Coevolution creates living complexity
A complex system that works is invariably found to have evolved from a simple system that worked. A complex system designed from scratch never works and cannot be patched up to make it work. You have to start over with a working simple system. (Gall’s Law)
Complex systems can evolve from simple systems only if there are stable intermediate forms. (Donella Meadows, 2008. Thinking in Systems)
Putting the quotes together, he coins the concept of a Gall-Meadows ladder.
Success needs what I think of as a Gall-Meadows ladder: a mechanism for jumping from one stable system to another (like the wall climbing mechanic in the first Ninja Gaiden game) in order to evolve your project…
The Gall-Meadows ladder suggests that to get from a simple working system to a complex working system you have to find a sequence of working systems that increase the level of complexity.
This ladder has a cyclicality to it. Simple systems evolve into complex systems, which generates specialization, producing simple building blocks for a new system one layer up. The result is an upward spiral of evolutionary complexity. A Gall-Meadows spiral staircase?
Life wants to spiral upwards
So, simple single-celled organisms join into colonies of cells, who specialize, and become building blocks for simple multicellular organisms.
Why and how does this happen? It seems that specialization and cooperation have economies of scale.
It is shown that an organism with a variety of differentiated cell types and a complex pattern emerges through cell-cell interactions even without postulating any elaborate control mechanism. Such an organism is found to maintain a larger growth speed as an ensemble, by achieving a cooperative use of resources, than do simple homogeneous cells, which behave “selfishly.” This suggests that the emergence of multicellular organisms with complex organization is a necessity in evolution.
(Furusawa, Kaneko, 2000. Complex Organization in Multicellularity as a Necessity in Evolution.)
The major evolutionary transitions, including those from prokaryotes to eukaryotes and from free-living cells to multicellularity, all increase the scale over which cooperative interactions dominate competitive interactions.
(Fields, Levin, 2018. Are planaria individuals?)
Life wants to spiral upwards. But how does this process start? What are the initial conditions, the primordial ooze from which upward-spiraling complexity emerges? And if we wanted to provoke the emergence of a new living ecosystem—a rainforest, or an economy, or a new web—where would we start? What’s in the seed? I have a hypothesis…
Coevolution creates upward-spiraling complexity
So, the path of evolution is always through the adjacent possible, right? This sounds a bit like hill-climbing. Follow the gradient until you get to the top.
But hill climbing will get you stuck at the top of the nearest hill. In the language of ML, we call this getting stuck in a local maxima. It doesn’t matter if there’s a bigger mountain next door, because you’ve already summited your hill, and from the top of that hill everything else is down.
But evolution isn’t just hill climbing. We can see this in practice, because living ecosystems seem to avoid getting trapped in maxima.
Living systems are perhaps best characterized as systems that dynamically avoid attractors. (Chris Langton)
Life evolves in open-ended directions, refusing to get stuck in any one place, continually finding new hills to climb.
How is this? One major difference is that evolutionary landscapes aren’t static. Why? Coevolution!
Ever since there were two organisms, life has been a matter of coevolution.
The moment two evolving organisms touch, they begin to coevolve. If you were to graph this out as a fitness landscape, you would see a complex landscape of peaks and valleys. This is because the organisms are interacting, and more interactions between variables, the more rugged a fitness landscape will be.
More than that, if you played the fitness landscape over time, you would see it is dancing.
Each organism influences the fitness of the other, and each adapts in response, provoking more adaptation, in a recursive dance. A Red Queen Dance.
As they dance, asymmetries emerge, and from asymmetry emerges specialization and cooperation. Whence predator, prey, pal, parasite, pollinator, niche-creator, producer, consumer, scavenger, decomposer, all tangled together.
At each step of this dance, the coevolving organisms collect new strategies, recording them in DNA. It’s as if they are collecting stepping stones in possibility space, and each stepping stone expands their repertoire of responses.
And this is another way in which evolution is not just hill-climbing. The path of evolution is always through the adjacent possible, but each step into the adjacent possible expands the adjacent possible.
Adjacent, i.e., nearby, possibilities constantly emerge in a multitude of settings for a multitude of entities. When these possibilities are explored, yet new possibilities emerge. (Björneborn, 2020. Adjacent Possible.)
So if we want to kick off a Gall-Meadows spiral, we need to create the initial conditions for coevolution.
Minimum viable coevolution
Coevolution takes two. Yet an ecosystem of two is likely to settle into a stable boring duopoly. More would be good! But how many more?
ARPANET started with four. So that’s one existence-proof.
Can we go lower? My sense is that three might be a minimum for kickstarting open-ended coevolution.