Author Topic: Evolution  (Read 7753 times)

Offline Anonomous Guest Person

  • Bot Builder
  • **
  • Posts: 85
    • View Profile
Evolution
« on: March 13, 2005, 05:21:42 PM »
New features are usually great, right..?
The only problem is they add up over time, turning something simple into something complex.
And complex things can become complicated.
And unfortunately, with the current mutation system, which is pretty much totally random as far as I know, only bad mutations seem to sprout.
Now, this will hopefully be fixed in version 3.0, but even so....
As a mere oppinion, I think that if someone decided to run an older version of DarwinBots, before Poison, and Venom, and such were added, as well as the current version of DarwinBots, the older version would have more good mutations then the newer.
So, for more "good" mutations, we either need a better mutation system which has more of a chance to make smaller, more related changes more frequently then larger, more drastic changes....
Or we need to make less commands. Which could be very difficult.
Or, on a third hand, we could consider it blessing that good mutations are rare. (Though, if you read another point I recently made, then hopefully you've figured out that good mutations can die off before they even start, thus making a mess of the whole natural selection idea.)
As yet another option, we could consider it a curse, and do nothing about it at all, and hope that programming methods will evolve in such a way that actual genetic mutations won't be needed for a species to "evolve."

Offline PurpleYouko

  • Bot God
  • *****
  • Posts: 2556
    • View Profile
Evolution
« Reply #1 on: March 13, 2005, 07:01:15 PM »
Here is an idea that I have been considering for a while.

How about we take the existing value and only allow it to mutate by a relatively small amount each time. This applies to both the value stored into a location and the value of the location.

The amount of change can be controllable from the mutations window.

The sysvars are reasonably well segregated into like commands at like memory locations so small changes would mean similar commands.

In this system a command like .repro could easily become .mrepro or .sexrepro but it would be much less likely to become .refpoison.

 :D  PY  :D
There are 10 kinds of people in the world
Those who understand binary.
and those who don't

:D PY :D

Offline shvarz

  • Bot God
  • *****
  • Posts: 1341
    • View Profile
Evolution
« Reply #2 on: March 13, 2005, 09:15:09 PM »
Good idea PY and AGP!  But large changes should still be possible - just less likely.  I am all for it!  This is a very good point and should improve DBs a lot without big changes!
 :clap:  :clap:  :clap:
"Never underestimate the power of stupid things in big numbers" - Serious Sam

Offline Zelos

  • Bot Overlord
  • ****
  • Posts: 707
    • View Profile
Evolution
« Reply #3 on: March 14, 2005, 06:26:32 AM »
I agree that big mutations should be very rarly, like in a dna friquency, ATGTCATG has a bigger chance to become ATGTTATG then TAGTTAA, it is of course possibly but very unlikly
When I have the eclipse cannon under my control there is nothing that can stop me from ruling the world. And I wont stop there. I will never stop conquering worlds through the universe. All the worlds in the universe will belong to me. All the species in on them will be my slaves. THE ENIRE UNIVERSE WILL BELONG TO ME AND EVERYTHING IN IT :evil: AND THERE IS NOTHING ANYONE OF you CAN DO TO STOP ME. HAHAHAHAHAHAHAHA

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Evolution
« Reply #4 on: March 14, 2005, 09:20:51 AM »

Offline Botsareus

  • Society makes it all backwards - there is a good reason for that
  • Bot God
  • *****
  • Posts: 4483
    • View Profile
Evolution
« Reply #5 on: March 15, 2005, 07:38:06 PM »
(There Bau you is , in conflict with everyone as usual)

If we dont add complex mutations then a robot will never develope new unique stategys.

Ex is my "firstbot": It simply repros and shoots and goes forward all the time.

But If we make the mutation system too simple and ristrictive then it will never learn to turn or use poison or shell or ties or eyes or setaim etc.

For stuff like that to work effisintly robots need to evolve a new "If BLA then BLA"
Instantaneously.

and math stuff too: in one shot it comes up with somthing like

For I = 0 to 10 'this is not a one cycle loop ;)
 c(I) = a + (b^( b-a))^(1/I)
next

will be great  :)
« Last Edit: March 15, 2005, 07:42:46 PM by Botsareus »

Offline Botsareus

  • Society makes it all backwards - there is a good reason for that
  • Bot God
  • *****
  • Posts: 4483
    • View Profile
Evolution
« Reply #6 on: March 15, 2005, 07:44:03 PM »
I like your ideas too guys , but we need the mutation to be evolutionary(your idea) and revolutionary(my idea) in the same time. Maybe its a good idea to combine the two.

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Evolution
« Reply #7 on: March 15, 2005, 11:58:05 PM »
The point of evolution is that it comes up with stuff gradually in small steps.  I think you're just impatient, Bots.  I mean, there are some improvements that could allow for some more realistic mutations, but they only (theoretically) increase the frequency of useful mutations.

You're using the same arguments most Creatonists use against evolution.  Like this guy.  Don't underestimate the power of, or time period of, natural selection.

Offline PurpleYouko

  • Bot God
  • *****
  • Posts: 2556
    • View Profile
Evolution
« Reply #8 on: March 16, 2005, 09:05:27 AM »
Where did you manage to dig up this idiot from Num?

Then again it isn't that hard to find people who know nothing about the TOE but it is difficult to find someone who knows so little about jellyfish.
Quote
And finally, the jellyfish & other species that were discovered to be 500 million years old (that's older than the dinosaurs) are JUST like the jelly fish today, and they have complex eyes, etc
HMMM?
Jellyfish with complex eyes  :blink:
Jellyfish don't have any eyes!

 :D  PY  :D
There are 10 kinds of people in the world
Those who understand binary.
and those who don't

:D PY :D

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Evolution
« Reply #9 on: March 16, 2005, 09:11:03 AM »
Ironically, this guy is the only one who's asked the question (exactly as I word it)

"why aren't all animals omnivores"

I think you can understand what I was hoping to come up with.  Such a dissapointment.

Offline PurpleYouko

  • Bot God
  • *****
  • Posts: 2556
    • View Profile
Evolution
« Reply #10 on: March 16, 2005, 09:27:59 AM »
I think it is just that if they were all omnivores then the competition would be too strong so they would all mostly starve.

If one species manages to evolve in such a way that it becomes more efficient at extracting food from a particular source (take humming birds) then it has no competition so it has the chance to thrive.

Evolution exploits niches. In a world without niches then they most likely would all become average omnivores.

 :D  PY  :D
There are 10 kinds of people in the world
Those who understand binary.
and those who don't

:D PY :D

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Evolution
« Reply #11 on: March 16, 2005, 09:47:32 AM »
Yeah, that's the general overview, but if I ask you the mechanics of how animals find niches, it'd take you a minute.  The links schvarz provided helped me out, but this was before then.

If an animal eats grass, why doesn't it learn to eat meeat too?  Well, to do so must mean it eats grass less well.  Or maybe not...  Arg!

Using google always gives me a headache.
« Last Edit: March 16, 2005, 09:48:12 AM by Numsgil »

Offline PurpleYouko

  • Bot God
  • *****
  • Posts: 2556
    • View Profile
Evolution
« Reply #12 on: March 16, 2005, 10:03:53 AM »
Num

Just remember what you keep telling Bots.

Evolution has no objective!

If something eats grass really well then it has no need to eat meat or anything else.

Also in order to efficiently eat meat it would have to change the type of teeth that it has so that it can tear instead of grind. This would make it less efficient at eating grass so the mutation would tend to be selected against.

It would be possible that such a mutation could be selected for if the particular animal that first gets it, also develops the tendancy to attack other creatures (aggressiveness). If this makes it more successful then it will pass those gense on until we get a herd of carnivorous cows.

It's just random dude!

Most random changes de-tune the species.

 :D  PY  :D
There are 10 kinds of people in the world
Those who understand binary.
and those who don't

:D PY :D

Offline shvarz

  • Bot God
  • *****
  • Posts: 1341
    • View Profile
Evolution
« Reply #13 on: March 16, 2005, 10:52:11 AM »
Quote
Don't underestimate the power of, or time period of, natural selection.


Just what my sig says :)
"Never underestimate the power of stupid things in big numbers" - Serious Sam

Offline Botsareus

  • Society makes it all backwards - there is a good reason for that
  • Bot God
  • *****
  • Posts: 4483
    • View Profile
Evolution
« Reply #14 on: March 16, 2005, 12:18:24 PM »
I got my stuff from my own (old dont have no more) evo. sim.

Basicaly a robot will take a long time to go from "doing nothing and spinning and shooting"  to  "chasing food and others" just because it evolves little bits of code per reproduction.

Although this problem can be fixed like this:

(This was talked about somewere in the other forums)

A robot keeps a gene [you]inactive[/you] and mutates it. The gene may actualy build up a nice code structure before the bot activates it.

So I am all for deactivating and activating genes.