|
Post by robertthebobert on May 24, 2016 23:04:51 GMT
From the moment an organism becomes multicellular it's cells can start to differentiate and specialize. I've been seeing talk of a caste systems for individuals of a species so you can create things like ant colonies (workers, queens, etc) but it's also important for the multicellular (maybe aware) stage's cells, so you can start making things like muscle cells, neurones, even stem cells. I'm thinking of being able to develop each type of cell independently kinda like in the microbe editor but being able to the drag and drop them into place to create a functioning multicellular organism . It's probably been discussed but I couldn't find a thread relating specifically to this.
|
|
|
Post by Atrox on May 25, 2016 1:44:01 GMT
Yeah it's definitely been discussed. I'm not so sure if we've ever talked about actually dragging and dropping premade cells into your organism. That's pretty cool.
|
|
|
Post by StealthStyleL on May 25, 2016 7:45:54 GMT
I'm pretty sure that it's planned to be able to drag and drop them. Once you get to the Multicellular Stage, you open a semi-new editor, where you can create more than one type of cell, mimicking differentiation. Then, you drag these pre-made cells onto the screen to create your multicellular organism.
It was probably discussed on the old forum which is why you can't find it.
|
|
|
Post by tjwhale on May 25, 2016 9:19:13 GMT
Yeah that's basically how it's going to work, it's going to shake out pretty complicated though.
|
|
|
Post by Atrox on May 25, 2016 11:58:07 GMT
Yeah that's basically how it's going to work, it's going to shake out pretty complicated though. How so?
|
|
|
Post by tjwhale on May 25, 2016 13:51:20 GMT
Yeah that's basically how it's going to work, it's going to shake out pretty complicated though. How so? I'd like to caveat this by saying I'm a bit uncomfortable saying this. It's come up before and we have talked a bit about it but no decisions have been made. I guess it's no problem for the wider community to be in on the discussion. It's also possible I have misunderstood and this is not a problem, or that the others will have clever ideas (they usually do they are awesome!) which will neatly get round all this. Anyway... So there's a general question with the game which is "how much impact should the choices you make in one stage have on the next stage?" As an example when you move into the Society stage what difference does it make what your species is like? So if you're Centaurs or Birds or Lizards does that make any difference? If so what? If you are birds it would be nice if you had much greater movement speed but reduced ability to transport goods, for example. Whereas if you are Centaurs then maybe you should be able to carry a lot of goods on your back as you are mostly a horse. Obviously Horseback riding doesn't really make sense if you are a Centaur already. But the problem with this is that the range of species you can generate is vast. We can't put in a rule that says "If you are a Centaur then horseback riding is out" because we would need to think of every possible species and make such rules for all of them. (In fact come to think of it why might there be horses? We will probably need some sort of "ridable animals" algorithm, but that in itself is a lot of work). So yeah if each stage has no effect on the next then it's going to be disappointing. However forging the links is going to be very complicated. This applies to the multicellular. So you start out with just copies of your original cell. Then can you select "turn this cell into a muscle cell"? If so does it have any memory of what it was before? Like maybe it was a hydrogen sulphide respirating cell with a thick plant cell wall, can it turn into a muscle? If not why not? Basically are we just going to offer a generic pallette of cells, muscle, neuron, bone etc which basically ignore all the choices you made in the cell stage. That seems kind of a shame as it wipes out everything that has come before. However how can that info be carried over in to the next stage? If you are a hydrogen sulfide respirator surely that info is important as it means you are stuck living by the vents and can never leave without going back to the cell stage. If you are a plant surely that means you can't get to the aware without going back and making an animal cell. But then how do we make an algorithm which can look at a cell and tell you what kinds of cells it can become (like bone for example) or not become (like muscle for example). If there are 10 optional organelles then there are 1024 different combinations you can make, if you could have between 1 and 10 of each of these then that's 10^10 different possible cells. Maybe it's possible to make rules to cover them (like if you have a cell wall then you cannot become a muscle cell etc) but what about later when you can design all kinds of crazy creatures? You don't want to be some massive dragon and get told "sorry you can't carry much in the society stage because you have wings and wings means low carrying capacity", that's going to feel pretty lame. Anyway yeah it's just a problem with no solution right now.
|
|
The_Wayward_Admiral
Spacefaring
The_Real_Slim_Shady
Atrox drew this awesome image of the Keldori!
Posts: 1,011
|
Post by The_Wayward_Admiral on May 25, 2016 20:20:29 GMT
tjwhale , has anyone proposed the following? In multicellular you start with two copies of your cell, but you can edit new cells with mutation points as you add them to your organism. Over time you start making distinct cell types to carry out specific duties, and eventually the player ends up creating tissue types de facto. At the end of multicellular, each cell type in the organism becomes the building block for actual tissue types in Aware. Then each organism either has very specific tissue types each game, or some algorithm examines a cell's main functionality and places it in a preset tissue category. let me know if this is unclear/foolish/impossible/already suggested. edit: So for example, if I made a cell type that had a lot of mitochondria and myomeres, then I could classify it as my "muscle tissue", or the game would do it for me after I spent the multicellular stage creating it. And perhaps there could be something like the symmetry editor to place multiple of the same cell type together.
|
|
|
Post by Atrox on May 25, 2016 21:29:20 GMT
I would have had this much earlier but my dog kept bumping into my laptop and erasing it. 1) But the problem with this is that the range of species you can generate is vast. We can't put in a rule that says "If you are a Centaur then horseback riding is out" because we would need to think of every possible species and make such rules for all of them.And why not? The game can have a rule that states if an organism has > 4 legs it is not able to ride another organism. Or it could be decided by weight. A horse sized centaur obviously would not be able to ride a horse because it'd break its back, but what about a schnauzer sized centaur? Just some thoughts. 2) This applies to the multicellular. So you start out with just copies of your original cell. Then can you select "turn this cell into a muscle cell"? If so does it have any memory of what it was before? Like maybe it was a hydrogen sulphide respirating cell with a thick plant cell wall, can it turn into a muscle? If not why not?
I definitely don't think that it should or would be as easy as pointing and clicking and "Oh hey look now it's a muscle cell." That's way too easy. Every change we make to our organism has to be slow to properly simulate evolution. We can't just make jumps like that, because then it might as well be Spore where we add 5 pairs of arms in one generation just because. 3) Basically are we just going to offer a generic palette of cells, muscle, neuron, bone etc which basically ignore all the choices you made in the cell stage. That seems kind of a shame as it wipes out everything that has come before. However how can that info be carried over in to the next stage? If you are a hydrogen sulfide respirator surely that info is important as it means you are stuck living by the vents and can never leave without going back to the cell stage. If you are a plant surely that means you can't get to the aware without going back and making an animal cell.
No generic palette! That's boring and makes the entire microbe stage pointless. I'm pretty sure Thrive is intended to have these dead-end pathways as well. We're going to be allowed to play as plants, why not hydrothermal vent life? Another member on the forum suggested that the microbe stage have notifications and warnings to let the player know if they're headed down one of these dead-ends. 4) But then how do we make an algorithm which can look at a cell and tell you what kinds of cells it can become (like bone for example) or not become (like muscle for example). If there are 10 optional organelles then there are 1024 different combinations you can make, if you could have between 1 and 10 of each of these then that's 10^10 different possible cells. Maybe it's possible to make rules to cover them (like if you have a cell wall then you cannot become a muscle cell etc) but what about later when you can design all kinds of crazy creatures? You don't want to be some massive dragon and get told "sorry you can't carry much in the society stage because you have wings and wings means low carrying capacity", that's going to feel pretty lame. Anyway yeah it's just a problem with no solution right now.
...I might have a solution? Let's say you begin as a 5 cell colony made up of the original cells. You've just begun the multicellular stage, where do you go from here? STEP ONE: SPECIALIZATIONYou enter the microbe editor like normal except what's this??? There are now two layers of editors :0000 By that I mean there should be a colony editor (where you place and remove cells) and the regular cell editor (you click a cell on your colony to enter and make changes to that specific cell). In order to specialize your clone colony (hereby dubbed clonony), you click the cells and gradually make changes until you unlock a cell type. Cell types can then be dragged and dropped into the colony editor, like organelles are dragged and dropped into the cell editor. An example: You're a 5 celled clonony. You enter the editor, click on Cell 1 and add an obscene amount of mitochondria. This becomes the first muscle cell and unlocks it in the colony editor. Cell 2 is fitted with calcium carbonate secretors and becomes the first bone cell. Cell 3 is filled with eyespot apparatus and becomes a photoreceptive cell (more so than an eyespot apparatus, even though the player will be able to see light anyway so it doesn't make much sense but that's for a different discussion). Cell 4 can be filled with lysosomes, so it becomes specialized as a digestive cell. Cell 5 is turned into a neuron somehow (I'm not entirely sure what could be added to a cell to make it a neuron. Maybe adjust the shape to give it dendrites and axons??). You now have a fully specialized colony. But how do you grow? STEP TWO: REPLICATION
Well you've been swimming around as a specialized colony now, but you want to grow. You take a trip into the editor and this time you use the colony editor. In the colony editor you should see a list of cell types that you have unlocked by specializing existing base cells. You can drag and drop these cells from the list, to your colony, showing an increase in the number of cells each generation. An example: You are playing as a 5 celled colony made up of a muscle cell, a bone cell, a photoreceptive cell, a digestive cell, and a neuron. You enter the editor and see the list of unlocked cells. In this list you see the original base cell (or a stem cell or something), and all the currently existing cells in your colony. You decide to spend your mutation points one extra digestive cell, one muscle cell, and one bone cell. And yeah you continue until you've unlocked things like muscle fibers (stringing muscle cells together), proper bones (having clumps of bone cells), and organs (clump a bunch of digestive cells and hey let's call that a stomach), etc, etc. Also for things like electric organs and stuff, there may need to be more than just the 10 organelles. What do you think? EDIT: All my work was for naught. Wayward explained it much better than I. I spent so long too
|
|
The_Wayward_Admiral
Spacefaring
The_Real_Slim_Shady
Atrox drew this awesome image of the Keldori!
Posts: 1,011
|
Post by The_Wayward_Admiral on May 25, 2016 21:36:04 GMT
Jibbers Crabst. I'm really sorry Atrox, that was the most villainous case of ninja'ing I've ever performed. Although I would say that you explained the concept far above and beyond what I did.
|
|
|
Post by Atrox on May 25, 2016 21:41:58 GMT
Eh don't worry about it, it's pretty funny honestly. And you know what they say, great minds think alike 
|
|
|
Post by tjwhale on May 25, 2016 21:51:41 GMT
It's interesting what you're saying. I personally feel unconvinced that we've got to the core of the issue. It's whether all neurons are the same and have the same function. Because if so that's a generic palette. Like Atrox when you say " This becomes the first muscle cell and unlocks it in the colony editor. Cell 2 is fitted with calcium carbonate secretors and becomes the first bone cell." Do you mean these are now generic muscle and bone cells or do you mean that they will still contain lots of the information from the cell stage? Like what happens if I add both mitochondria and calcium carbonate secretors? What kind of cell is it then? How does the game know? Again same with you The_Wayward_Admiral "or some algorithm examines a cell's main functionality and places it in a preset tissue category." Do you mean "it's now just become a generic version of that cell, like a generic muscle cell" or do you mean writing an algorithm that can intelligently look at any cell with any combination of organelles and tell you what it is? Like what if it's 50% muscle and 50% neuron? What if it's not really anything recognisable? Do you see what I mean? Do you see how hard it is for it not to be a generic muscle cell but also to mitigate the need for being able to recognise an arbitrary cells abilities? If, in the end, you have a generic palette then that works well but is a bit lame. If we're keeping layer upon layer of information it's difficult. We need, for example, to be able to model the effectiveness of different types of hypothetical cells at being neurons. That's hugely complicated. How is a neuron affected if you add a flagella to it? How is it affected if you make it's cell wall 8% thicker? Do you see how complicated that question is?
|
|
|
Post by Atrox on May 25, 2016 21:54:23 GMT
Yeah.. wow okay. I've got nothing at the moment tjwhale , what if it had to do with a ratio of certain organelles. The more mitochondria you put in a muscle cell the more effective it is? I dunno maybe generic cells is the way to go. When you said generic cells I thought you mean we have them all from the start.
|
|
|
Post by tjwhale on May 25, 2016 22:06:28 GMT
"When you said generic cells I thought you mean we have them all from the start." I agree that would be a bit dull and feel like a big jump from the cell stage. Maybe one solution is for the game to have, pre-programmed, the generic cells. Like what we think the optimal muscle cell is, for example. It would have one of these for each tissue type, muscle, bone, neuron etc. Then when you are designing your cells the cell you make could be compared to each of these generic cells and then given a score. So it might say "this cell is 57% like the optimal muscle cell and 12% like the optimal bone cell etc". It would be labelled as whatever it was closest to and given a functionality proportional to that. So it might be 57% as effective at being a muscle as "the optimal muscle cell". Which is kind of boring for the experienced players who know how to quickly make their cells exactly how they should be. But is cool for noobs who get to play around with trying to make better muscle cells. The problem is people might get frustrated trying to optimise to these hidden variables. A larger issue is that whatever algorithm we make to test the cells will always boil down to this system. So say we make a magical, super complicated, neuron simulator which can tell you exactly how effective any arbitrary cell is at being a neuron. Well (assuming it's convex) there'll be some cell the simulator thinks is the best and everything else will be less good than this and so you're always working towards this "best cell" anyway. We could try making it proportional to the environment but then the complexity is getting out of control again. It's all pretty complicated, and that's the fun of making Thrive! 
|
|
The_Wayward_Admiral
Spacefaring
The_Real_Slim_Shady
Atrox drew this awesome image of the Keldori!
Posts: 1,011
|
Post by The_Wayward_Admiral on May 25, 2016 22:07:15 GMT
What I am about to say will not address every problem you pointed out (because this issue is a lot more complicated than I at first imagined), but I do have a slightly less nebulous definition of "the Algorithm" *epic music plays, flock of birds fly over head* In the scenario described in my post, the computer would have a rough idea of what was needed for a cell to perform a designated function. So it would require a muscle cell to have, say, twelve mitochondria, and any cell meeting this requirement can be considered muscle tissue (although the requirement in game would ideally be more complex). Under this system, cells could be considered multiple types of tissue if the player just stuffed every organelle possible into them, but hopefully energy constraints would encourage them to avoid such clutter. In this case the player could just sort of choose which of their cell types they want to include in pre-made organs, which would allow for tissue classification, but perhaps allow a bit of the microbe stage to flow into the multicellular stage. In the less formulaic scenario I put forth, in which there are no predefined tissues, then the computer would know what biological functions each cell was capable of carrying out, and the player would mold and sculpt the tissue arrangement so that the creature had the right distribution of cells that did the right job. This scenario would probably not allow for pre-made organs. Exempli gratia: painstakingly placing a column of "bone cells", which as far as the computer is concerned are just super strong, apatite reinforced cells, to form a structure that can support muscles. let me know if this was unclear/unhelpful/still not quite addressing the issue. edit: tjwhale the complexity of the mechanics is my favorite part of this game's development! edit 2: There are a lot of problems with my second paragraph, most notably that I'm not sure how the computer would figure out a creature's movement with player sculpted muscle clusters.
|
|
|
Post by Atrox on May 25, 2016 22:11:34 GMT
Mmmm... would it be too boring if we had pre-programmed generic cells, and you can designate which cell will ultimately become a certain type? From that point forwards, that cell is barred from getting certain organelles and sort of forc- nope.
Okay but let's say we designate Cell 1 to ultimately become a muscle cell right? What if auto-evo took over in this case and the game began making its own changes to the selected future muscle cells? Or is that what you meant by proportional to the environment?
|
|
|
Post by Atrox on May 25, 2016 22:14:44 GMT
Exempli gratia: painstakingly placing a column of "bone cells", which as far as the computer is concerned are just super strong, apatite reinforced cells, to form a structure that can support muscles. Support this. Side idea: What if we can save certain bone structures (when they get large enough) to be able to use later in the game (multicellular and aware). It'd be similar to tech object in the tech editor, whereas the cells themselves would be the function parts. EDIT: This addresses edit 2
|
|
|
Post by robertthebobert on May 26, 2016 0:42:30 GMT
This talk is kinda what I was going for, where the cell you create isn't necessarily labeled a specific type of cell, but is measured for it's capabilities and then used accordingly. I was also thinking that in the drag and drop style menu you can label each different kind of cell to whatever you need, so if you create what you want to be a neurone you can just label it as such. I feel this allows for more freedom while creating and adds more fun. Let me know what you guys think (I'm also amazed my post got so much attention  ).
|
|
|
Post by tjwhale on May 26, 2016 9:01:47 GMT
"but is measured for it's capabilities and then used accordingly" - got any clues about how this measuring might work?
Another issue is how fine detailed is the building process. Like say you are building a human are you really going to put in every bone and muscle and never ending? Like will you have to design teeth one by one and put nerves in each one and put gums on and put a tongue in etc. That seems too detailed and like too much work.
But if you get given "it's a mouth" and you can put it on your creature then don't we have the same problem again? Like you spent all this time designing custom neurons and custom muscles and custom bone segments and then it all jsut gets smushed in to a "generic mouth".
Or you can label whatever you want as "this is the mouth" but then we need a magic algorithm for determining how effective something is at being a mouth.
It looks like we might need a lot of magic algorithms!
|
|
|
Post by NickTheNick on May 27, 2016 9:08:54 GMT
I'm surprised I missed this topic, but now I'm here, and hopefully for the best because this is right up the alley of something I'm been thinking and planning out for a while now.
It's not that you select a group of the cells from the colony and you reassign them from default cells to, say, muscle cells, it's that you select a group of cells from the colony and assign them to their own group. Originally, when you first start multicellular gameplay, your colony is comprised of identical cells. A change to one cell is an identical change to all the cells. However, after your first generation of playing as a colony you can section off specific groups of the colony towards their own groupings. These groups evolve independently of each other. Also, a change to any cell of a group applies identically to all cells of that group. So if you edit a cell of type A to add a mitochondrion to it, this would give all cells of type A the same modification, but all cells of type B and type C and any other types you have would remain completely unchanged. This would form the basis of specialization, the fact that you can group cells together and evolve them as a whole.
It is in the player's best interest to form several types of cells, and specialize these different types. They don't necessarily have to fall into the categories of tissues we are familiar with, like bone, muscle, neural, etc, although they most probably will, since it is a proven approach.
This sort of gameplay would only have to be maintained until the 2D-3D transition, which would be probably after reaching 100 cells. Therefore we won't really have to worry about this feature causing micromanagement when colonies grow bigger, because by that point the mechanics will change anyways. When converting the colony to a 3D organism, this is where we'll need the mentioned "magical algorithms". The computer will have to be able to look at this colony of 100 cells and determine which cells are the most muscle-y, and classify those as muscle cells and convert that into muscle tissue when producing the 3D organism, and etc. for all the other cell types. A simple approach would be to just convert the colony into an organism of a single tissue and ignore an accurate conversion of the specializations from the colony into the organism.
If we are to take the approach of finding an algorithm to recognize and convert 2D colony cell types into 3D organism tissue, then we should identify specific things which can be used to identify cells. For example, to identify a cell as counting as digestive tissue, some important criteria I can think of are: 1. Plenty of lysosomes, possibly digestive agents 2. Lack of defensive mechanisms (thick membrane, cell wall, toxic agents, etc.) 3. High surface area to volume ratio
|
|
|
Post by tjwhale on May 27, 2016 10:07:58 GMT
When converting the colony to a 3D organism, this is where we'll need the mentioned "magical algorithms". The computer will have to be able to look at this colony of 100 cells and determine which cells are the most muscle-y, and classify those as muscle cells and convert that into muscle tissue when producing the 3D organism, and etc. for all the other cell types. So once this conversion happens you're back to a generic palette? Like there's muscle and bone and nerves etc which are the same for all players? Because if we're going with generic palette what's the purpose of the intermediate phase where you specialise the cells? I guess it smoothes out the transition but a shift to a generic palette is always a shock as it forgets the past completely.
|
|