Interactive Evolution Simulator 

imagine being able to see 3 miles in every direction this post was made by the primer blob gang

Thank you so much! I'm a biology teacher that just discovered both the Primer video and your browser simulator. When I first saw the primer video I was actually thinking, "I wonder if I can get access to this simulator, and I hope it's in a browser so that I can let my students play with it." Then of course, I found your work. I've been playing with it quite a lot. I've been struggling to try and create a functioning mini ecosystem. I was hoping to make blue blobs act like herbivores and orange as carnivores, and maybe even get the pink in there somewhere, but have not had any success creating two stable populations. It occurred to me after several attempts that the reason is the competitive exclusion principle which says that two species can't coexist if they occupy the same niche. Even though I was seeking to make one blob type act as a carnivore, in reality they are all omnivores with all the same possible resources. So that brings me to a request. Do you think that you could program in two more food types with their own generation regrowth rates, and allow us to set for each blob species which food types they can eat? For example, maybe we could set up two food types, one that only blue blobs could eat and another that both blue and pink blobs could eat. The second could have a much smaller regrowth rate. Then if we could set it so that the orange blobs could only eat blue and pink blob, not the regrowing food pellets. If we could make such a situation then we could see the arms races between predators and prey under different situations. We would also see the tension between pink evolving specifically to it's particular resource while blue evolves to collect as much food of any type as possible. And lastly, we would hopefully be able to construct a system with multiple stable populations. I realize that what I'm asking for might be quite complicated and that this is just a side project, so no hard feelings if you can't, but I just think that that expansion would add some valuable capabilities to your simulator.

That bit with how less mutation in size actually makes evolution faster was kind of unintuitive, yet you explained it very clearly and the visualization was pretty easy to understand! Amazing simulation, I'll definitely play around with it when I have the time!

Me: aww the blobs are so cute Minutelabs : so let we make them cannibals

Makes me wonder how many tweaks the model would need to start exhibiting speciation into predator and prey blobs. Sexual reproduction would probably help a lot to make this happen. Making reproductive success depended on mutual similarity.

You should introduce stamina. This could allow for really strong species to evolve because through more stamina, they are able to use their other traits more. Keep up the great work! Also, maybe there should be an option where children and parents don't eat each other. Grandparents and grandchildren can each each other, but one generation is always liked.

The age of the blobs steadies at 2 generations and humans (and a lot of animals) live also for 2 generations, amazing

Wow great job! I've just stumbled across this video and really love those simulations of Primer. I don't get it why you only get this little attention? However, amazing work and keep going!

Can you please add a few features that I believe would be amazing for this Simulation (In no particular order) I love this tool: 1.) The ability to make some species strict Carnivores (aka blob eaters) and some strict Herbivores (Pellet Eaters) 2.) Make the value of food/survivability and Reproduction adjustable 3.) The ability to create different types of food so we can have different size and quality of food generated I feel that these 3 additions to the Simulation would really flesh it out...

the size should have a bell curve effect on the speed, as very large animals can usually be outrun by "normal" ones due to gravity

Linking speed to size helps, but I still think that this simulation undervalues carnivorous behavior relative to nature. Sure, you have more food options, but no matter how big or fast you are, catching a running target will still be harder than getting the food pellets. Because of this, there's really no benefit to going after other blobs. I think a good way to improve this simulation would be to make eating blobs better than eating food pellets- eating meat gives more energy than plants after all, and the blobs are bigger than the pellets. If eating another blob was worth two food, ensuring reproduction on their own, I think that would be more realistic. Or if you'd rather, maybe eating a blob is worth one food for two days- carnivores often live for quite a while off of one kill in nature after all.

wait there was minutes lab too? totally forgot

You deserve more credit for all that programming, coding and data analysis.

I don't agree with the size-speed thing, some species are faster because of their small size, specially if they are closely related

Amazing, the speed of evolution is based on selection pressures in the environment. Taking a counter intuitive graph and explaining it beautifully. Simulations can really help us gleam the mechanisms of the natural world. You’re fulfilling the mission, brother.

very good work! really amazing, it also runs smoothly. will you be adding other features like an option to change the quantity of food or even to intoduce a new specie?

I'd be interested in messing around with the source code, when I have time. Might play around with things like adding a good bit of poisoned food, that cost energy next round, but allowing varying degrees of immunity/resistance to it.

Cool! I thought the drastic change due to the size mutation rate was really cool!

2:27 i LOVE how this rule just once again somewhat creates agario :D

I had a lot of fun with this and even found a really strange effect of have 1000 food and 0.1 change in all stats i got to different strategist for food those who rush and get the food close to them and have almost no sight and slow blobs that can see crazy far and go for the food in the middle.

The meta is extremely small but fast creatures (10% size, 125% speed compared to the default), since they use speed to get food before big creatures, and don't get eaten because they're fast enough to run away from big creatures. The exception is when there's tons of food, then the small but fast creatures get spawnkilled by big creatures.

I never usually comment on RU-vid but you have made my day. Thanks for the simulator. I loved thr primer videos was interested in it and was planning to code one myself.

differentiating the colours so people can see the difference between size and sense range would be really helpful

Dude I loved your idea!

Good job man!

I found it interesting how the sense range wouldn't go down even if it was set to crazy high, to the point at which it was totally unnecessary. Could be seen as a vestigial structure. Also interesting to play around with exponential growth starting from just one blob, and trying to make bottle necks and seeing how that affected stuff. Really well made simulation!!!

Hi, is there any chance we could get a simplified version of this video? I am a Biology teacher and am going to incorporate your simulator into my class. A short video on how to use the simulator would be extremely helpful. Maybe something 3-5 minutes in length and just explain how to use the different settings and what they do. Just enough information so kids can understand how to use the simulator and draw information from it. I'm planning on having them play around with the settings and then use data to develop their own conclusions. Anyway, thank you for all the hard work you've put into this! I've been looking for something like this for a long time now! Awesome work!

You are teh best I was waiting for some kind of evolution game and you have nailed it.

Super cool thanks

Wow I love the simulation. I have a suggestion can you please add a indicator for size. bigger blobs are brighter green and the small blobs are dark green.

Oh my god, i can't believe he is back!

Oh my god, i can't believe he is back!

It would be so cool if someone made a game like "Species" with better and more relatable looking creatures. The animals in that game look pretty uncanny. Hopefully there will be more simulators like this coming out for a broader audience.

Hope you get hooked on the topic of evolution & simulaton :) I started working on something similar 10 years ago during my bachelor thesis and since then I could never let go of this topic again. There is so much to discover!

Really didnt expect a video from you, this just made my day. :)

This is sooo awesome I didn't know you guys knew about the primer channel I love that channel sooo much

Ants have a higher speed value but we are faster lol, I like that.

So cool, was searching for this for a long time... Thanks a lot dude.. Liked, subscribed, now sharing (And commenting :3)

From playing with it a bit, it seems that the population always converges rapidly to nearly match the amount of food. This makes sense, since if the blobs were perfect they would be able to maintain exactly 1 blob per piece of food. If we really maintained population = food, then the average age would be exactly 2 if every blob either reproduced (and ate 2 food) or died (and ate no food) in every generation. On the other hand, if every blob just maintained its own life and ate one food, then the age would be unbounded. In practice, the population becomes dominated by blobs that reproduce, so the average ends up pretty close to 2. Probably the fact that not all food gets used is enough to counter the effect of any blobs "stuck" in a state in which they just keep eating one food and going home. (Also, I'm not sure if this survival strategy even makes sense in this model.) I tried to make a world in which finding exactly one piece of food was the best strategy, but I mostly just ended up with extinct blobs :-/. It would be nice if we could see the number of new blobs and dead blobs produced each turn to get a better feel for this.

Hey, I’m so happy I’ve found your website ! Thank you so much, since I have watched Primer videos a few years ago I wanted to try but I haven’t enough skills to do it so thank you ! I am quite late, but I think I have the solution for the mean age. First, I think that it was because of selection: children and grandchildren and etc were better adapted to the environment, so to prove it I stopped the fact that the blobs evolved between a generation and an other (variance = 0). But the average age was still around 2. So, the selection was not taking any part of that constant mean age. Actually it’s just maths, at each generation every individual will have approximately the same chance to die (because in 2 or 3 generation the blobs can’t be that different, they have practically the same advantages /disadvantages). So when you reach kind of steady state (the size of the population change just a little around a constant value), 25 to 45% of the blob will always die (I take this number by observing the simulations on the website), whatever is their age, so in the population they would be always quite the same amount of blob with 0 generation, 1 generation, 2 generations… And so the mean stay stable around 2. I’ve made kind of a simulation in python to see if it works (and it is !). I considered that the size of the population was constant, the number of deaths is equal to the number of births. And at each generation, around 1/3 of the population in renewed by new individuals with an age of 0. At the beginning each blob (which is a number in my list ‘pop’) have the value 0 and each generation I had it one unit (if it survive). The function returns the list of the mean age at each generation. And then you can represent the graph and you will see that the mean age fluctuated around 2 but rarely goes above 3 or below 1.5. As you may have seen I’m not a native English speaker, I am sorry for my language and I hope you have understood everything ! (‘effectif’ for size of the population, ‘age_moyen’ for mean age, ‘pop’ for population, ‘nouveaux’ for new blobs, nb_gen for number of generations) from random import * from math import floor import matplotlib.pyplot as plt import numpy as np def primer(effectif,nb_gen): liste_effectif=[k for k in range(effectif)] pop=[0 for k in range(effectif)] age_moyen=[] for gen in range(nb_gen): pop=[ind+1 for ind in pop] nouveaux=sample(liste_effectif,floor(effectif*np.random.normal(0.33,0.05))) for nouveau in nouveaux: pop[nouveau]=0 age_moyen.append(sum(pop)/effectif) return(age_moyen) nb_generations=150 effectif=100 generation=[k for k in range(nb_generations)] plt.plot(generation,primer(effectif,nb_generations),'r') plt.grid() plt.ylim(0,5) plt.show()

I think what you have done is great! I wish I had the time to mess around with the code. I find it strange that every blob has around the same size and I would have thought that some blobs would have specialized into eating other blobs. If I had to explain it, I would say that it can be explained by reproductive reasons. It seems a bit simplistic that you have to eat two pieces of food to reproduce. I would rather say that each piece of food gives you some energy and you have to eat twice your energy in a turn to be able to reproduce. But when you eat another blob, you are eating his energy, so there is clearly an incentive to specialize on eating blobs instead of food.

Wow. I am amazed

With the speed thing, smaller isn’t always slower. Small prey animals are wicked fast!!

Fascinating work! Keep it up. I wish I could meet you someday.

Wow awesome work!

This is incredible! I'm amazed!

The average age being stable at 2 means mortality is at 50%/gen. And the one thing you don't have sliders for is the value of each piece of food and the maximum reproductions per generation.

So cool man I always wonder to create a simulation with my own configurations when I was watching your videos thanks to do this site

it makes sense to make each generation smaller if the mutation is smaller because even if you as small as possible other blobs would need atleast 20% of your size but till that happens the bigger ones would die becouse they were not capabale of finding enough food

as others have pointed out, average lifespan is ~2 because coinflips with offspring. following from that, average age across 100s of generations can be pushed above 5.0 by limiting reproduction capabilities-- that is, putting trait variance to 0 and setting max energy such that blobs only have time to collect a single food before returning.

this is awesome! You should turn this into an app for Ios/android

Just turn this into a full blown game

Idea: food appearing throughout the day in addition to the preset food at the beginning.

It seems like, no matter the starting conditions, the creatures always tend to evolve higher sense range and speed while lowering their size

2 parameters that could be added: Greed, percentage chance that blobs will stay looking for another piece of food. Reproduction, the chance that an extra piece of food will produce another offspring. Lower then 100% will be reproduce or not, higher than 100% will create a chance for twins. Twins should probably each get a random stat change and a negative multiplier for all stats, because using less energy to produce two offspring should make the offspring weaker over all

A principal component analysis would be nice to visualize the traits on each generation

Thank you sooo much!!!!

In 14G my blobs ganged up to find food in different groups.

I would love a version of this where speed and energy consumption vary. We never move at top speed the whole time

I'm curious about the effect of field of view vs view distance with relation to bigger and smaller blobs,as in nature many predators have a narrower field of view (front mounted eyes) and prey have a wider field of view (side mounted eyes).

Amazing! Great job creating this simulator!

I really like the simulator but I would like it even more if you could adjust more settings to simulate custom diseases, markets and emotions/personality types like in some of Primer’s other videos. I don’t know how complex this would be to code but I’ll assume your skills are off the charts, hope you can do it. But even if you don’t, thanks for the simulator anyway.

Me: Looks at date Also me: Oh, too boring quarantaine

there should be a feature with which you can spawn blobs between days

i personally think that the reproduction system should work like that: The offspring inherits some traits form their parents and has a change of random mutation this would make only the strongest survive

Why don't smaller blobs run away from larger blobs to keep from getting eaten? Or was that shown?

i wonder if species separation would occur naturally. for instance, many small blobs that have lots of energy and can reproduce quickly, and a few large blobs that can sustain themselves off the plentiful small blobs.

If more games were like this

This is beautiful

Prime really at here spreading his awesome content

Big creatures needs more food in the real world and I can't figure it out, if bigger mammals are less or more energy efficient (Talking about the real world only)

The Bibites game has similar mechanics with the eyesight range, but it is about evolving an AI for the creatures through simulations.

This is a good starting point, but obviously I feel like this could benefit from being able to simulate a few more physical variables.

I wish there was an identification number for each blob and you could name them. The first generation blobs would be assigned integers between 1 and the total number of blobs, inclusive. So if there were 5 blobs they would be 1, 2, 3, 4 and 5. Offspring would be x.1, and survivors x.0, where x is the identification number of the blob. So blob 3 would be 3.0 in the 2nd generation. If it reproduced at the end of the 1st generation, the offspring would be 3.1. If blob 3.0 reproduced at the end of the 2nd generation, the offspring would be 3.0.1. And if blob 3.1 reproduced at the end of the second generation, its offspring would be 3.1.1. A 0 at the end means that the blob survived from the previous generation, while any other number (always 1 unless it's the 1st generation) means that it is new.

You should add more to this

the thing I would love to see are the different specieis,cause I don't like canabilasim:it would make it if a blob is too similar to another,then they can't eat the other(all stats have to be in an amount similar enough (custamisable)) Edit:you should also add carnivore,herbivore,and omnivore as an option in the settings

Thank you so much!

I'd imagine the average age being 2 is just a product of the likelihood of the offspring being unviable offset to the likelihood that a blob will likely always be successful.

Rust! Yay!

I thing that i found using this simulator is when theres lots of food to go around the population expands when it gets close to its maximum they all just get bigger and start eating each other then they get bigger and bigger trying to eat each other and the population gets smaller and smaller with more and more food being wasted. It would be cool if you experimented with sme different mechanics or let the user do that such as what would happen if food not eaten was left behind for the next genration or the plants reproduced to simulating more of an ecosystem, i know its meant to simulate a species but the same things happens every time, i think it would be cooler simulating different species giving opportunity for different things to happen. Going back to creatures just looking out for thier own self intrest which is not optimal for the species and ends in blobs dying unnessercarily and food being wasted a mechanic to help smaller blobs or species wroking together against an ecosystem vs eating each other would be pretty cool i think and the relationship between size and speed dosent make a whole lot of sense, in the same way that a human is faster than an ant an elephant is slower than a human and there is no limiter to size being advantagous for individuals it makes them faster and more likely to eat while not being eaten except for energy which causes lots of them to die. one thing i wanted to mention about energy is that i would have assumed it would be based of food but idk i was thinking a way to balance big blobs would be to have smaller blobs require less food this would work tying in with energy so small blobs dont have to eat as much as big blobs i think this makes sense the energy per generation and 1 food to survive 2 to reproduce is pretty basic and not very robust, i think it could be improved I would love to see some cool changes like this to the evolution simulator or for you to give some thought about it, just some suggestions im open to talking about them if anyone actually bothered to read my essay lol

I have a suggestion, can you make it so bigger blobs require more food to survive? I know this is a direct nerf to size, but in reality, being big requires more energy as you said, but it also requires more food to maintain that energy.

World is all black on Chrome in Mac, but renders properly in Safari. Interesting.

3:38 you should allow us to manipulate the equations as well if it's exactly like the primer video, we're not doing something new...

For 2 generation factor: It is matching reproductivity rate. 1st phase - ramp-up. Amount of creatures increases to the food limit amount. 2nd phase - competition. Blob double up each time. Thus each second one shall be out of food this round each time. Why not 2 generations (0, 1), but 3 generations (0, 1, 2). I think this is because of one stash of the food is basically stored within starting creatures. There are huge peaks though, so one can live long. You shall put on graph mean, not avarage. The mean will differs.

the constant age is maybe related to the rythm of the coninuous evolution... the blobls older than aprox 2 generations might be out-of-date, unadapted compared with the new ones? Due to this they have more chances to die then. In folowing versions you could consider a variability (following Poisson's distribuition maybe) in the contribution on the next generation! I gess this would accelerate evolution by natural selection...

You should make this a video game

My guess is that due to the high rate of mutation, new blobs have evolutionary advantage over older blobs and therefor no one blob lives very long.

Funny. I started something like this in Unity. Never got it to stabilyze. I even had some of the same ideas about size, speed, energy consumption etc.

regarding the age, what is the probability of getting tails 3 times in a row when flipping a coin ? that's your answer

would be interesting if we could change the energy cost function or how speed, sight is calculated or how large you must be able to eat smaller blobs and perhaps make starting energy be dependent on size

Amazing

nice to know im not the only one that learns like this

Ever thought about that a common mechanic is that bigger creatures eat smaller ones but the reality isn't always this way also size and speed corelates not that 1 dimensional a elephant isn't faster as a leopard cause wouldn't need to a Antilope is bigger as a leopard but can eaten by it... I know a simulation of reality is to complex but maybe a mutation to carni/omnivore and something attributes like strength and durability could already fix some of the mentioned issues and swing the simulation already in some different ways

Amazing content! Thank you

Ever try to catch a rabit?

Natural selection

wrong. wrong? When you re small you need less energy to move fast, you re accelerating a smaller mass. when you re big you need more energy to move the same speed? Right? I know I might be missing fwe things in these statements but it's a quick comment(observation). I just strongly feel he is incorect in some of the logic behind what he said about speed and size. Smart, patient people pls colaborate!

If you like learning by being entertained matpat got you

Low variance in size means it's rare for a blob to achieve the 20% size advantage it needs to become a predator, so there's no benefit to gaining size. I wonder, if you run the simulation long enough how long will it take a predator to emerge by chance from a population of similarly sized small blobs. Will it then take over like an invasive species?

TANSTAAFL!