Chris posted some cool questions on my earlier post about decentralized behavior in social insects:
so what kinds of things do computer scientists do w/this idea? is sort of what evolutionary computing is about? i picked up the evolution of cooperation recently (tit for tat wins prisoner’s dilemma) you must be familiar w/this, yes?”
There’s really three different (but related) ideas in this question:
Decentralized and emergent behavior
Game theory and cooperation
I’ll take them one at a time, with more on the first of the three, at least for now.
As many of my students can attest, I’ve always loved emergent behavior where you have a whole bunch of “stupid” agents working with simple rules, but the end result is amazingly complex and wondrous. There are termites, for example, that by following simple rules (and with no “leader” in charge) manage to build large mounds with extraordinarily precise climate control in certain chambers where they farm special fungi. If you held a gun to my head and told me to use dirt, sticks, and other bits lying around to build a chamber that would be nearly constant temperature day and night, winter and summer, I’d have to tell you to pull the trigger. Yet these insects, with no blueprint or foreman, pull it off over and over again across thousands of years. (Doesn’t evolution just rock!)
No surprise, then, that I was smitten with this recent TED video (of a talk from four years ago) where Deborah Gordon talks about her research into how ants figure out how to allocate tasks (i.e., who does what when).
If you want to play with this sort of thing, I recommend both NetLogo and Breve, as well as this cool Breve-based screensaver, which evolves walking creatures while you’re momentarily distracted from your random web surfing. Both NetLogo and Breve can happily suck up way too much of your life (there are literally days to be spent playing with all the canned models in NetLogo), but it’s great fun, and working with tools like this can really help build your intuition for how complex systems of agents can work.
Below the fold I list some of my favorite NetLogo simulations that I often use in demos.