Terminally Incoherent

Utterly random, incoherent and disjointed rants and ramblings...

Thursday, July 14, 2005

Virtual Worlds

This is the kind of news I like to read - stories about ambitious, intriguing and interesting projects. Nice change from the barrage of stupidity that I have been posting about lately :P

Bunch of researches from NEW-TIES (New and Emergent World models Through Individual, Evolutionary, and Social Learning) are planning to build a virtual world, inhabited by thousands of individual characters. They will be able to interact with each other and with their surroundings in order to find food, reproduce and build things. The researches also built in a random word generator, hoping that these virtual sims can develop their own language by learning (ie. pointing at things, and saying random stuff). This is reported in this New Scientist article.

Oh, and they are using Counterstrike engine! How wild is that :P

I find this very, very interesting. I did some work on emergent systems when we were learning parallel processing. We adapted cellular automata model (for the uneducated: think Game of Life) model to study spiral wave phenomena in excitable media. Here is the ppt presentation we gave at Sigma Xi. While this is not my main research area, I always found this to be fascinating. We literally sat there for hours watching spiral waves form triplets. Game of Life itself is mind blowing!

This project might be one of the most significant experiments in the study of emergent systems. The NS article offers a critique from Edward Castronova, an alleged expert in online gaming worlds from Indiana University. He claims:

Inferences from an entirely artificial system are always going to be weakened by the artificiality.

I think he is missing the point here though (and so does the editor, I'm afraid). This study is not so much about making inferences about human society from an artificially created model. It is more a study of emergent behaviors using a very complex model. This is not really about observing how a real society forms. It is about putting thousands of complex individual bots into the wild, and seeing if by means of random interaction they can conjure something ever resembling a society.

While this can give us some insight on how civilized societies might, might not form, this does not seem to be main focus of the research. This is more of an AI project answering questions such as - can little virtual bots develop their own language via random interactions? If they do, will they be able to develop abstract concepts such as worship, ritual or religion? Will this artificial society organize itself hierarchically? Will some bots emerge as leaders? Ultimately, this research will give us some answers on the future of AI!

This is fascinating stuff! Of course a sociologists from Indiana who specializes in analyzing the MMPOG's "societies" would not understand. Now I think that social implications of the grind, or impact of monies whorin noobs on virtual economies might be fascinating... :P No really - observing how people act in virtual worlds is a good research topic for sociology and psychology people. This one, however is a computer science project!

0 Comments:

Post a Comment

Links to this post:

Create a Link

<< Home