NPC Auto-Evo Thread

So wait all this is for one thing in the game? Okay it is nice to plan all this but lets think about something else for just a second, is a computer this day and age going to handle a high process game such as this. I mean from what I'm hearing this is going to be a high quality graphic, complicated evolving game. Do you think that a computer can actually handle all the processes that could go on here? But any ways these are very nice Ideas.

Background maths are easier to run than you might think. If we were using any music or graphics for these, it would be a different story, but the plain old math file is pretty small and simple.
Compare the size of a 4 page word doc to a 1 minute video (with sound) and you'll see why no one else is worried.

..and thus a tiny fish evolves onto land simply because random deviations in its stomach evolved the ability to eat grass? It would need to have some ability to walk too.

I say each organism has a list of attributes. Each niche has a list of requirements. An organism fits more or less based on the attributes it has. Attributes could be weighted and could replace eachother too. I don't see the problem with this.

No, if the fish's stomach was able to digest grass, it would still be limited by it's fins and stuck in the water, as it couldn't travel there. Should it emerge on land, it would die, and either [extinction: tiny fish] or its comrades that live in the water would happily ignore their grass-eating ability for years to come. We did mobility limits a while back, it should hold.
Besides, there's no specific [accept: food type: grass]. It would be something much more general, which could cover multiple species of food. [accept: food type: general vegetation] or [accept: food type: woody stems] would be more like it. Just because a species can do something doesn't mean it has to.

roadkillguy wrote:

..and thus a tiny fish evolves onto land simply because random deviations in its stomach evolved the ability to eat grass? It would need to have some ability to walk too.

I say each organism has a list of attributes. Each niche has a list of requirements. An organism fits more or less based on the attributes it has. Attributes could be weighted and could replace eachother too. I don't see the problem with this.

Yeah. I agree that there has to be a set of rules. However, that's going about defining niches, not the evolution process.

Something in the game has to tell a fish it cant walk.

roadkillguy wrote:

Something in the game has to tell a fish it cant walk.

That would be taken care of by animation procedurals. They wouldn't figure out a way for the fish to walk if it couldn't.

Could someone please start collecting all these procedurals together so we don't have to keep flopping across the same ground without feet? I for one think that, with Auto-Evo in a concept, we should grab all the pertinent data to cross-check situations just like these, in case there's something that we haven't thought of yet.

Let's start with:
Everything concrete about the OE
Other stuff about pre-sentience, such as behaviors
Whatever else we have lurking about Org and early civ.

Mysterious_Calligrapher wrote:

Could someone please start collecting all these procedurals together so we don't have to keep flopping across the same ground without feet? I for one think that, with Auto-Evo in a concept, we should grab all the pertinent data to cross-check situations just like these, in case there's something that we haven't thought of yet.

Let's start with:
Everything concrete about the OE
Other stuff about pre-sentience, such as behaviors
Whatever else we have lurking about Org and early civ.

Does this include the multicellular transition and template organism concepts?

A good proc site: http://pcg.wikidot.com/

@ Pez: Sure, why not?
@ Keen: Over my head, but I'm sure someone else here will find it helpful.

ADMIN wrote:

roadkillguy wrote:

Something in the game has to tell a fish it cant walk.

That would be taken care of by animation procedurals. They wouldn't figure out a way for the fish to walk if it couldn't.

I would think animation procedurals would come after niche finalization in the code.

It could go like this:

Organism randomization
Organism analysis
Organism placement and population (Based on analysis)
Final organism animation (honestly I have no experience with this)

Also, AFAIK that link doesn't have any actual algorithms, only lists of games that use them...

Hm. Finally checked through the site, and even though there is a page called pcg algorithms, it seems to contain brief info on the term or just a wiki link. Next time I'll have be more careful when someone suggest me a link...

Naah it's ok. I saw all the links and I honestly thought it would share how everything works.


Anyway, I think the problem is that NPC auto evo needs to know what will eventually happen in a given scenario. In other words, it needs to know all the possible evolutionary mismatches that would mean certain death to an species in the simulation, within a minute after it started. It should check for these after randomizing the creature.

Wrong atmosphere (Fish in air, monkey underwater)
Wrong food type (The animal evolves to eat something it cant digest)
Wrong transportation (Fins on land, legs in water)

I know there's more of these but I can't think of any more at the moment..

If a creature happened to evolve one of these randomly it would die in the simulation and become extinct unrealistically. (An animal would never evolve something that kills itself.) The game would then have to check for and fix them to 'simulate' evolution. This way, randomization wouldn't destroy the game.


My subconscious tells me that this isn't necessary, so feel free to fill me in if I'm missing something.

Mmm... excepting mutations which are one-offs which don't allow organs to properly funcion (and which aren't usually passed on because the creature dies off, or they're caused by errors when the DNA is replicated,) I can't think of much else.

Atmosphere/respiration actually isn't strictly tied to air vs. water in real life. Worms breathe through their skin, so some of them can breathe underwater too, and snails are pretty darn good at going from one to the other. Of course, some respiration types are limited by size. And of course, you could have gills and lungs.

.... why the animal would "evolve to eat something it can't digest" is honestly a mystery to me. Eating behavior is ruled by whatever the stomach type is in the OE. Niches for predators will be filled by pretators, niches for herbavores by herbavores. Omnivores are classified under predator (because of the consumer stage thing) but they have the ability to accept plants as food, and can be tagged to plants as a food resource.

I don't think the random "mutation per generation" would suddenly swap something so complex as the whole species' stomach chemistry. It would make legs longer, patterns more distinct, claws sharper, venom stronger - but not mess around with a whole organ system. Conservation of change.

Wrong transportation... valid. However, legged creatures often can and do swim, but they're not living in there... and some things can't swim at all. We'll have to figure out restrictions on that one.

roadkillguy wrote:

Naah it's ok. I saw all the links and I honestly thought it would share how everything works.


Anyway, I think the problem is that NPC auto evo needs to know what will eventually happen in a given scenario. In other words, it needs to know all the possible evolutionary mismatches that would mean certain death to an species in the simulation, within a minute after it started. It should check for these after randomizing the creature.

Wrong atmosphere (Fish in air, monkey underwater)
Wrong food type (The animal evolves to eat something it cant digest)
Wrong transportation (Fins on land, legs in water)

I know there's more of these but I can't think of any more at the moment..

If a creature happened to evolve one of these randomly it would die in the simulation and become extinct unrealistically. (An animal would never evolve something that kills itself.) The game would then have to check for and fix them to 'simulate' evolution. This way, randomization wouldn't destroy the game.


My subconscious tells me that this isn't necessary, so feel free to fill me in if I'm missing something.

Those would all be mutations that would make it jump into another niche/environment to compete.

~sciocont wrote:

roadkillguy wrote:

Naah it's ok. I saw all the links and I honestly thought it would share how everything works.


Anyway, I think the problem is that NPC auto evo needs to know what will eventually happen in a given scenario. In other words, it needs to know all the possible evolutionary mismatches that would mean certain death to an species in the simulation, within a minute after it started. It should check for these after randomizing the creature.

Wrong atmosphere (Fish in air, monkey underwater)
Wrong food type (The animal evolves to eat something it cant digest)
Wrong transportation (Fins on land, legs in water)

I know there's more of these but I can't think of any more at the moment..

If a creature happened to evolve one of these randomly it would die in the simulation and become extinct unrealistically. (An animal would never evolve something that kills itself.) The game would then have to check for and fix them to 'simulate' evolution. This way, randomization wouldn't destroy the game.


My subconscious tells me that this isn't necessary, so feel free to fill me in if I'm missing something.

Those would all be mutations that would make it jump into another niche/environment to compete.

My subconscious always knows... but I never do.

Quote :

I don't think the random "mutation per generation" would suddenly swap something so complex as the whole species' stomach chemistry. It would make legs longer, patterns more distinct, claws sharper, venom stronger - but not mess around with a whole organ system. Conservation of change.

Random is random. It will do whatever it wants. That's why I proposed these restrictions in the first place. Somehow we need to predict what will happen.

roadkillguy wrote:

Quote :

I don't think the random "mutation per generation" would suddenly swap something so complex as the whole species' stomach chemistry. It would make legs longer, patterns more distinct, claws sharper, venom stronger - but not mess around with a whole organ system. Conservation of change.

Random is random. It will do whatever it wants. That's why I proposed these restrictions in the first place. Somehow we need to predict what will happen.

My point being that we wouldn't necessarily program a change in the whole GI tract as only "one mutation" as it would be more like several - more than five, at least. After all, we wouldn't necessarily go from gills to lungs in one generation either, or suddenly jump from being a fish to having usable hind legs. We'd be a lungfish first, with strong back fins to scoot us through the mud, then have frog legs.

Random is random - but only to whatever variables we program to be equivalent to "one mutation." When it would take several mutations to achieve the affect IRL, we need to have steps between.

I see what you mean..

That means we need to have a tree of possible mutations for every part. We could then cascade the tree as necessary.


@scio
You propose that one species will be given the chance to evolve every generation no? What happens when a species' food source A. Becomes extinct or B. Becomes inedible? Is it given the chance to evolve?

roadkillguy wrote:

I see what you mean..

That means we need to have a tree of possible mutations for every part. We could then cascade the tree as necessary.


@scio
You propose that one species will be given the chance to evolve every generation no? What happens when a species' food source A. Becomes extinct or B. Becomes inedible? Is it given the chance to evolve?

Yes, that triggers a more likely "reaction mutation".

So effectively it's chosen first? What about the rest?

This is what I'm understanding so far:

You're a little frog hopping through a swamp. You primarily rely on a strange grass at the bottom of the ocean. Then, after a few generations, your special kelp evolved some toxins to prevent you from eating it. You were chosen to evolve right after your kelp.

My question is now, what would auto-evo do? Will it correct the frogs species to keep it living or randomize it's abilities and hope for the best? Maybe it should randomize multiple times based on the population and find the best fit? In that case, what is the best fit? What if you were chosen to evolve before your kelp? Would you be able to react?

roadkillguy wrote:

So effectively it's chosen first? What about the rest?

This is what I'm understanding so far:

You're a little frog hopping through a swamp. You primarily rely on a strange grass at the bottom of the ocean. Then, after a few generations, your special kelp evolved some toxins to prevent you from eating it. You were chosen to evolve right after your kelp.

My question is now, what would auto-evo do? Will it correct the frogs species to keep it living or randomize it's abilities and hope for the best? Maybe it should randomize multiple times based on the population and find the best fit? In that case, what is the best fit? What if you were chosen to evolve before your kelp? Would you be able to react?

Auto evo, finding that
-Your species is dependent on seaweed species
-Seaweed species has evolved a toxin

Will
-Increase your chances of evolving
-increase your chance of becoming immune to that poison

Notice that the seaweed evolving does not force you to become immune, and it also does not force you to evolve at all. Your species could
-evolve resistance
-evolve a different adaptation to deal with this (can't think of any others for poison, but in other respects thee would be something to fit)
-find a different resource (may involve invading another niche)
-do nothing (may cause extinction)
-move to a different biome

All of these are chosen at random as possible reactions to the original seaweed mutation.

Now, you may say that this will cause lots of extinctions. It will. But the truth is, we need environments to change quickly so that the game is interesting and dynamic, and so that you constantly have the chance to evolve your species (which I still think should be done either through this NPC procedure or direct edits).

Finally I understand it well, thank you Scio!

I like this a lot, especially the bit with extinctions.

I see... What if the frog species was chosen to evolve before the seaweed? The frog is nearly guaranteed to be extinct due to the fact that the seaweed technically hasn't evolved toxins yet.

roadkillguy wrote:

I see... What if the frog species was chosen to evolve before the seaweed? The frog is nearly guaranteed to be extinct due to the fact that the seaweed technically hasn't evolved toxins yet.

No, the frog developing toxin resistance doesn't affect it unless there are said toxins in its environment. If the frog develops toxin resistance out of the blue, it wouldn't affect the environment at all unless something in the environment directly linked to it has that toxin.

However, if the frog develops its own toxin, then any creature directly above it in the food chain gets an increased chance of evolving next, and in that evolution, an increased chance of toxin resistance.

So say the frog becomes toxic. An eagle, a large lizard and a snake all rely on the frog as a food resource. This means that each of these animals gets a heightened chance of evolving, say 10x more likely.

The probability of any one species evolving is based solely on their population which is determined by their size. Smaller size generally means higher birthrate, so smaller animals are likely to have higher populations. Therefore a mouse has a higher chance of evolving monstrous fangs that an elephant does of evolving a lack of fear for mice.

Back to our eagle, snake and lizard.
The lizard population of the biome is- 56
snakes- 47
eagle- 33

total orgs in the biome- 2087

therefore normally, their chances of evolving are
lizard- 2.6%
snake- 2.2%
eagle- 1.5%

If nothing happens to their food sources, these animals have a low chance of evolving (they would be apex or near apex predators)
Now that the frog has evolved, their chances of evolving increase.

total orgs in the biome- 2087
minus original populations= 1951

Multiply each target population (the populations of the predators) by the constant of reaction (the factor by which their chances are increased- we'll use 10 for now)

lizard-560
snake-470
eagle-330

new population- 3311 orgs

new chances of evolving
lizard- 16.9%
snake- 14.1%
eagle- 9.9%

Therefore, their chances of evolving at all are increased dramatically. Now, their chances of evolving toxin resistance versus any thing else- probably about 50/50.

If you'll notice, even if all of their chances are added up, it's still more likely that something else will evolve. This is because the frog either
- was not very important in the food web
- developed a mutation that did not affect many other populations

In this case, we don't really know much about the frog, so it could have been both or just #2, since becoming toxic doesn't affect whatever it eats. However, lets say a different animal, a small predatory mammal, becomes faster. This affects both its predators and its prey. Therefore the chance of a reactionary evolution because of this is very high. This is due to
-larger number of species affected
-lower trophic levels (its prey) contain more organisms.

It follows that the eagle snake and lizard, all of which prey on this mammal, get heightened chances. In addition to them, there a few species of small reptiles and insects that this mammal feeds on that get a heightened chance at evolving.

Now, simply because of the trophic level of the mammal and its mutation, the chances of a reactionary evolution for this mutation is over 70%

Now, remember a few things:
-Any organism smaller than 8cm is ignored.
-The difference between trophic levels is made smaller than the usual 90% population difference.
-The populations aren't real. The computer won't actually place 2087 different orgs in front of you at once, their populations are determined by some randomness and their trophic levels.


tl;dr
It works.