Chaos, collapse and complex systems in volatile times

This post is the first in a series looking at how we can apply systems thinking to the omnicrisis

I haven’t written for ages. Not for lack of ideas but mostly because I can’t get them to sit still long enough in my head for me to write them down. There is so much I want to talk about and so many places to start this account but wherever I begin, I find myself reversing to explain what happened before or forwards to explain what happens next. Then the whole undertaking overwhelms me.

But today I’m going to persevere!

I must start somewhere and since I’m writing about how to respond to chaotic situations in my other world (the business bit), I may as well start with collapse. So here is my longish introduction to system collapse for people unfamiliar with systems thinking.

Imagine a pile of Lego. Now give this Lego to a creative young brain. This creative human will build a panoply of structures by pushing together small pieces to create larger pieces. Some of these constructed objects will be displayed, but within a short period of geological time, they’ll mostly be collapsed down into bits and re-sorted into something new. Much of the Lego will be surplus to requirements. It will find its way into tiny spaces, clogging up metal runners, merging with organic matter and generally making a nuisance of itself. This is my son’s bedroom. A bedlam of disordered Lego pieces spilling from every surface and recombining into his crazy creations.

Lego is a good metaphor for a complex system. The pile of Lego also represents a non-adaptive system. The pieces of Lego cannot self-organise into new structures while my son sleeps. The structures my son creates are never bigger than the sum of their parts. But they do have an energy source. This is the easily manipulated grandparent who keeps feeding this mess with more packages of new Lego.

As the complexity of my son’s creations has expanded, he has reached a limiting constraint. To build bigger, structurally more complex objects, he must move beyond trial and error learning and apply some embedded knowledge. To construct such objects de novo, from the chemical soup of pieces that lie scattered in his room, has become an overwhelming task. He commences ambitious projects - the White House, the Colosseum, a Boeing aircraft. Then collapses them in frustration when he discovers that complex systems exhibit interesting properties:

• path dependency

• sensitivity to initial conditions

These mean that if he can’t get the design exactly right at the beginning, the whole structure will collapse halfway through the build process. He could, of course, build some modular supporting structures and use these to underpin his designs. He has grasped some basic laws about weight distribution and structural strength of different shapes. But it seems a lot to ask him to divine 5,000 years of rules in engineering and design, in his untidy little bedroom. That’s what knowledge is for and knowledge is stored in heuristics, rules, textbooks, processes and norms. All things that disinterest my son.

In the Lego HQ, someone worked out these limiting constraints a while back. They recognised that most creators would hit them early in their trial and error career and would seek out order and rules. Lego began to design ‘whole’ complex structures that could be copied. Instead of fostering trial and error exploration, they gave creators the chance to build something big and complex - if they followed a design. This was enormously satisfying for children who wanted to make big things, and a huge relief for parents, sick to their teeth of impaling vacuum cleaners in search of lodged parts.

This Lego metaphor introduces some characteristics of complex systems:

• they are the sum of their parts

• they can be recombined in a vast number of ways to create new objects

• the objects can be collapsed back down

• to make complex objects requires top down design or,

• bottom up modularity based on rules, standards, norms and knowledge

Complex adaptive systems are like a stack of Lego that builds itself. The system still needs energy to survive (either embedded in solid parts or captured through clever co-operative processes, from the sun) but as it grows, the parts will evolve into new shapes. A complex adaptive system can always make new structures that are greater than the sum of their parts (a process known as emergence).

If you can imagine a pile of Lego that keeps re-ordering itself into new, greater than sum of parts shapes it should also be intuitive that more energy (parcels of Lego) will have to be fed into the system. As the Lego outgrows its niche (my son’s bedroom), it must adapt to the new environment. The more the Lego has been structured into big, rigid, complex objects, the more likely they will snap as the Lego tries to escape its niche.

Biological systems are very good at moving from one niche to a neighbouring niche. They form structures that are not so rigid that they can’t adapt to new, similar-ish, environments. They also consume their energy carefully, ensuring that big things collapse down into reusable energy captured and re-used by the small, modular parts. This enhances adaptability, allowing parts to recombine into new shapes. I have borrowed this picture of a trophic pyramid from Wikipedia to illustrate this point (copyright: Swiggity.Swag.YOLO.Bro):

In this model, some energy is lost as heat dissipates, but most of it is recycled into new growth. The system is balanced and can sustain big shocks because of its adaptive qualities. Instead of being top heavy with rigid structures that can’t easily absorb change, it is bottom heavy with new growth coming from small, modular elements. But even this system can’t absorb any shock. If an asteroid hits, the big, rigid things will be wiped out very quickly and the small, modular things will take generations to re-organise themselves back into big, complex things (especially if energy supplies have been sharply curtailed by a dust cloud).

I have now rushed into the collapse scenario without explaining how we get there. Since I’m not a biologist, I’m going to try and explain evolution through the lens of the BBC television series: The Traitors. In this show, strangers are holed up in a baronial Scottish castle, and seek to eliminate one another in pursuit of a financial prize. Each episode consists of a series of challenges with prizes that are shared collectively (a big pot of money) and individually (protection from being eliminated by ‘traitors’ in the group’s midst).

In one episode, the contestants are gathered outside a building and told that within the building, they will find 4 doors. 1 of the four doors leads to another room (again with 4 doors). If the contestants pick the wrong door, they will be eliminated. If they pick the correct door, they can pass through to the next room. The building contains 4 rooms. Successfully navigating all 4 rooms leads to ‘escape’ and a cash prize.

Imagine you are one of the contestants. You are confronted with 4 doors, knowing that one will lead you to a room closer to your prize while the others will eliminate you. Our gameshow participants use trial and error learning to identify which doors will open up a path that takes them closer to winning the prize. As each door is tried, the contestants call out to the people behind them:

“the harlequin door with the white lines is the one to choose!”

This is knowledge and it is transferred to other participants (who share a common goal to win prize money for the group). As each door is unlocked and the knowledge is shared, the groups create a safe path for others to follow. If you can imagine this as a multi-dimensional space with lots of doors, lots of trial and error learning and lots of paths (that can be encoded as knowledge, rules, DNA, norms etc;) you will have started to understand the theory of evolution.

While the participants in the gameshow have the pleasure of choosing their door, evolution is blind. Different doors are explored randomly through mutations, recombination of genes, transcription factors etc; If an organism picks the correct door, it will achieve a higher level of ‘evolutionary fitness’. If it picks the wrong door, it may find itself stranded at a lower level of fitness and eliminated.

You will also see why my son’s attempts to build complex models from scattered parts is too big a task. He needs a path. He needs some rules. He needs some knowledge. Once he has acquired these, he will find himself further along the path (able to construct more complex objects).

Engineers represent this space as a lattice. It may describe a set of decision choices in a computer algorithm or a phase space in physics. For biologists, the lattice conjures up a set of random choices that will be explored by evolving species. (Illustration of lattice courtesy of Wikipedia: https://commons.wikimedia.org/wiki/User:El.vegaro):

This lattice may not look like a building with 4 rooms and 4 doors but it performs a similar role in this explanation. As a non-physicist, I find it easier to understand rooms with doors than multi-dimensional spaces and ‘random walks’. But either way, the participants in the gameshow are acting like self-organising parts in a complex adaptive system. They are trying out doors, finding paths and getting closer to a desired outcome. Once an adaptive path has been found, knowledge of the path can be stored in rules, norms, codes, DNA etc;

Over time, these paths can become pretty stable. Ancient footpaths likely started as animal routes leading to resources. As the path deepens, other animals will follow. Once a cohort of graduates have earned more money doing an MBA, more of their species will follow.

For established, ordered pathways, all other information (e.g. about the other 3 doors) becomes surplus. It becomes ‘noise’ rather than ‘signal’. Trial and error learning is no longer needed. Herd instincts and conformity are a more reliable strategy for obtaining the prize. Biological systems and human engineers may be tempted to streamline this discovery by eliminating all the wasteful information (the noise) pertaining to the other doors. Indeed, they will often write a set of rules (or DNA code) to guide participants safely to the higher level of fitness.

But (good) designers and self-organising natural systems also recognise that eliminating all surplus information (variation) from a system, makes it rigid. Just as turning a pile of Lego into one huge structure will make it liable to snap when it needs to migrate to a new niche, eliminating all the informational noise in a complex system, will make it harder for participants to adapt to change.

Engineers build in redundancy and safety valves for their systems to help them navigate unusual events. Complex adaptive systems do the same - but to a much greater degree. DNA will encode information that may never be needed. It will retain memories of pathways long since discarded just in case it comes in useful. It will duplicate parts of the system (lungs, kidney, genes) in case they suffer malfunction. And even if the participants in a complex adaptive system settle into a natural, stable rhythm of life, each participant will retain just enough knowledge to respond to novel challenges if required.

Now imagine that our clear, ordered pathway through the four doors is stirred by a small perturbation. Although the lattice / doorways wobble, the rules of the system are still intact:

“follow the signs to reach the prize”.

The system continues unchanged. But now imagine something HUGE hits the structure (the multi-dimensional lattice shape). In this case, the doorways are flattened. The participants are flung to the ground and cannot differentiate up from down.

Something odd happens - because all the doors have been flattened, participants begin to run in all directions. Some will go the wrong way and be instantly eliminated … but others will find a new path and run along it to the prize.

Instead of finding the correct pathway gradually, through trial and error learning, these participants will arrive at a higher level of fitness (win the prize) without modularity (bit by bit adaptation). This is a little like my house being hit by a thunderbolt and a fully constructed Lego DeathStar tumbling out of the ruins.

If this sounds unlikely. Crazy. Bizarre. You’re right, but it’s a theory of evolution known as ‘punctuated evolution’. Somewhere, within the incredible modularity of evolutionary growth, the blind watchmaker (may have) designed a wormhole. This means that groups of participants in the system bifurcate (split in two) and run through the collapsed space to a new level of fitness (showing up in the fossil record as a new species).

Punctuated evolution is highly contested but the gist of it is that instead of species experiencing slow, gradual change (going through door after door to higher levels of fitness), they find a comfortable niche (a room with 4 doors) and stay there until jolted by a huge external shock. When this happens, modularity in the system is lost and instead of slow, adaptive, trial and error exploration of pathways, huge cascades are set off.

I have taken a long time to get here but it really is time to talk about chaos theory. In our stable, ordered, predictable system, changes are slow, gradual and modular. Although some parts of the system are bigger than others, the whole system responds to feedback loops that redistribute energy from the large, rigid organisms back down to modular, adaptable parts. This creates homeostasis and balance and allows the system to keep renewing itself (recycling its resources) with little change over long periods of time. But when a sudden large shock (external or from within the system) perturbates the system, the lattice (phase space) collapses and parts of the system run (cascade) through the space. The characteristics of a complex system entering chaotic territory may seem familiar:

• Oscillations - the system wobbles back and forth without settling into a clear pattern

• Volatility -the situation is fast moving and changes rapidly

• Interconnectedness - separate problems all appear to synchronise

• Cascades - problems in one area lead to escalating problems elsewhere

• Power laws - apparently insignificant events become the ‘straws that break the camel’s back’

• Emergent properties - new, unforeseen problems / outcomes arise

• Tipping points - the situation goes below (or above) a previous established threshold

• Fractal behaviour - as a threshold is breached, other layers of the system start to show stress

• Feedback loops amplify - the more 'x' happens, the more 'y' happens

• Disintegration - the system fragments and stops responding to top down control

If you are finding it difficult to imagine this in a real life situation, think about the sinking of the Titanic. After hitting the iceberg, at first the ship appears to bob back and forth within its normal range of operation. But the situation is volatile and fast moving. A tipping point is reached and everything on the ship begins to cascade violently in the same direction. Previously disconnected parts of the ship now synchronise (deckchairs, furniture and people all tumble downwards). All this is fed by an amplifying feedback loop: the more water that rushes into the ship, the worse things get. As the ship begins to sink, other layers of the system start to exhibit stresses. Steel buckles, the crew stop obeying orders, passengers jitter around randomly. The ship breaks apart and loses its tensile structure, unable to respond to its controls.

Now not all systems will collapse in response to a large shock. The ability of a system to withstand large perturbations depends on its adaptability. Has the system been designed with sufficient redundancy or slack to absorb a sudden hit? Does the system have energy resources it can call upon to shore up the damage? Can these resources be acquired quickly enough to prevent collapse? Does the system have modularity such that the perturbation wipes out rigid upper layers of the ecosystem rather than collapsing the entire system? Is the system composed of self-organising parts, able to quickly reassemble or does a collapse in one part of the system cascade through all parts bringing the whole ship down?

These questions are critical for understanding resilience in complex systems in volatile times. I intend to write more about this particularly with regard to economies, businesses and even teams, to explain why many of the systems we have constructed (designed) are overly rigid, top heavy, unable to recycle energy efficiently and lack the necessary variation needed for continuous growth. This is inhibiting our ability to evolve to changing conditions, and increasing the likelihood of a very unpleasant collapse.

My own interests lie in economic units but anyone who is interested in climate change will recognise some of the points I raise in this post. What I hope to do, slowly (!) is try to make all this a bit less complex and easier to understand. If I succeed, you will eventually see the world as a set of interconnected complex adaptive systems, responding to energy inputs. Instead of trying to fix parts of the system, you will see the whole system. Many of the economic systems we have designed have lost their ability for variation, are overly specialised, too ‘efficient’ and are exhibiting some of the early signs of chaotic behaviour (power laws, interconnectedness, volatility and cascades). While 99% of people are distracted by political noise, I fear that the 1% are very aware that phase space is already flattening. If we are not careful, they will run through and collect the prize (again). With these posts, I hope to offer a new way of viewing our problems so that we don’t keep rebuilding the same fragile systems.