Auto-Evo in Multicellular Stage

Current Auto-Evo is working well, but in multicellular stage things are getting more complex since tissues will be availiable and
these tissues should be placed in some order, right? Right now Auto-Evo is working like this:
Place organelles randomly
2. Check how fit species is with newly placed organelles
But in case of multicellular stage placing cells in random places might not be best idea.

Organisms are adapting to perform or improve one of these:

  1. Reproduction
  2. Defence against decay
  3. Resource acquisition
    We can give cells roles depending on their functions, for example digest cells which help in resource acquisition. We can
    then split these cells into 2 groups: inner layer and outer layer. And now we can put these cells in correct order to make
    species stronger.

I drew some image in paint where i segregate some examples of cells.

Do you like my idea?
I’m open to hear your suggestions.

(if there is already existing topic/reply or this thread should be moved to existing thread, you can do it (if you are moderator))

(my 1st post by the way)

4 Likes

That is not a bad idea to start in the macroscopic stage.
Although I’m not directly involved in game development, I know such a division can make some order.
But the problem I saw here is not related to you, these are one of the things I remember about the development stage of this stage:
They plan to center the player as the basis for all the creatures that come after him.
At first, it was based on what the player created, but later it seemed that Auto-Evo would encounter a number of issues that would require developers to think deeper.
But let’s leave that aside, the idea of organizing the types by categories is to be something that will help the center a few things, but I want to share something that might help you further develop your idea since some of it has been talked about in the past (by me):

But in short: there are 3 layers from which all things in the body develop. Maybe you can use that as a basis to develop this idea.
As for Defense against decay, this is something that happens regardless as part of the natural way for cells to maintain themselves from breaking down, so I think it’s less relevant and if it’s related to the skin or dermis, I think it’s worth considering as the ‘outer layer’.
As for reproduction, this is probably the first thing in specializing in the process of creating the first multicellular, and there are many ways to reproduce, and I think it’s a great idea to separate it from the rest.

Besides, it’s a great idea, but I have no idea what’s going on there.

1 Like

Kind of sounds like the germ layers idea posted recently on the dev forums:

Another thing: placement of cells with unique functions, such as defense.
When you are looking at Ankylosaurus and you are thinking why it has this armor-like cover at top, instead of bottom or somewhere else, here’s the answer: that point on its body is very vulnerable.

Same goes for Triceratopses. Their collar is where it is to protect the neck which is also very vulnerable point.

I want to suggest attaching some piece of code to current auto-evo algorithm which checks what spot is being attacked the most of time and then put some bonus protective cells over it.

(And for organisms that have esophagus I suggest something that will place the mouth in point where most food is encountered i guess?)

(sorry if my English is bad)

2 Likes

ok i came up with something:
4 things for multicellular auto-evo:
1 - Structure
2 - Layer map
3 - combining them together
4 - adding insides

Structure is basically shape for organism. it can be anything

Layer map is order in which layers are placed.

If we combine these we can get a nice organism with nice layers.


But it needs insides, like stomach
so we add it
image

and we have this:

of course auto evo can’t add 40 cells in one editor session. What it does then is giving itself goal of how organism will look like, but also changes its plan a bit depending on its changing environment.

Why not add individual variance, each creature is slightly different than it’s parent creature, and the ones that survive carry on the species. The idea is that auto evo is pretty much the same, but we add a gene varience script that allows for a few more mitochondria to be placed in the cells, or a little difference in color from creature to creature, before the creature is no longer the same species, and if all of the species with a particular variant die out, it will more heavily prefer varience in a different direction.

Very, very, very early on in Thrive development it was known that we can’t do a full simulation where you’d have genetic variants going around and the game would simulate them reproducing and passing on the best genes.

If someone wants that conclusion to be wrong they have to show how that approach can be computationally feasible.

5 Likes

correct me if I’m wrong but couldn’t the organism’s structure be compressed into 2 parallel strings of letters with the same rules as our DNA and the locations of the organelles also encoded but only when parts are skipped or moved a line down as well as the rotation being encoded into the string of letters representing it’s DNA and further compressing them into a string of 1s, 2s, 3s, and 4s or any other 4 characters with AT turned into 1, TA turned into 2, CG turned into 3, and GC turned into 4, and it is only read when trying to spawn new organisms of the species that the DNA belongs to and generate the template for it’s arrangement and what it has and the organism just multiplies it’s organelles if it is a single cell or uses it’s template generated from it’s genetic code to produce new cells in the correct order(for early multicell) and just grows to its next life stage based on it’s template generated from it’s genetic code. this system would make organelle hoarding less viable and make it possible to not have to run auto evo before loading the editor on lower evo rates even if you have run auto evo during gameplay off (at least for your current patch which encourages finding a good patch and staying in it) thus cutting down the time it takes to edit your cell/cell cluster but a slightly different but still having the same base idea method would likely still have to be used for macroscopic as there would be a whole new axis added. there could also be markers for the layers of the cells (if cell layers are added) to stop the automatic merging of cell layers. the player’s DNA file would only be read when splitting to give variation from the player to the new AI cell of the player’s species. it would probably be best to not have this active in any patches the player is not in and just run normal auto-evo for all patches other than the player’s patch as it would only be visible if the player is in the patch it is occurring in and would otherwise waste computing power that could be used for rendering the current patch.

the game could also flip a 1 into a 2 or 3 in it’s DNA file if the organism went near a radioactive object(for AI cells) to simulate radiation damaging DNA and the cell repairing it with a few errors sometimes and obviously each species in the player’s current patch would get it’s own DNA folder as well as any species that the player had fossilized so they could be imported into the world in whatever patch the player wanted. the folders for the DNA would have something to tell the game which DNA file was the most common in every patch (to simulate geographical separation causing speciation) and use only one file for non player occupied patches and use that to spawn members of the species if the player enters the patch that DNA file occupies

You are bunching up a ton of different points into a hodgepodge of ideas. Here’s what I can see:

  • Switch the game to represent species with a genetic code where organelle information is encoded with short genes. I see at least one huge problem: how are organelle upgrades going to be saved?
  • In multicellular the gene code could be extended to add a third dimension to positions, though I fail to see how that would actually work… the convolution surface approach means that all macroscopic creatures have to be made up of a graph of metaballs.
  • Not run the auto-evo for the player patch for some reason? (well I kind of get the point but making 2 separate algorithms one full and one cut down for the player patch doesn’t sound like a very productive use of time, right now even with just one algorithm we have many gameplay features it hasn’t been updated to know about) This would make it so that a player speedrunning through generations would not even see any species really changing.

In conclusion, my opinion is that due to various constraints with the game design and realities of restricted resources, none of these ideas are usable as is. Some of these could be tweaked to be used, but then again I don’t see the point (for example as finding ways to represent species colouring, organelle upgrades, behaviour values as gene code just adds a ton of complexity but doesn’t really improve anything, mutation rates of various things can already be tweaked). These are the kind of suggestions that sadly a person who isn’t both a programmer and a bit familiar with the Thrive code and future goals, can’t make correctly.

1 Like

i will try to think of a solution for that problem. maybe another line to keep track of the upgrades? if not i will reply later with a better solution

that’s why i said there would need to be a slightly different system for that as there would be a new axis to work with.

that would only be at low enough values that otherwise auto evo would basically be off(anything below 1).

i also added a bit to my previous reply

All of this intresting honestly I really like that there was a disscussion like this.