I made a video to discuss how the current auto-evo works, as well as my personal experiments with the technology. If you want to understand the system better (and it’s okay if you don’t) give it a watch.
If you read my writeup on auto-evo before, this video reflects a few updates I’ve made to the algorithm.
Very nice video. I had to turn up my volume a bit compared to other Youtube videos I usually watch. So if you make more videos you could make them a tiny bit louder.
Very nice explanation on the miches, I think I finally understood the point about them. I am kind of worried though how much maintenance it would need because each potential niche will need to be separately programmed and the “evolution guidance system” needs to be extended for it.
Why is sunlight above water, and not the other way around? Why arent they next to each other? I guess it doesnt matter, as long as the map doesn’t branch out.
Lets consider adding a niche map to Thrive. How would it be automaticly created for a given environment? To run away from a predator, you can[1] either dig underground, or climb a tree, which becomes a new niche, but trees are also organisms that evolve. The niche map would heavily depend on the organisms already present.
If it isn’t automatic, it would require maintanence as hh said.
among other things ↩︎
They can’t be next to each other because that makes them an either or case: some plants that don’t need sunlight to survive, while some plants don’t need water to survive. The vertical ordering is a little more subjective (but it does make a difference in code), but I think my mindset was that since I was limited this model universe to strictly plants, so photosynthesis was the core of all survival. I think there’s a good case that water retention could in fact be on top.
In my implementation on my Thrive fork, I do generate them programatically: https://github.com/adQuid/Thrive/blob/master/src/auto-evo/simulation/selection_pressure/MicheFactory.cs#L20
As I recall it also iterates over existing species to make new miches for predation then adds a species-specific selection pressure to avoid predation to every species that ends up with a filled predation miche. I wouldn’t think different survival strategies like digging underground or climbing a tree would constitute distinct miches, since if there’s already a species in town that can climb trees to avoid the predator really effectively, you’re not making a space for yourself in the ecosystem because you can dig underground half as fast as they can climb (unless the environment has some big treeless areas where the other strategy is the only option, in which case maybe that opens up)