[ismail]

The concept he discusses of “the edge” is very similar to “edge of chaos” from physics, but has been studied extensively in complexity sciences, specifically complex adaptive systems.

The theory proposes that all complex adaptive systems (CAS) naturally adapt to a state at the “edge of chaos” which is a transition zone between order(stability) and disorder.

The theory proposes this is the zone where maximal learning/innovation/creativity in social systems occur.

We studied complex adaptive systems in 2019, at the time I changed my LinkedIn tag line to : “learning at the edge of chaos” , still have not changed it since then.

https://en.m.wikipedia.org/wiki/Edge_of_chaos

[Jerrrrrrry]

all systems that have accumulated complexity did it in an effort to create resiliency to survive/reproduce/continue to exist, which is innately a method of accumulate self-supervised 'learning', when left to dwell to its own emergent methods.

this applies to all systems, from biospheres to food-chains to cells to human evolution.

resiliency is needed for systems to be less fragile against chaotic perturbations, as the most 'complex' (sub) systems are the most impacted by any change without it.

complex systems would fail catastrophically instantly if its resilient sub-systems weren't able to postpone it.

resiliency is the ability to respond to varied input, to face dis-order.

The universe, uncaring, is a dis-orderly increase in entropy. It accumulates, and averages to eventually to act as a sieve, a selective pressure, an edge....a particularly varied input.

Anything that would pressure the system - such as an environment change or competition against itself for a resource constraint - and this selective pressure culls the weakest variations of the system from the pool. Those variations that had the least effective resiliency features, now gone, are quickly replaced, and the system continues to exist.

All complex systems in adversarial conditions must then incentivize resiliency, and the generalized property of being self-reliant; adaptive to variation in input.

This incentive/reward is essentially an iota of agency, a flash of an of objective goal.

intelligence is the ability to reach a goal given varied input states.

The ability to 'learn' is really just how to compile ways to reach a goal, inferring relations between the solutions, then internalizing that inference to later better increase its ability to generalize / respond to varying input.

learning is _only_ possible at the edge of chaos

[Nevermark]

> learning is _only_ possible at the edge of chaos

Which is what makes math so interesting. The constant stream of finding predictable islands in the unpredictable, and then unpredictable islands (hard problems) in the predictable (seemingly simple easily defined systems).

Math is chaos.

I suppose as we get smarter, and our understanding of the world gets more sophisticated, our survival/growth progress becomes more and more a math exercise.

[coffeecantcode]

It has also been posed that the hemispheres of our brain operate at this edge of chaos with one hemisphere aligned with the novel and chaotic, and the other hemisphere aligned with routine and order. Then in this theory as we sleep/dream our brains take the novel and chaotic things we experience and transfer them to the routine and orderly side of the brain. Or something along those lines.

I’m no psychologist but I find the existence of chaos and order in the physical word to be really interesting. It seems like something we would make up as a species, but really there is proof for it in the universe around us.

[jumping_frog]

Brain Criticality - Optimizing Neural Computations

https://youtu.be/vwLb3XlPCB4

[taneq]

The boundary between predictable and chaotic is always there most interesting place to be.

[ls65536]

For some reason this also made me think of the boundary of the Mandelbrot set...that place where all the interesting structure reveals itself.

[taneq]

Yep, you're getting it. That same universal principle where the interesting things all happen in that boundary between boring and random (ie. trivially-predictable, and not-predictable-under-any-conditions).

If you haven't already, I heartily recommend James Gleick's book "Chaos" (https://en.wikipedia.org/wiki/Chaos:_Making_a_New_Science) which is, to me, the introduction to both self similarity and sensitive dependence on initial conditions.

[Jerrrrrrry]

https://en.wikipedia.org/wiki/Rule_30

[taneq]

Wow, that's really different to rule 34. :P