Darwinbots Forum
General => Biology => Topic started by: Numsgil on May 29, 2006, 04:17:15 PM
-
This blog entry (http://www.greythumb.org/blog/index.php?/archives/109-Plateaus-in-complexity-growth-in-artificial-life-systems-wheres-the-problem.html#extended) talks about why population complexity tends to "plateau" in ALife systems.
Really doesn't tell me anything we haven't discussed in the forum before, but it does support the idea that Darwinbots will achieve more complexity if:
1. The uniform blue field isn't so uniform (env. grid)
2. One bot's waste is another's food (metabolism)
While I'm at it, the blog linked to here (http://www.sparknotes.com/biology/evolution/naturalselection/section1.html).
-
Very interesting, especially the info about signaling. Some stuff there I've never thought about.
-
We seem to have at least a couple of the types of selection, stabilizing and directional. The disruptive selection is where we're most lacking I think. The bots in even the most successful ecosims seem to stagnate into one particular "best" form, depending on enviroment and initial dna.
I'm going to try and see if walls can be modified to function as mobile blocks, creating a basic movable object for the bots to use/adapt to. I still need to make them(walls) stop being destroyed first though.
Sexrepro in one of my versions is working somewhat correctly now, seems to provide some novel effects.
50 300 2 rnd add store
This bit of reproductive code was conserved by one species during its evolution. Seems evolution likes the occasional re-mixing effects it brings.
Anyways it seems like we just need to provide more texture in a whole bunch of different areas. Just takes awhile to make it work.
-
GreyThumb's blog is one of the best I have encounted yet! Very interesting views, links, articles etc.