a thaumaturgical compendium

A few times this week, I have chatted with folks about the use of humans by computers. To what degree does the introduction of a new user interface or, for that matter, any technology change the nature of individual thought or social interaction?

There are plenty of specific and general examples of this that are commonly called up. Cell phones and email have changed the way students move around a university campus. So much so, I suspect, that on campuses that have more grass than we do up in frozen Buffalo have likely seen new patterns of erosion. In Japan, they refer to “thumb tribes” (親指族), or youth for whom the thumb has become king. They reportedly have stronger thumbs than the norm, and use them more often for pushing other buttons.

I also recall a science fiction novel in which the characters speak “IBM” and other languages based on the brand of computer they use. I think it would be fun to do a content analysis of Mac users’ blogs and PC users’ blogs to see if one uses certain terms (choose, start, crash) more often. Andrew suggested that Palm Graffitti might also change the way people write in a similar fashion. Certainly, some forms of internet dialect have leaked into everyday conversation, but that isn’t really the same thing.

Amazon has introduced what they call their Mechanical Turk in beta. The original Mechanical Turk was an artificial artificial intelligence, an Ottoman automaton chess player that beat both Benjamin Franklin and Napoleon Bonaparte in games. Of course, it was operated by a hidden human.

Amazon’s project falls on a number of attempts to harness collective micro-efforts by individuals. It pays participants a few cents for each “HIT,” each use of your cerebral computer. Some things are just easier using human brains, and until they figure out a way to scoop it out of your head, they are stuck with you as well.

In the past, this has been used as a clever way of defeating CAPTCHA-like systems. The bot (a spambot, for example) need only “outsource” the recognition part to a human. It turns out there are lots of things that can be so parted out to a human. The Wikipedia is only one example of this micro-effort. If any one person were to sit down an encyclopedia, people would think they were crazy. Sure, drafting full articled takes more of a macro-commitment, but most edits are only worth a few cents-worth of time, and people consider it to be a “give-a-penny-take-a-penny” form of knowledge distribution.

Amazon’s project is designed to provide classification of images, and other things humans seem to be better at. It will allow for a fairly easy programmatic interface to this process. The first thing I thought of, when I saw this, was content analysis in the traditional human-coded form. Of course, you are usually able to train coders, but if you have a really good codebook for a single item, and a little cash to pay for the process, you could do some pretty cool content analysis project this way. Intercoder reliability would inevitably go down, since there is no time afforded for decent training of your “recognizers” but, as they say, you can make it up in volume.

This entry was posted on Friday, November 4th, 2005 at 2:50 pm and is filed under Research, Systems Seminar, Technology, Wikipedia.