[MUD-Dev] Game Economies

[Matthew Mihaly]

On Thu, 10 Jun 1999, Jo Dillon wrote:

>   I'll see if I can dig it out. It's about time I handed it to the Devmud
> people anyway ;) I believe there have been economists working on agent-
> based simulations - I don't suppose you've heard anything about that?
> I'd bet their code's a lot more sophisticated than mine!

Yes, in fact, JC and I posted about this sort of thing either yesterday or
the day before (ahh, how they blur together). The Santa Fe Institute in
New Mexico was formed as an inter-disciplinary organization in order to
study complexity. It managed to attract some of the absolute top people in
their fields, like Murray Gell-Mann (Nobel winner in particle physics I
believe), Phil Anderson (Nobel winner for his work in solid-state physics,
itself sort of a field studying emergent properties of particle physics),
Ken Arrow (Nobel winner in economics), and a bunch of brilliant young guys
who had previously been working largely independently on ideas like
genetic algorithms, cellular automata, neural networks, classifier
algorithms, co-evolution, "self-organized criticality", etc. 

Their first major product was commissioned by the head of Citicorp, who
was disgusted that Citicorp's in-house economicsts (presumably using
mainly Keynsian-style economics) were so incredibly useless. Citicorp had
lost TONS of money (like 1 billion in profits alone), from defaulted Third
World debt, and was sitting on another 13 billion in loans that they may
never see paid back. The economists had been utter failures at predicting
the crisis and their advice apparently just made things worse. (and given
the silly assumptions that classical economics makes, it's no surprise). 

So, they got to work, and basically the Santa Fe approach, instead of
emphasizing static equilibrium points, decreasing returns, and perfect
rationality (neo-classical view), they emphasized rationality that is
bounded, increasing returns, and evolution and learning. Basically, they
were going to look at an economic system as being a living thing that
adapts and changes and has no static equilibrium point. (wasn't really
totally new approach. Joseph Schumpeter had pushed for approaches like
this in the 30s). It was quite a difference in approach, as most economics
operated purely deductively, whereas this was going to use agents that
operated inductively (ie realistically).

Anyway, it's all very interesting, even to someone like me who isn't an
economist and doesn't know enough math to understand all the equations and
such. You might try reading Brian Arthur's "Positive Feedbacks in the
Economy" from Scientific American, February 1990, or see if you can find
"The Economy as an Evolving System" by Phil anderson, Ken Arrow, David
Pines, and some others. It's volume 5 in the Santa Fe Institute Studies in
the Sciences of Complexity and contains a summary of the institute's first
large economics meeting sometime in the late 80s. Published by
Addison-Wesley.

I think it would be beyond excellent if a mud was ever coded with truly
sophisticated adaptive agents, though I'm not really sure how much I'd
want this to happen in a for-profit game. Who knows what kind of
structures they would form. Would they invent central banking and options
trading, or would they come up with entirely new ways of expressing
economic motives and desires?

--matt




_______________________________________________
MUD-Dev maillist  -  MUD-Dev at kanga.nu
http://www.kanga.nu/lists/listinfo/mud-dev