April 2, 2002

12:09 PM

"They are, it is true, almost laughably simple by comparison with real people and real societies, but that is exactly the point. If even the crudest toy societies take on a life and a logic of their own, then it must be a safe bet that real societies, too, have their own biographies." Things have certainly gotten more interesting in the years since John Conway invented the game of life.
posted by tdismukes (20 comments total)

Part of the practical use for this line of research will probably depend on whether human societies are governed by the laws of chaos theory (arbitrarily small differences in starting conditions resulting in arbitrarily large differences in ending conditions.) However, the simulations listed seem to show that certain dynamics may occur consistently even when starting conditions are varied. If so, this could be the best new lead I've seen for someday making sociology into an actual science. Thoughts?
posted by tdismukes at 12:15 PM on April 2, 2002

You can download the software to play with here
posted by i_am_joe's_spleen at 12:49 PM on April 2, 2002

I thought this was going to be about the Sims!
posted by acridrabbit at 12:55 PM on April 2, 2002

i_am_joe's_spleen - Thanks! I was wondering what it would take to code some of these simulations and I was hoping there would be some web resources so I wouldn't have to start from scratch.
posted by tdismukes at 1:01 PM on April 2, 2002

it's a new kind of science :) basically, i think the trick is just to find CA models that fit RW data patterns. like majority voting for figure 2 in the article: "agents seek only one neighbor of their own color."

what i find really interesting is power law behavior in certain kinds of entropy which could lead to new measures of complexity. there's going to be a workshop at the santa fe institute next week about it. check out the abstracts!

IBM has been doing some work on agent based artificial economies and gulfstream recently linked to this firstmonday article by a guy at RAND "on the connexion between game theory and drafting in NASCAR races."
posted by kliuless at 1:05 PM on April 2, 2002

btw, cosma shalizi proposes his own measures of complexity :) also, his paper on self-organization and CAs "is the primordial soup done yet?" is really great!
posted by kliuless at 1:26 PM on April 2, 2002

This is an absolutely fascinating article. Ranks up there with the best stuff I've ever ran into on the 'Net, education-wise, from the homeless children's stories of Bloody Mary to the Onion's interview with KRS-One.

Thanks so much, tdismukes.
posted by Marquis at 1:36 PM on April 2, 2002

I think alan turing pretty well proved the futility in trying to formulate methods of determining what any given system will do in the long run (halt/not halt).

Complexity is fun. In a way I'm not surprised that random noise isn't ever generated with these things. The larger and more complex a system gets, the greater chance that it is built up from self similar parts. It is very easy to provide instructions for a life form which is very simple and very large, but very difficult to do the same with instructions that create something without any patterns in it, ever. In fact, one could argue that the data required for such a system would equal the complexity of that system...although you can get pretty nonsensical with simple instructions as well.
posted by Settle at 1:39 PM on April 2, 2002

This is an absolutely fascinating article. Ranks up there with the best stuff I've ever ran into on the 'Net, education-wise, from the homeless children's stories of Bloody Mary to the Onion's interview with KRS-One.

Thanks so much, tdismukes.
posted by Marquis at 2:11 PM on April 2, 2002

oh my. how embarassing.
posted by Marquis at 2:12 PM on April 2, 2002

Fascinating stuff. I wonder how the spread of speedy, reliable communications technology affects the conclusions, with respect to addressing some of this emergent behavior. If we're smart about it, can't we find ways to educate and inform the "agents" in our societies (i.e., us), so that they better understand their circumstances, and the impacts of their actions?

That is, it should be possible to find a model that is sustainable -- and then find ways to foster that type of society.
posted by mattpfeff at 2:57 PM on April 2, 2002

In a way I'm not surprised that random noise isn't ever generated with these things.

interestingly random noise may be integral to life, the universe and everything! and a newscientist article in the web archive :)
posted by kliuless at 6:09 PM on April 2, 2002

(If you decide to read the New Scientist article that kliuless links to above, there is something else you should read. John Baez, an old netizen from sci.physics, is quoted in the NS article but has this to say about it.)

I am not quite clear on how a lot of this society modeling differs from efforts in Cellular Automata (CA) or how current CA software cant be used to do this modelling. The major difference seems to be that the CA guys arent yet claiming to be solving any real-world problems.

Also, Zipf's Law (aka Benford's law) is presented as some sort of great mystery. It is really strange, true, but it can be explained as the natural appearance of power laws in human-created statistics.

My own feeling on this is that real-world problems are too complex, that real people and their behaviours cannot be accurately modeled in any meaningful way. I can see how this might produce some quick insights and Aha! type results but nothing that a simple statistical model could not have produced. Twenty years ago these were simply called 'simulations'. I dont see how today they are now self-organizing atomistic models that show emergent behavior. Maybe its just me, but I started losing interest when I saw the self-organized criticality sandpile analogy being trotted out for the umpteenth time.
posted by vacapinta at 8:56 PM on April 2, 2002

thanks for the pointers vacapinta. i just learned about john baez from a paper cowlix linked to about loop quantum gravity in relation to greg egan's new book.

i actually wrote cosma shalizi about that newscientist article after it came out and like baez he didn't think too much of it. me<--disappointed. having really no math or physics background and no means or experience to evaluate their work, it sounded really cool to me. and i still think it's neat to think about. but yeah, i guess it is pretty buzzword compliant :) process physics did make it to the cover of the physicist too though, does that count for anything?

also btw, it seems kauffman and smolin (based on penrose's spin networks) have their own version, twister theory!

How do these discrete bits of space assemble themselves into a smooth structure that looks like the space we see around us? This turns out to be very much like asking why atoms often assemble themselves into solids, like plastics or metals, that look smooth when examined at scales larger than the atoms. It seems to be the case that without some special organization, the discrete bits of space—the networks—do not assemble themselves into big smooth structures that could describe the featureless space we observe. Instead, they typically form chaotic structures that do not resemble any previous notion of space.

Thus, we are faced with the very real possibility that the fact that the world has any spatial extension at all is a contingent historical fact, that also requires explanation by some principle of self-organization. We are working on this now, and there seems to be good progress. I also expect that the outcome of this work will be a unification of the different approaches that have led to an expectation that space is discrete, including string theory and black hole thermodynamics. What for me is most provocative is the possibility that, for this to work, we will have to extend the Darwinian idea that the structure of a system must be formed from within by natural processes of self-organization—to the properties of space and time themselves.

kinda reminds me of this thing i read on chris crawford's erasmatazz basically positing some law of information conservation 'it' from 'bit' type deal with like qubits in the transition or something.
posted by kliuless at 5:02 AM on April 3, 2002

Great, great link... Bravo tdismukes.
posted by talos at 5:21 AM on April 3, 2002

Thanks, kliuless!

I think because of Brian Greene and his elegant universe, many people are under the impression that string theory is the only revolution happening in physics right now.

What is interesting is that ideas like Penrose's spin networks, the EPR paradox and the DeWitt equation, these old anomalies in Physics, are being regarded as serious clues in the hunt for the ultimate nature of the universe. I admire what Smolin has been doing.

I also recommend the book End of Time by Julian Barbour, a collaborator of Smolin and others. He's a physicist who spent 30-odd years working as a translator and developing his idea that time does not exist! The way I would put it is that time is a psychological shorthand and has no true meaning in the way that 'up' and 'down' mean nothing in outer space.

Regarding Artificial Life, I'm surprised nobody has mentioned Avida. I once read some sci-fi novel where Avida from the twentieth century had given rise to a whole new set of organisms who then tried to obliterate humanity etc. Avida has been studied intently.
posted by vacapinta at 10:59 AM on April 3, 2002

Thanks to everyone that posted links to software downloads. I just downloaded the Ascape software that joes_spleen linked to. My girlfriend just finished her master's degree in social psychology and is heading to a doctoral program in the fall. I plan to get some real-world social dynamics problems from her to play with and try to model. (I don't necesarily expect to discover anything amazing, but at the very least it'll be some good programming practice.) As far as whether anything "useful" actually come from this field of research, I guess that would depend on whether anyone can find characteristic patterns relevant to real world problems that emerge consistently even when the detail of the simulation is not perfect. (For example, the original article suggests that a smaller number of police focused in the right areas have much more effect than a lot of police spread out everywhere. Of course, real world police probably know that already, but we might find other patterns which are less obvious.)

If anyone else is has experimented or is thinking about playing with the software, drop me an e-mail to discuss.
posted by tdismukes at 12:29 PM on April 3, 2002

I always suspected those games of go were trying to tell me something...
posted by roboto at 3:04 PM on April 3, 2002

tdismukes,

I've been trying to find more information on algorithms, models etc. that can show the spread of disease. This note by Ed Pegg intrigued me a long time ago- you may want to write him. Researchers are using advanced mathematical models but seem reluctant to make them publicly available.
posted by vacapinta at 4:02 PM on April 3, 2002

vacapinta - I don't have a sufficient math background for the sort of mathematical modeling the researchers you linked to are doing. The computer simulations you can do with the Ascape software don't seem to require that level of math however - you just need to be able to program in Java. I just started to look at the code today,so I'm not really up on everything it can do, but it seems like it could model disease spread just as easily as it does societal behavior.
posted by tdismukes at 8:57 PM on April 3, 2002