Experiment in Evolution.

The Artifical Life group now has 3 members, which is enough to keep it in place. I've confirmed with a Linden that we need 15 members in order to request a group forum, so as soon as it gets to that number I'll put a request in so we can have a forum for discussing the many aspects, possibilities and ideas for ALife in SL.

I love this kind of stuff. Richard Dawkins talked about his own efforts briefly in The Blind Watchmaker. What it showed was that evolutionary change could, in fact, occur at a very rapid pace given the right set of circumstances.

Computer Models of Evolution


In SL, though, short of creating tectonic shifts and occasional climate changes I'm not sure you'll have anything that's scientifically valid. For one thing, the most basic speciation usually takes thousands or even millions of years -- Gallapegos and the Faroe Islands being the exception. Those were cases where one segment of a species became isolated from its peers, preventing further genetic intermixing. The result was a hastened divergence in DNA, but one that still required several generations to observe.

Social Darwinism, however, would have a field day with SL. Even though it isn't exactly a one-for-one transfer from pure biological Darwinism over to social Darwinism, and the subject has endured some controversy, I think it's still a fascinating field of research.

I'm going to try to remember to join your group and keep up with what you're doing. Good luck

Cindy

Welcome to SL Cindy

I wasn't hoping to mimic RW evolution exactly. I listed the RW evolutionary pressures just to illustrate the point that we would need changing conditions of some kind. In the SL experiment, the conditions may have no RW equivalent.



I'm sure we could have thousands of generations evolving in a very short time

You got it
My critters most certainly do not use listeners lol, we were just making a point. In fact my bees do use simple tags. Named objects placed at critical locations around the place can be used to great effect.

The neatest idea just occured to me. I can script and I think I might join the group if it is open enrollment.

This is an idea for an algorithm and I do not have the entire implimentation yet. Here goes.

There is one Species.
Fish

There are two entities
Male
Female

There are 3 interactions
Male - Male.
Male - Female.
Female - Female.



There is a distance between two entities.

There is an invisible Prim called "The Center"
All entities want to be near "The Center" because it affords the best protection in the group.

Females want to be near Males (distance of 0)
Males want to be near Females (distance of 0)

Males do not want to be near other Males (MAX Distance)
Females do not want to be near other Females (MAX Distance)

The need to be near "The Center" overrides the need to be away from the opposing sex.
There is some percentage split here that we will call "Mood" which is a random variable for each entity. Sometimes it is 90% important for a creature to be near "The Center" sometimes it is only 20% necessary to be near "The Center" and they will venture farther from the center to be away from others of the same sex.

Now...
as far as intelligence in movement, all you have to do is move "The Center" The AI lifeforms will move with the center to maintain proximity.

I am going to start working on this. What do you all think ? To show behavioral patterns Female Fish will be brown and Male fish will be Gold,

oh hell... I have to get land first. *sigh*

anyway, what do you all think?

It was on System 9, written with code warrior on an old G3, I'll have it backed up on CD somewhere, thinking about it again today has made me want to dig it out again.

:: sounds of boxes shuffling around in the loft::

It certainly is open enrollment And it's not just for scripters either, it's open to anybody with an interest in discussing artificial life and SL. The purpose of the group at the moment is really to get a group forum where the subject can be discussed much easier, from there we'll just see what evolves

I can sort out a plot of land for you, or anyone else to use for these projects In fact when I am next in world, I'll find a suitable plot that we can use as an AI sandbox

if you look at any interaction with the environment as stimulus, then any reaction could be considered a response. it's a very mechanistic way of looking at things. it's one way of looking at mutagens.if you want unicelluar to become multi-cellular, then you probably need to go one step further down and deal with the "physics" (rules of interaction at the building block level) first. aim high for the moon and you might reach the sky.

learn to script change according to your environment

this has been fun, thanks.

Oh yes I agree. The thing is, mutations do not occur as a result of an interaction with the environment, they're pretty much random. The environment only really comes into play when it comes to determining the success or not of that mutated organism. Granted, mutations can be accelerated by certain environmental conditions, but in the main they play no part.



hehe it's certainly moved up to the top of my list of things I want learn next. Hopefully you guys maybe able to help me

Even more thoughts (sorry to continue this, I am just excited)

Each fish is assigned a random number between 1 and 20

Addition one: Selective Breeding and Reduction in Sexual Competition
Females want to chose males that are closer to their number, but do not have many other females near (remember females want to be a greater distance from other females)
This represents 'selective breeding' found in nature. Further it represents the desire to reduce competition.

Addition two:
Any male and female fish that attain a distance < 1 has a chance to produce offspring.
The chance for offspring is 20% - the absolute value of the difference between the two.

The Male will 'chose' the closes genetic matching female that is a distance less than 1

if both fish are a 5, then they have a 20% chance to produce a viable offspring.
If one fish is a 20 and the other fish is a 1 then they have a base chance of they have a
20 - (20-1) = 1% chance of producing an offspring.

The Offsprings Genetic Number is (Male + Female) / 2. Sex is randomly chosen.

Addition three:
Death: Each fish has a 3% chance of dying per cycle. This is modified by it's distance from "The Center" (EDITED TO ADD) chance of death increases by .5% each cycle (age) ... thus eventually we will see generations

The chance of death for each fish every cycle is base + 2*(Distance from "The Center"

I am thinking about adding a 'search for food' function that will force that fish from the center to go eat but I think I have enough to implement right now, that might come later. Now I have to figure out a working communication model for the fish and the center

Damn, I am excited LOL

Jessica

Your basic idea is good -- I'd only suggest that the "Center" occasionally and randomly move. There isn't any solid, stable target here except survival, and survival is dictated by the current location of that hypothetical "Center".

Cindy

Of course Cindy!

The Center will actually be in constant motion. That is how the group of fish "swim" they are forced to follow the center. The coding of the group movement and collision avoidance will only needed to be coded in the center. Well, the fish will be coded for avoidance detection of each other. The fish are coded to follow the center and interactions with one another through communication channels.

It wouldn't be much fun to watch the fish swirling about a static object. The center is the cause of the "schooling" activity of the fish.

The Center will be coded for movement, collision detection with the outside environment and avoidance, etc...

That's one way to do it. I wonder if it's possible to cover the parcel with tiles that would record the level of pheremones in the tile.. or perhaps one big tile that has some method of keeping track internally the values of pheremone at various points on it's surface, and would permit entites running around in the 'world' to query the values. This would allow you to construct decaying gradient fields of pheremone, and form the basis for the grouping, flocking, and search for food behavior described in Jessica Robertson's part of this thread.

Fun stuff.

--v

Ok, I misunderstood you (was trying to read while my boss wasn't watching )... I was into the whole genetic drift train of thought, and thought you were talking about how the school of fish's DNA would drift

Sorry. I'll shut up and listen now

Cindy

LOL it's okay, actually that was a continuation of my earlier post. It's all one idea, I just broke it into two parts because.... well... when I posted the first I hadn't actually thought of the second part that could be done as well.

Damn, 27 minutes till I get off work and can start working on this!

Jess

I wonder if we could have a Science Fair/Symposium covering Alife, Cellular Automota, Genetic Algorithms and Genetic Programming?

I'd be happy to provide a modest amount of land for the symposium.
It would be great to see posters for all the interesting work that's happening.

My personal interests lie in the areas of Genetic Programming, evolution, memetics and molecular biology.

Can anyone put together a basic genetic programming (GP) toolkit in SL?

John Koza is my hero.

(To think I almost gave away my GP textbooks. I'd better go save them.)

I'll sign onto the group as soon as I arrive inWorld.

The magical center exists for my fish though with a major difference. The 'center' is different for each fish.

Some fish can't see the center (or other fish) and so follow a 'random' path untill they spot other fish to shoal with.

For anyone wishing to experiment with fish the lake in Aretias/Hypatia is available.


Jessica, I really hate to say this, but most of what you're describing my fish already do...

Cool! Is your code permissioned to (copy)(mod) so we can all benefit?

Group is now up to 7 members, Hiro Queso has made some land available in Parce for the group as an AI sandbox (actually it's about 1/2 the sim!) which is very kind of him. It's set to group and there's an auto-return for non-group objects.

We need 8 more members before we can ask the Lindens for a group forum, so hopefully that won't take much longer to happen

heh. interesting thread. especially since I was working on some breeding ai fish myself for krittannia based on a simulation I wrote about a decade ago. actually, the ai bit was modular and for 'any' wildlife to react to any other, but i was starting again with fish

If I wasn't leaving SL I'd be interested in joining. But I shall be interested in seeing your progress sometime anyway!

The group is always welcome in Aretias and Hypatia (Collectively, the Amazon Nations). While permanent building is limited to Amazons, auto-return is set for two hours for non Amazons, so AI sandbox activity is not a problem.
Hypatia is effectivly empty of builds (there are a few things scattered around) as it is intended as a nature sim. Lots of trees, lots of water, etc. and a nice high sim fps as a result.

Any female members are welcome to join Amazons, guys...ummm, sorry.

If 2 hours is too short I'll happily increase the auto-return time.

*bites tongue*.........hmm I think I will stick to Parce then

Hiro, I didn't mean you're not welcome Everyone is welcome, it's just that by definition Amazons are female.

I was swimming with your fishes last night. Managed to get a piccy of one of the shoals onto Snapzilla (after about 20 attempts to capture them in a screenshot - fast moving little thingies that they are!):

http://www.sluniverse.com/pics/pic.aspx?id=9066