles to get at otherwise hard to example concepts, crowdsourcing and emergent behavior being just a few of them. I’m happy that stumble upon lead me to them, and provoked some interesting thoughts I wouldn’t have otherwise had.