Lambda the Ultimate

Watching Super Walker is pretty entrancing. It's interesting how many things go through your mind.

For example, if you turn down the gravity, long-legged walkers quickly start to dominate, as they tend to sail spastically through the air. (In fact, they tend to develop jumping and cartwheel behavior, although this comes and goes.) But I'm pretty sure they have an unfair advantage, because their long legs create more torque than short-legged organisms. So one wonders what would happen if you fixed the power available to an organism so that becoming airborne has a cost associated with it. Similarly, what if impact on landing could cause damage?

Also, for example, finer-grained sorts of genomes would be interesting. I want my daddy long-legs to learn how to jump and land, for example, but the demo just seems to loop the limb actions at regular intervals, and air travel is so chaotic and mutation so severe that it is very unlikely my organisms will ever learn that in my lifetime, and if they did, it would be only for a generation, and then mutation would upset the delicate balance required.

So what would be interesting from a programming point of view is to develop a much simpler simulation, with less realistic physics, and a more declarative language, where there is enough information to prune undesirable evolutionary branches, so that we don't spend a lot of time wasting time, and effectively accelerate the sim. I can see fixpoint theory would be handy here.

Anyway, I encourage you to play with this thing. I would if I had the time.

http://www.spiderland.org/breve/
(there's a typo in the topic text)