Author Topic: Co-Evolution  (Read 20035 times)

Offline Endy

  • Bot Overlord
  • ****
  • Posts: 852
    • View Profile
Co-Evolution
« Reply #15 on: October 17, 2008, 01:45:52 PM »
Maybe adding in Day/Night cycles, would help out. If the veggies are abusing the free nrg they get during the day then the night cycle would cause those to die off. Maybe implement the seperate costs for the veggies idea, if their movement costs were high enough they'd be less likely to develop as extreme evasion behaviors.

I wonder why real world veggies don't evade animals more, pretty much everything they do is a passive defense. Anything even close to active motion seems to require pretty extreme circumstances for them.

Offline jknilinux

  • Bot Destroyer
  • ***
  • Posts: 468
    • View Profile
Co-Evolution
« Reply #16 on: October 17, 2008, 02:15:17 PM »
Quote from: Endy
Maybe adding in Day/Night cycles, would help out. If the veggies are abusing the free nrg they get during the day then the night cycle would cause those to die off. Maybe implement the seperate costs for the veggies idea, if their movement costs were high enough they'd be less likely to develop as extreme evasion behaviors.

I wonder why real world veggies don't evade animals more, pretty much everything they do is a passive defense. Anything even close to active motion seems to require pretty extreme circumstances for them.

It's too huge of an evolutionary step in the real world to go from a non-moving multicellular organism to a moving one. Animals started out moving, so that's why it was easy for us to evolve limbs, etc... Plants are just stuck where they are because the chances of making an efficient way to move are awful, as opposed to DB, where all you need is one instruction.

Besides, there isn't really a big reason to evolve to move- the worst killers of plants are bugs, which cannot be evaded by moving, as opposed to moose etc.., which never harm too many plants. The only plants that I'm aware of that move are predatory, but their mechanism of movement is horrendously inefficient- they actually move by making their outer cells soak up water to expand. In order to move as fast as a slug, they would waste far more energy than if they just let the moose eat a few leaves and regrow the tissue. On the other hand, if they were poisonous or tasted like crud...



Anyway, this brings up a good point-

movement, etc.. in DB is way too easy. It's so unnatural, and so limiting. We need a much more complex movement system, like maybe a setting where single bots can't move at all- they have to form ties and the ties causing drag against the water makes them move.

Offline bacillus

  • Bot Overlord
  • ****
  • Posts: 907
    • View Profile
Co-Evolution
« Reply #17 on: October 17, 2008, 03:43:42 PM »
This has already been suggested before and won't be the last time; this really should be top-priority. Possibly adding currents would also make for interesting movement patterns.
"They laughed at Columbus, they laughed at Fulton, they laughed at the Wright brothers. But they also laughed at Bozo the Clown."
- Carl Sagan

Offline jknilinux

  • Bot Destroyer
  • ***
  • Posts: 468
    • View Profile
Co-Evolution
« Reply #18 on: October 17, 2008, 05:14:59 PM »
Quote from: bacillus
This has already been suggested before and won't be the last time; this really should be top-priority. Possibly adding currents would also make for interesting movement patterns.

I also just realised something- If reproduction was more complex, then we might not have to worry about mutation protection either. In living organisms, reproduction is so complex that if the young has problematic mutations, it'll automatically die because it can't do all the complex processes involved in development. If all a person had to do was say the magic word ".repro" and they'd reproduce, then we'd have tons of mutations and we'd also "evolve" into slugs that can't eat.

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Co-Evolution
« Reply #19 on: October 17, 2008, 05:37:54 PM »
You can turn off the .up, .dn, etc. movement by setting the voluntary movement efficiency in the costs panel to 0.  Bots will spend nrg to move, but nothing will happen.  Ties, etc. will still cause movement.

Offline jknilinux

  • Bot Destroyer
  • ***
  • Posts: 468
    • View Profile
Co-Evolution
« Reply #20 on: October 17, 2008, 11:17:53 PM »
Quote from: Numsgil
You can turn off the .up, .dn, etc. movement by setting the voluntary movement efficiency in the costs panel to 0.  Bots will spend nrg to move, but nothing will happen.  Ties, etc. will still cause movement.

OK- there are still two problems, though:

1: Where will I find a bot that only moves with ties?

2: How can I make/evolve a complex organism without cellular movement? There don't happen to be any amoeba-bots out there, right?

However, it does seem like the best temporary solution we have...



By the way, do you remember how you got E. Comesum and A. Minimalis co-evolving? What settings did you use? Did you tweak the DNA?
« Last Edit: October 17, 2008, 11:21:00 PM by jknilinux »

Offline Moonfisher

  • Bot Overlord
  • ****
  • Posts: 592
    • View Profile
Co-Evolution
« Reply #21 on: October 18, 2008, 09:28:13 AM »
I agree that it may be a good idea to have seperate costs for veggies in order to balance the energy spendt with the energy gained.
But it also seems like a good idea to have day and night cycles to keep the veggies from getting too strong.

I don't think veggies will always be shy... if the veggy cap is high then a veggy that kills other veggies could be able to slowly take over the sim and then kill off the bots... ofcourse the first step will always be to run away, but eventualy if there's enough veggies to fight the bots they will become agressive and get rid of their natural predators for good (This could also be something as simple as major shell production when shot at).
But a veggy will not always evolve to outrun the bots, this usualy happens because the bots kill the veggy, but often the veggies can evolve to a point where they escape alive and grow big again, and this is often enough. Ofcourse it would be better to loose nothing at all, but as long as the veggy survives and reproduces as well as the other veggies then natural selection will no longer encourage the alge to be more ellusive unless the bot gets better at eating it...

So with a litle luck and balanced size dependant energy gain and such you can end up with an alge that won't have a reason to evolve unless the bot gets better at chasing it, creating better chances for the bots and veggies to go hand in hand...

Still... in the end it is a messy buisness... there probably should be some sort of disadvantage to the nergy gain, either with seperate costs or a requirement in order to gain energy. For instance not gaining any energy while moving or shooting or creating resources... so a veggy only gains energy while it's doing nothing.
Or maybe having the veggies start out fixed and only gain energy when fixed... although I would imagine it would be hard for veggies to evolve past the first stage, having to unfix to escape predators and then fix again to gain energy...

But if you remove volontary movement it goes for both bots and veggies I think... and in the end the question remains, if it should be hard to move then how did the bots ever evolve movement. I know you can handcraft a bot that moves using ties and it would hold a great advantage over the alge, but in nature noone ever handcrafted the bots, so somehow the evolution of movement must have come at the expense of gaining energy or something like that...
I mean you're just creating a larger gap between bot and veggy, but eventualy with enough time the veggy would still surpass the bot...
I think it would be interesting to have either seperate costs, or some requirements for veggies, either having to be fixed to gain energy or just cutting the nrg gain whenever the veggy does something, forcing it to only do anything when strictly nesesary.
But even then I still suspect it will get ahead at some point since storing a litle bit of shell now and then would be enough to ward off the bots, just to take an example...

But I'm actualy wondering... what's the difference between veggies and bots in nature ? Is there realy a difference or is it just more or less the same thing where the veggies spend all their energy on creating more energy where the bots spend their energy on stealing it from others ?
I know we need to use "magic" when it comes to this sort of thing in DB, but I'm just curious as to how this stuff works IRL.

Offline Endy

  • Bot Overlord
  • ****
  • Posts: 852
    • View Profile
Co-Evolution
« Reply #22 on: October 18, 2008, 11:28:09 AM »
Quote
1: Where will I find a bot that only moves with ties?

Several of the inchworm type mb's can move without using up, using a combination of fixing/unfixing while lengthening/shortening their tie. May still be able to move using tieang, haven't seen anyone try it since Vermis_P though. Might be interesting to see if the zerobots could come up with something on their own.

Offline jknilinux

  • Bot Destroyer
  • ***
  • Posts: 468
    • View Profile
Co-Evolution
« Reply #23 on: October 18, 2008, 01:58:02 PM »
Moonfisher- I absolutely agree- when you run an evosim, whether it takes 1M cycles or 1B cycles, the simpler bots that get free energy WILL evolve to kill whatever it is that keeps killing them, even it it started out as some F1 superbot. I like your ideas, but the only problem is that they discourage interesting behavior. That's why I liked the pollination idea- it encourages interesting behavior, without using too much magic. What do you think about it?

And what do you mean by "What's the difference between veggies and bots in nature?" There's alot of them differences...

Endy- Are there currently any "Amoeba" MBs that move and eat with pseudopodia?
« Last Edit: October 18, 2008, 11:32:56 PM by jknilinux »

Offline Endy

  • Bot Overlord
  • ****
  • Posts: 852
    • View Profile
Co-Evolution
« Reply #24 on: October 18, 2008, 04:20:56 PM »
Inchworm2 still works somewhat well. I think .tielen may be broken, with the version changes. Could probably improve the MB's using the new eyepos, could scan around the MB and then use setaim to turn.

Offline Moonfisher

  • Bot Overlord
  • ****
  • Posts: 592
    • View Profile
Co-Evolution
« Reply #25 on: October 19, 2008, 08:43:12 AM »
I see the point of polination, but it doesn't realy seem natural that single cells would need the other species. And when something comes to polinate you it's also coming to eat you...
In the other scenario it's true that the veggies won't get too advanced since they need to take reglar breaks from doing anything in order to gain energy... but considering that a veggy that always moves will still have no disadvantages conpared to a normal bot then I don't think it's realy being hidered, I'm still concerned the veggy would eventualy take over, it definately needs to be outnumbered by ther bots before it starts to get too clever...

And this is also what I ment with the difference between veggies and bots IRL... as I understand it all life forms evolved from one kind of cell, wich I asume is a veggy... so why do the bots at some point stop gaining energy... why are the carnivorous species not able to gain energy from the sun, what happened to them ?
It just seems like the whole gaining energy deal is something that comes at a great cost, or rather that the behaviors of the bots are preventing them from gaining energy. So what is the difference that defines when you get energy and when you don't when it comes to nature ?

Offline Numsgil

  • Administrator
  • Bot God
  • *****
  • Posts: 7742
    • View Profile
Co-Evolution
« Reply #26 on: October 19, 2008, 04:16:52 PM »
Quote from: jknilinux
By the way, do you remember how you got E. Comesum and A. Minimalis co-evolving? What settings did you use? Did you tweak the DNA?

Nothing special.  It was E Comesum and Algae Minimalis, without modification.  Either F1 costs or I added some extra costs.  A little friction (not a lot though), with veggies getting fed 3 or 4 nrg per veggy with a reasonable cap.  Fixed bot radii, too.  Once the veggies become cancerous, it should be impossible for the bots to kill em all off.  Without the veggy repop, the alga in the sim will go through lots of generations very quickly.  Something like 10x faster than the bot generations.  You'll need a veggy cap to prevent an epic slowdown, of course.  E Comesum will keep its population in check pretty well (if there are too many, they'll eat each other), but you'll need at least a 3 or 4 sized field or the population won't be high enough.  With speed improvements, I'd go for the max size, actually.  The Enitor Comesum are fairly efficient at finding food (they don't sit still like animal minimalis), so you should be able to set up a fairly reasonable evo sim and let it run indefinately.  Tweak things as needed, but try to keep the bot population above 100 if you can.

Quote from: Moonfisher
And this is also what I ment with the difference between veggies and bots IRL... as I understand it all life forms evolved from one kind of cell, wich I asume is a veggy... so why do the bots at some point stop gaining energy... why are the carnivorous species not able to gain energy from the sun, what happened to them ?

In RL, I think the limiting factor is roots.  If you've ever been buried in sand at the beach you know that only a few inches of soil can be really, really heavy.  So a plant can't leave the spot it's in.  It could theoretically still have muscles and things for the above ground portion, except that muscles and the like have a constant metabolic cost.  Humans burn in excess of 2000 calories a day for only about 100-200 pounds of material.  A tree can weigh tons.  If it was made out of animal tissue the metabolic costs would be higher than it gets from photosynthesis.  So it minimizes metabolic cost by using plant tissue, which is strong and rigid and light to hold the weight without being expensive to maintain.

Microscopic single celled plants actually have a bit of limited mobility.  But even on the microscopic level plants try to maximize their surface area for light, which makes them so big that the cost of moving through water (which at that size is more like molasses than water) is prohibitive.

Offline ikke

  • Bot Destroyer
  • ***
  • Posts: 300
    • View Profile
Co-Evolution
« Reply #27 on: October 20, 2008, 02:35:18 AM »
Last save from my co-evolution

Offline jknilinux

  • Bot Destroyer
  • ***
  • Posts: 468
    • View Profile
Co-Evolution
« Reply #28 on: October 25, 2008, 08:21:34 PM »
Quote from: Moonfisher
I see the point of polination, but it doesn't realy seem natural that single cells would need the other species. And when something comes to polinate you it's also coming to eat you...
In the other scenario it's true that the veggies won't get too advanced since they need to take reglar breaks from doing anything in order to gain energy... but considering that a veggy that always moves will still have no disadvantages conpared to a normal bot then I don't think it's realy being hidered, I'm still concerned the veggy would eventualy take over, it definately needs to be outnumbered by ther bots before it starts to get too clever...

And this is also what I ment with the difference between veggies and bots IRL... as I understand it all life forms evolved from one kind of cell, wich I asume is a veggy... so why do the bots at some point stop gaining energy... why are the carnivorous species not able to gain energy from the sun, what happened to them ?
It just seems like the whole gaining energy deal is something that comes at a great cost, or rather that the behaviors of the bots are preventing them from gaining energy. So what is the difference that defines when you get energy and when you don't when it comes to nature ?

Well, it doesn't seem natural that bacteria shoot little bullets at each other either, but we have to live with it... Anyway, so long as it encourages complex behavior, I'm for it, and since we won't be simulating anything as complex as a bee anytime soon, I think we should let them pollinate each other.

And animals don't photosynthesize because they didn't evolve from a veggy, so your assumption is wrong.

Offline Endy

  • Bot Overlord
  • ****
  • Posts: 852
    • View Profile
Co-Evolution
« Reply #29 on: October 25, 2008, 10:58:21 PM »
I think it has to do with relying on Mitochondria vs. Chloroplast.

Still weird to think about having a long-term symbiotic relationship with it. Is it part of our species or not? Makes me wonder how a sexrepro'ing bot would do with an asex helper bot, maybe the union is somehow better than either alone.

We do actually use solar nrg to some degree for vitamin E. Pigmentation is probably the only real evolution going on with the movement of people from different areas with differing UV levels.