NPC Auto-Evo Thread

GhengopelALPHA wrote:

The Uteen wrote:

I'm not sure how we can do it without just preventing extinction completely... Scio's idea or reacting to a mutation would mean about anything could change, and some could be much better at survival at other's, and things would therefore change. With this, they would all end up unable to either flourish or go extinct, just all be stuck with exactly the same threat level. Or even worse, they all become rocks, with no threat level at all.

Is this an overlooked flaw, or can you convince me otherwise? I give you 25.13274 hours to formulate a point I've overlooked or revision as necessary... Plus VAT.

Longishpost

@Uteen: I believe that our mutations will be small enough changes to the code that this shouldn't be an issue.

@Ghen: Careful of your wording. The mutations themselves have to be random. The species really has no "goal." Threat levels merely help streamline our method for determining success. This does, however, bring me to my next point. If we want to include this, we will have to be very, very careful with the equation. This system helps quantify many things, but as of yet I fail to see a method of determining the pure success of mutation. For example, the way you originally described this makes it seem that r-type orgs would be significantly favored over k-type ones.

@GhengopelALPHA
If I understand you correctly, the "threats" system would be a different way of modifying odds after a census.

Now, the idea behind having threats drive the entire population is good, since it does mimic reactionary evolution quite well. However, as a population decreases in the real world, it's chances of evolving also decrease, since the size of the gene pool is directly proportional to the probability of useful mutation. In this way, the model doesn't mimic evolution very well.

But what your model can do-extremely well-is predict useful mutations. By determining how a creature is threatened, we can determine what traits are selected.

So to modify the evolution process by implementing threats, we would run auto evo as is, but after step 4 (selection) we run the threat evaluation process on each of the selected species.

Tenebrarum wrote:

GhengopelALPHA wrote:

The Uteen wrote:

I'm not sure how we can do it without just preventing extinction completely... Scio's idea or reacting to a mutation would mean about anything could change, and some could be much better at survival at other's, and things would therefore change. With this, they would all end up unable to either flourish or go extinct, just all be stuck with exactly the same threat level. Or even worse, they all become rocks, with no threat level at all.

Is this an overlooked flaw, or can you convince me otherwise? I give you 25.13274 hours to formulate a point I've overlooked or revision as necessary... Plus VAT.

Longishpost X) sorry

@Uteen: I believe that our mutations will be small enough changes to the code that this shouldn't be an issue.

@Ghen: Careful of your wording. The mutations themselves have to be random. The species really has no "goal." Threat levels merely help streamline our method for determining success. This does, however, bring me to my next point. If we want to include this, we will have to be very, very careful with the equation. This system helps quantify many things, but as of yet I fail to see a method of determining the pure success of mutation. For example, the way you originally described this makes it seem that r-type orgs would be significantly favored over k-type ones.

I didn't really mean that the species has a goal; rather that evolution, when taking the whole population and a long period of time into account, gives a species a natural tendency to attempt to thrive. That is sort of a goal, I guess, so forgive me of my wording if you disagree.

And yeah, a lot of study and testing is probably going to accompany this method to get it balanced, but it helps the computer decide what kind of mutations have the greatest chance, as ~sciocont says it could.

Maybe we could do a for-gameplay exception to science here and make the more threatened species evolve faster? It would be very frustrating for the player to know his species are going to slowly die out because there aren't much of them left and they are probably not going to ever get an useful mutation because of it. And now that I think of it, what about crocodiles? They are very succesful predators, yet they have evolved little from the time they have appeared (same with sharks I think).

Commander Keen wrote:

Maybe we could do a for-gameplay exception to science here and make the more threatened species evolve faster? It would be very frustrating for the player to know his species are going to slowly die out because there aren't much of them left and they are probably not going to ever get an useful mutation because of it. And now that I think of it, what about crocodiles? They are very succesful predators, yet they have evolved little from the time they have appeared (same with sharks I think).

Perhaps there could be an exception for the player? And with the crocodiles and sharks, they haven't evolved much as a result of being so successful, so wouldn't creatures in game that are successful not evolve much? Even if they did evolve some, it would still mirror the real world, since sharks and crocodiles HAVE evolved somewhat.

Commander Keen wrote:

Maybe we could do a for-gameplay exception to science here and make the more threatened species evolve faster? It would be very frustrating for the player to know his species are going to slowly die out because there aren't much of them left and they are probably not going to ever get an useful mutation because of it. And now that I think of it, what about crocodiles? They are very succesful predators, yet they have evolved little from the time they have appeared (same with sharks I think).

I was under the impression that the player would have partial control over their mutations, aren't they?

As for crocodiles, case and point for using threats to chose which species to evolve. They are successful, and thus not threatened, so they don't get selected for auto-evo. In Scio's current concept, perhaps if the threat level is low enough, the selected species is kept the same? (skips the actually mutation step?)

I never liked the concept of player-edit evolution, but if it's the best way, then so be it. My knowledge lies elsewhere and I don't want to mess up the concepts of people who know better than me.

Leaving out mutations for the most successful species would mark the best spot somewhere in the middle between extinction and thriving, as long as Scio's concept of fewer population = less mutations appear. The mutation rate would have to be proportional to the threat value for it to really work.

Commander Keen wrote:

Maybe we could do a for-gameplay exception to science here and make the more threatened species evolve faster? It would be very frustrating for the player to know his species are going to slowly die out because there aren't much of them left and they are probably not going to ever get an useful mutation because of it. And now that I think of it, what about crocodiles? They are very succesful predators, yet they have evolved little from the time they have appeared (same with sharks I think).

That exact example has been bugging me for months. To allow successful species to stay successful, we might have to add something. I think if a species' threat levels are low enough when it is chosen to evolve, it will just not evolve. For this, we'll need to do things with death rate.
Perhaps if [births /deaths] in a generation is over 1, the species does not evolve? So If a species in severe decline is chosen to mutate, it will mutate, but if a successful species is chosen to mutate, it does not? This would make the mutations each generation less numerous (so it would make the mutation step less taxing on the computer), but definitely solves the problem.

Why not just make it less likely to be chosen for mutation as the threat level drops and cut out the middle man?

Tenebrarum wrote:

Why not just make it less likely to be chosen for mutation as the threat level drops and cut out the middle man?

Because we don't have a good definition of threats yet, and because you can still be threatened by things and have a growing population.

~sciocont wrote:

Commander Keen wrote:

Maybe we could do a for-gameplay exception to science here and make the more threatened species evolve faster? It would be very frustrating for the player to know his species are going to slowly die out because there aren't much of them left and they are probably not going to ever get an useful mutation because of it. And now that I think of it, what about crocodiles? They are very succesful predators, yet they have evolved little from the time they have appeared (same with sharks I think).

That exact example has been bugging me for months. To allow successful species to stay successful, we might have to add something. I think if a species' threat levels are low enough when it is chosen to evolve, it will just not evolve. For this, we'll need to do things with death rate.
Perhaps if [births /deaths] in a generation is over 1, the species does not evolve? So If a species in severe decline is chosen to mutate, it will mutate, but if a successful species is chosen to mutate, it does not? This would make the mutations each generation less numerous (so it would make the mutation step less taxing on the computer), but definitely solves the problem.

If we get the births/deaths, then make a higher number less likely, we could get the average of this and the threat level to determine whether it evolves. We shouldn’t stop the mutations for thriving species completely, if they are doing successfully they should still change sometimes.

The Uteen wrote:

~sciocont wrote:

Commander Keen wrote:

Maybe we could do a for-gameplay exception to science here and make the more threatened species evolve faster? It would be very frustrating for the player to know his species are going to slowly die out because there aren't much of them left and they are probably not going to ever get an useful mutation because of it. And now that I think of it, what about crocodiles? They are very succesful predators, yet they have evolved little from the time they have appeared (same with sharks I think).

That exact example has been bugging me for months. To allow successful species to stay successful, we might have to add something. I think if a species' threat levels are low enough when it is chosen to evolve, it will just not evolve. For this, we'll need to do things with death rate.
Perhaps if [births /deaths] in a generation is over 1, the species does not evolve? So If a species in severe decline is chosen to mutate, it will mutate, but if a successful species is chosen to mutate, it does not? This would make the mutations each generation less numerous (so it would make the mutation step less taxing on the computer), but definitely solves the problem.

If we get the births/deaths, then make a higher number less likely, we could get the average of this and the threat level to determine whether it evolves. We shouldn’t stop the mutations for thriving species completely, if they are doing successfully they should still change sometimes.

It will take some tweaking.

I'm not sure if this was resolved.
Concerning a scenario where your population number is so small you just no it will become extinct due to no useful mutations being produced

Smaller populations, in the real world have obviously, smaller gene pools.
Because of this, a single mutation occurring in the species is much more likely to become more frequent.

And as for this and the crocodile problem, consider the evolution of a species like a soldier. There are long periods of waiting, where there is no threat (or need to evolve due to high success, little directional selection). Random mutations may occur due to genetic drift or weak selection but only if largely beneficial.
Then there are very short periods of frantic rapid evolution when the species is under large threat (A new species in their habitat, loss of a resource, natural disaster, high death rates) and undergoes heavy selection. Only a few individuals survive which have the capabilities of surviving in the new environment. Their new population will be smaller, and thus less genetically diverse so more susceptible to genetic mutation in future generations should it occur.

It seems to be a common misconception that lower population numbers means less chance of evolving. While the mutation rates will be lower, they should have much more of an effect

Pezzalis wrote:

I'm not sure if this was resolved.
Concerning a scenario where your population number is so small you just no it will become extinct due to no useful mutations being produced

Smaller populations, in the real world have obviously, smaller gene pools.
Because of this, a single mutation occurring in the species is much more likely to become more frequent.

And as for this and the crocodile problem, consider the evolution of a species like a soldier. There are long periods of waiting, where there is no threat (or need to evolve due to high success, little directional selection). Random mutations may occur due to genetic drift or weak selection but only if largely beneficial.
Then there are very short periods of frantic rapid evolution when the species is under large threat (A new species in their habitat, loss of a resource, natural disaster, high death rates) and undergoes heavy selection. Only a few individuals survive which have the capabilities of surviving in the new environment. Their new population will be smaller, and thus less genetically diverse so more susceptible to genetic mutation in future generations should it occur.

It seems to be a common misconception that lower population numbers means less chance of evolving. While the mutation rates will be lower, they should have much more of an effect

What you just described is called punctuated equilibrium. However, lower population DOES mean lower chance of a helpful mutation (or any mutation, for that matter), so the current concept still holds. Lower population=lower chances of evolving. We're evolving an entire species at once, so we can't take population size into account they way that you are describing.

~sciocont wrote:

Pezzalis wrote:

I'm not sure if this was resolved.
Concerning a scenario where your population number is so small you just no it will become extinct due to no useful mutations being produced

Smaller populations, in the real world have obviously, smaller gene pools.
Because of this, a single mutation occurring in the species is much more likely to become more frequent.

And as for this and the crocodile problem, consider the evolution of a species like a soldier. There are long periods of waiting, where there is no threat (or need to evolve due to high success, little directional selection). Random mutations may occur due to genetic drift or weak selection but only if largely beneficial.
Then there are very short periods of frantic rapid evolution when the species is under large threat (A new species in their habitat, loss of a resource, natural disaster, high death rates) and undergoes heavy selection. Only a few individuals survive which have the capabilities of surviving in the new environment. Their new population will be smaller, and thus less genetically diverse so more susceptible to genetic mutation in future generations should it occur.

It seems to be a common misconception that lower population numbers means less chance of evolving. While the mutation rates will be lower, they should have much more of an effect

What you just described is called punctuated equilibrium. However, lower population DOES mean lower chance of a helpful mutation (or any mutation, for that matter), so the current concept still holds. Lower population=lower chances of evolving. We're evolving an entire species at once, so we can't take population size into account they way that you are describing.

Actually, I think we can, sort of: higher population = higher chance of species branching when a mutation occurs?
So when a species is under threat it is more likely the entire species will get the change, while larger populations which don't particularly benefit or lose from a mutation just branch out.

This simulates the faster spread of the mutation in a smaller population, and allows for unchanging species when successful without a (unrealistic) lower mutation rate for successful species. Plus, thriving species branching out adds to biodiversity. Thoughts?

The Uteen wrote:

Actually, I think we can, sort of: higher population = higher chance of species branching when a mutation occurs?
So when a species is under threat it is more likely the entire species will get the change, while larger populations which don't particularly benefit or lose from a mutation just branch out.

This simulates the faster spread of the mutation in a smaller population, and allows for unchanging species when successful without a (unrealistic) lower mutation rate for successful species. Plus, thriving species branching out adds to biodiversity. Thoughts?

For what we have right now this is very good.
But wouldn't the affected species with the mutation in the large population require genetic/geographic isolation? Otherwise there would be too much gene flow between the affected and unaffected population and the mutation would be unlikely to lead to a new species unless it is hugely beneficial. Whereas with a small population they can already be considered isolated.

Pezzalis wrote:

The Uteen wrote:

Actually, I think we can, sort of: higher population = higher chance of species branching when a mutation occurs?
So when a species is under threat it is more likely the entire species will get the change, while larger populations which don't particularly benefit or lose from a mutation just branch out.

This simulates the faster spread of the mutation in a smaller population, and allows for unchanging species when successful without a (unrealistic) lower mutation rate for successful species. Plus, thriving species branching out adds to biodiversity. Thoughts?

For what we have right now this is very good.
But wouldn't the affected species with the mutation in the large population require genetic/geographic isolation? Otherwise there would be too much gene flow between the affected and unaffected population and the mutation would be unlikely to lead to a new species unless it is hugely beneficial. Whereas with a small population they can already be considered isolated.

We can just boost the odds of a splitting mutation as population size increases.

Remember, all evolution happens within the biome, so a species is isolated to that biome if they are evolving within it.

~sciocont wrote:

Pezzalis wrote:

The Uteen wrote:

Actually, I think we can, sort of: higher population = higher chance of species branching when a mutation occurs?
So when a species is under threat it is more likely the entire species will get the change, while larger populations which don't particularly benefit or lose from a mutation just branch out.

This simulates the faster spread of the mutation in a smaller population, and allows for unchanging species when successful without a (unrealistic) lower mutation rate for successful species. Plus, thriving species branching out adds to biodiversity. Thoughts?

For what we have right now this is very good.
But wouldn't the affected species with the mutation in the large population require genetic/geographic isolation? Otherwise there would be too much gene flow between the affected and unaffected population and the mutation would be unlikely to lead to a new species unless it is hugely beneficial. Whereas with a small population they can already be considered isolated.

We can just boost the odds of a splitting mutation as population size increases.

Remember, all evolution happens within the biome, so a species is isolated to that biome if they are evolving within it.

I think this is more of a question of how the species-split works... ...How does it work? If they split within the biome, that isn't under isolation...

What if a larger population is more likely to branch off into other biomes, and when that happens the population is divided (aka. they undergo a species split) and each species is isolated to the biome, and so evolve separately from one another using the current biome-specific evolution system? A smaller species also has a raised chance of going to another biome when under threat, but the entire population is more likely to go together, and thus no split. Species doing moderately well, under no extremes, would be more likely to stay where they are (with a more equal chance of split and migration if they do go). When all species are doing moderately well, the biome is stable, and this also keeps biomes from mixing when both are doing well as they are.

When a threat level of a species in one biome is higher than the potential threat level they would have in a neighbouring biome, this could raise the chances of a migration (and split) into that biome. This will mean that if a neighbouring biome is suitable for life, but contains very little, plants will enter due to a very low potential threat level, then followed by herbivores when there enough to support them, again based on threat level, followed by predators. Any empty niches will quickly be filled out. Niches filled by inadequate species will be replaced by better one from nearby. The threat system would work very well in this respect. I think this has convinced me the threat system was a very good idea after all.

The Uteen wrote:

~sciocont wrote:

Pezzalis wrote:

The Uteen wrote:

Actually, I think we can, sort of: higher population = higher chance of species branching when a mutation occurs?
So when a species is under threat it is more likely the entire species will get the change, while larger populations which don't particularly benefit or lose from a mutation just branch out.

This simulates the faster spread of the mutation in a smaller population, and allows for unchanging species when successful without a (unrealistic) lower mutation rate for successful species. Plus, thriving species branching out adds to biodiversity. Thoughts?

For what we have right now this is very good.
But wouldn't the affected species with the mutation in the large population require genetic/geographic isolation? Otherwise there would be too much gene flow between the affected and unaffected population and the mutation would be unlikely to lead to a new species unless it is hugely beneficial. Whereas with a small population they can already be considered isolated.

We can just boost the odds of a splitting mutation as population size increases.

Remember, all evolution happens within the biome, so a species is isolated to that biome if they are evolving within it.

I think this is more of a question of how the species-split works... ...How does it work? If they split within the biome, that isn't under isolation...

What if a larger population is more likely to branch off into other biomes, and when that happens the population is divided (aka. they undergo a species split) and each species is isolated to the biome, and so evolve separately from one another using the current biome-specific evolution system? A smaller species also has a raised chance of going to another biome when under threat, but the entire population is more likely to go together, and thus no split. Species doing moderately well, under no extremes, would be more likely to stay where they are (with a more equal chance of split and migration if they do go). When all species are doing moderately well, the biome is stable, and this also keeps biomes from mixing when both are doing well as they are.

When a threat level of a species in one biome is higher than the potential threat level they would have in a neighbouring biome, this could raise the chances of a migration (and split) into that biome. This will mean that if a neighbouring biome is suitable for life, but contains very little, plants will enter due to a very low potential threat level, then followed by herbivores when there enough to support them, again based on threat level, followed by predators. Any empty niches will quickly be filled out. Niches filled by inadequate species will be replaced by better one from nearby. The threat system would work very well in this respect. I think this has convinced me the threat system was a very good idea after all.

A species split happens as a specific type of mutation. a species is simply split in two. A species can move into an adjacent biome if their current biome is no longer suitable, meaning that it doesn't have enough resources for them to live there anymore. In any case where a species moves biomes, that movement splits the population 50/50 in between the two biomes, so a movement acts as a split. This means that we simulate both allopatric and sympatric speciation.

~sciocont wrote:

The Uteen wrote:

~sciocont wrote:

Pezzalis wrote:

The Uteen wrote:

Actually, I think we can, sort of: higher population = higher chance of species branching when a mutation occurs?
So when a species is under threat it is more likely the entire species will get the change, while larger populations which don't particularly benefit or lose from a mutation just branch out.

This simulates the faster spread of the mutation in a smaller population, and allows for unchanging species when successful without a (unrealistic) lower mutation rate for successful species. Plus, thriving species branching out adds to biodiversity. Thoughts?

For what we have right now this is very good.
But wouldn't the affected species with the mutation in the large population require genetic/geographic isolation? Otherwise there would be too much gene flow between the affected and unaffected population and the mutation would be unlikely to lead to a new species unless it is hugely beneficial. Whereas with a small population they can already be considered isolated.

We can just boost the odds of a splitting mutation as population size increases.

Remember, all evolution happens within the biome, so a species is isolated to that biome if they are evolving within it.

I think this is more of a question of how the species-split works... ...How does it work? If they split within the biome, that isn't under isolation...

What if a larger population is more likely to branch off into other biomes, and when that happens the population is divided (aka. they undergo a species split) and each species is isolated to the biome, and so evolve separately from one another using the current biome-specific evolution system? A smaller species also has a raised chance of going to another biome when under threat, but the entire population is more likely to go together, and thus no split. Species doing moderately well, under no extremes, would be more likely to stay where they are (with a more equal chance of split and migration if they do go). When all species are doing moderately well, the biome is stable, and this also keeps biomes from mixing when both are doing well as they are.

When a threat level of a species in one biome is higher than the potential threat level they would have in a neighbouring biome, this could raise the chances of a migration (and split) into that biome. This will mean that if a neighbouring biome is suitable for life, but contains very little, plants will enter due to a very low potential threat level, then followed by herbivores when there enough to support them, again based on threat level, followed by predators. Any empty niches will quickly be filled out. Niches filled by inadequate species will be replaced by better one from nearby. The threat system would work very well in this respect. I think this has convinced me the threat system was a very good idea after all.

A species split happens as a specific type of mutation. a species is simply split in two. A species can move into an adjacent biome if their current biome is no longer suitable, meaning that it doesn't have enough resources for them to live there anymore. In any case where a species moves biomes, that movement splits the population 50/50 in between the two biomes, so a movement acts as a split. This means that we simulate both allopatric and sympatric speciation.

I think the system I suggested would be much more compatible/interconnected with other systems. Basing it on threat level really makes it work more realistically, rather than leaving it to chance. And a 50/50 split seems too organised for animals. Even more so for plants.

You know what? let's roll with that.

Tenebrarum wrote:

While we're on the topic though, threat levels could actually help remove the extinction monkey from our backs by letting species go extinct if their threat level reaches an arbitrary point.

Still wading through the stuff that happened while I was gone here, but how about an environmental stress/threat level? Taking into account the above, (posts surounding Tenebrarum's) population, and the availability of necessary food resources as parts of determing the level. We could assign some things some arbitrary numbers, and when the total stress passed an arbitrary point, higher chances of evolution. Then, if it passed the arbitrary maximum point, extinction.

Just tossing this out here.

An Organism/Species doesn't evolve by being threatened. It's just that if a population is under threat, it will generally decrease in numbers until a point where mutations in this smaller genetic pool will be more visible and less diluted (such as mixing a small amount of paint into a cup rather than the same amount in a vat).

The reason it seems like this happens is that a population that is threatened and doesn't evolve will go extinct.

GamerXA wrote:

An Organism/Species doesn't evolve by being threatened. It's just that if a population is under threat, it will generally decrease in numbers until a point where mutations in this smaller genetic pool will be more visible and less diluted (such as mixing a small amount of paint into a cup rather than the same amount in a vat).

The reason it seems like this happens is that a population that is threatened and doesn't evolve will go extinct.

We realize this, but threats are a good way to direct mutation, since a threat will select for certain characteristics that mitigate it.

How will the program be able to tell what the threat is caused by. Not to mention that there could be multiple solutions to a single Threat.