Author Topic: Diversity in plain environment  (Read 2529 times)

Offline shvarz

  • Moderator
  • Bot God
  • *****
  • Posts: 1341
    • View Profile
Diversity in plain environment
« on: March 15, 2007, 07:13:21 PM »
Hello everyone!

It's been a while since I came here last time.  Hope things are going well...  although I do see that it appears to be one of those "slumps" that we get.

Anyway, I have a question that I'd love to talk through with someone.  We've all seen that the flat and boring environment of DBs leads to selection of bots that are very alike.  That's not surprising, because that's exactly what a darwinian selection should do.  Even if you start with two bots that are exactly the same, one of them is going to win in the end because it would be the lucky one that got a mutation that gave it a temporary edge over the competitor.  

Yet I have an idea on how a fairly homogeneous environment can maintain survival of two (or more) species for fairly long periods of time. The idea is that instead of having one single continuous well-mixed environment we create an environment that consists of a large number of small temporary niches of food that allow only a limited number of progeny to be produced.  Kind of like a veggie spawning in a random spot in the sim, dividing for a while and then dying.  

This would achieve two things:

First, the competing species would rarely directly interact.  Each organism would randomly encounter a niche and use the available resources completely.  Whoever is first one, gets the whole niche.

Second, the stochastic effects of encountering a niche would cancel out any beneficial mutations that may be acquired by one of the bots.  This is because a more fine-tuned bot, a bot that is better adapted, has a lot more chances to create a progeny that is less fit than parent, than a less-fit bot.

Does that make sense?  I feel like I did not explain it very well, so please ask questions to clarify.

I'm actually thinking about testing this hypothesis in a number of ways. I can do some real experiments with viruses to test that.  I have a math-wiz guy working with me who is maybe interested in doing some modeling of this situation.  And I also have a buddy who is interested in writing a cellular-automata program to simulate that (kind of like DBs approach to modeling life).

But I just want to run this idea by as many people as possible to see if there is some flaw in the logic that I'm missing.  So I'd really appreciate your comments
"Never underestimate the power of stupid things in big numbers" - Serious Sam

Offline Endy

  • Bot Overlord
  • ****
  • Posts: 852
    • View Profile
Diversity in plain environment
« Reply #1 on: March 16, 2007, 12:45:09 AM »
It might be possible to do that currently. Have a veggie only last a limited time by storing large quantities of shell/slime/etc. The initial veggie could set an epigenetic location to some value, then all the bots could "count-down" to zero and then run their auto destruct.

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7718
    • View Profile
Diversity in plain environment
« Reply #2 on: March 16, 2007, 12:32:17 PM »
Well you also have to ask the question of how useful two species coevolving are if they rarely interact.  You could for instance, use shapes to totally segregate two populations of bots.  You would effectively just be running two simulations at the same time.  It could still be interesting to see if they followed concurrent evolutionary lines, but it's still limitingly interesting.

The real grail, in my mind, is to achieve two or more species interacting to create a simple ecosystem, especially if they come from the same stock.  Have a base species that splits into an herbivore and a carnivore, for instance.  Or one turning into a shot feeder while another turns into a tie feeder.  Just so long as they can cohabitate the same physical space-time.

Offline MacadamiaNuts

  • Bot Destroyer
  • ***
  • Posts: 273
    • View Profile
    • http://www.franontanaya.com/blog/
Diversity in plain environment
« Reply #3 on: March 17, 2007, 11:26:21 AM »
The problem is it doesn't matter much what bots try to feed from. There's no methabolism so there isn't food a chain at all, just eat-all-you-see.

That's why I suggested before that bots had a membrane and enzyms sysvars, so the closer the enzyms to the target membrane, the highest feeding efficiency for the bot.
Sometimes you win, and sometimes you lose...

Offline shvarz

  • Moderator
  • Bot God
  • *****
  • Posts: 1341
    • View Profile
Diversity in plain environment
« Reply #4 on: March 17, 2007, 02:10:16 PM »
Well, the idea is to show that under certain conditions you can have non-exclusive co-existence of two species without spatial or any other separation. The competitive exclusion principle is one of the major rules in population genetics and ecology. There are well-known exceptions for that (like plankton), but they are still explained by some kind of separation between species.

Actually, I am thinking that the system that I'm proposing does not have to rely on the fact that species rarely interact.  That was a wrong approach.  The actual idea is this:

The average fitness of a species is set by two opposing forces. Natural selection leads to continuous increase in fitness. Mutation and stochastic events lead to decrease in fitness. At some fitness value these two forces equalize and the average fitness does not change anymore. So if there are two competing species and one of them gains some advantageous mutation then stochastic events will push its fitness down, back to the optimal level.  This creates a situation where both species have fixed fitness and one cannot out-compete the other.

I would like to test this idea.  For most species that have large population sizes and low mutation rates, the optimal fitness value is pretty high, because stochastic effects are minor.  Any experiment with such species would require a lot of time.  But for viruses (and I'm interested in viruses) the mutation rates are very high. Add to that the conditions under which the stochastic effects are important and the average fitness will be at some average level.  So it will have a potential for a going up or down a lot and will be fairly dynamic. This would allow to do experiments over fairly short time periods.

Here is an example of experiments with viral fitness in tissue culture: http://www.plosone.org/article/fetchArticl...al.pone.0000271
The time scale is several months, which is quite doable. And viral fitness changed 1.5-fold, showing that it is indeed very dynamic.  So I would start with two viruses that have exactly the same fitness and then let them go the way they did in this paper (large population sizes, deterministic increase in fitness) or in a system with a large number of stochastic events (that can be accomplished by transferring low number of virions from one culture to another).  I would expect to see that in the first case one virus outcompetes the other pretty quickly, but in the second case the viruses would co-exist for some time.
"Never underestimate the power of stupid things in big numbers" - Serious Sam

Offline Peter

  • Bot God
  • *****
  • Posts: 1177
    • View Profile
Diversity in plain environment
« Reply #5 on: June 27, 2007, 12:19:27 PM »
Can't you have two different specie's if you're using two total different kind of veggies. Example
veggie A using shell to protect himself, so shoot -6 does work less.
veggie B using poisonl to protect himself, so shoot -1 does work less.

This would favor 2 kinds of specie's and they would interact with each other without extincting the other specie. This is an example, I have'nt tested it jet but I gues this could work.
Oh my god, who the hell cares.

Offline Endy

  • Bot Overlord
  • ****
  • Posts: 852
    • View Profile
Diversity in plain environment
« Reply #6 on: June 27, 2007, 05:25:50 PM »
Maybe it'll work if the bots are likewise using shell and poison to defend themselves. Otherwise the body feeder will probably win out.

Offline Jez

  • Bot Overlord
  • ****
  • Posts: 788
    • View Profile
Diversity in plain environment
« Reply #7 on: June 27, 2007, 08:10:38 PM »
I imagine the repopulate bit for veg would also cause problems, it only works for the first bot on the list giving one of the veg an advantage.
If you try and take a cat apart to see how it works, the first thing you have in your hands is a non-working cat.
Douglas Adams

Offline Peter

  • Bot God
  • *****
  • Posts: 1177
    • View Profile
Diversity in plain environment
« Reply #8 on: June 28, 2007, 04:40:54 PM »
I am now running the simulation I was talking about 2 niches one using -1 shots the other -6. After just 20000 cycles not that long I now. The ones that are using -6 are just dominating the simulation with 500 bots against maybe 10 bots from the -1 shots, maybe I run it another with diffent settings to favor the -1 more. First see if the few -1 shot can keep up. Then I will see.


Thank's Jez for the use of stop instead of end for ending a gene.   Not anybody can make a mistake like that.
« Last Edit: June 28, 2007, 04:41:49 PM by Peter »
Oh my god, who the hell cares.

Offline Jez

  • Bot Overlord
  • ****
  • Posts: 788
    • View Profile
Diversity in plain environment
« Reply #9 on: June 28, 2007, 05:02:42 PM »
Quote from: Peter
Thank's Jez for the use of stop instead of end for ending a gene.   Not anybody can make a mistake like that.

I've spent years making mistakes like that, you are in good company!
If you try and take a cat apart to see how it works, the first thing you have in your hands is a non-working cat.
Douglas Adams