System Dynamics | Foundation 15: Resilience
You'll Only See It Once It's Gone
On where we are. Resilience, self-organization, and hierarchy are three structural properties that explain why complex systems work as well as they do, and why they're so hard to see until they're gone. This is the first of the three.
Until now, we have explored the vocabulary of systems dynamics: system, element, relation, purpose, stock, flow, causal loop diagram, stock-flow diagram, reinforcing loop, balancing loop, coupling… We have also covered some simple systems that produce their own elegant, characteristic behaviors. If we push too far, systems can fall apart or show us unobserved behavior. But in general they are remarkably stable, and it is hard not to notice the quiet coordination in the way they function.
The irritating, repetitive question I have been asking myself for years is why systems work at all, in this very frictionless way. I can map the structure. I can trace the loops, identify the stocks, name the dominance. And yet, intuitively, I still feel amazed. Meadows offers three properties as the answer: resilience, self-organization, and hierarchy. This article is about the first of them.
Resilience has many definitions across different domains, and the one I love the most is short:
Resilience is the ability to recover strength, spirits, good humor, or any other aspect quickly.
Every word in that definition points at motion -recovery, springing back, elasticity. Resilience, read carefully, is a measure. It measures the capacity of a system to keep going inside an unsteady world, it is an ability to bounce back. Its opposite is stiffness, the system that has run out of ways to give.
Before we go further, there is a confusion worth sitting with. When people hear resilient, they tend to picture something unchanging. Maybe a wall that holds, a person who does not flinch, an institution that looks the same after a crisis as it did before. This is where the common use of the word leads us astray. Resilience, in the systems sense, is not about holding still. A resilient system can be extremely active, swinging through highs and lows, absorbing sudden outbreaks, moving through long arcs of growth and breakdown. Those movements can be the system’s normal, and resilience is what brings the system back to them. A system that appears to hold still, on the other hand, can be one shock away from breaking; it simply has not been hit yet.
The reason this distinction matters is that the two qualities show up on completely different schedules. Stability (the kind that looks like sameness) is legible on the surface. You can see it in the weekly numbers, the quarterly reports, the year-over-year comparisons. Resilience is not like that. You can run a system for years without knowing how much of it there is, because the information you would need is hidden inside what would happen under strain. You usually find out how resilient the system was only after it has been asked to prove it, and by then the answer may be unfixable.
What resilience actually is
So where does resilience actually live in a system?
In the loops, but not in any one of them. A single balancing loop is a correction machine. It pulls a stock back toward its desired level and stops. Resilience needs more than that. It needs many loops, of different kinds, running at different speeds, sometimes on the same stock and sometimes on each other. One loop corrects in seconds. A second corrects across days. A third only engages when the first two are overwhelmed. The system’s capacity to absorb a disturbance comes out of that layering.
The clearest place to see this is our own body. Take body temperature. When you are cold, the small blood vessels near your skin constrict first. If that is not enough, you shiver. If the cold continues, your metabolism shifts, and your behavior changes: you look for a jacket, you curl up. Each of these is a balancing loop running at a different speed. If one fails, the next is already running. The rest of the body runs on the same plan. Immune response engages in layers. Collateral circulation opens when a vessel closes. And we carry literal backup hardware, two kidneys, two lungs, two of most things that matter.
We do not live strictly inside our bodies either. We learn, we make tools, we form bonds whose resilience can be borrowed in an emergency. Nobody has yet been resilient enough to keep dying from happening eventually.
Ecosystems run on the same logic at a larger scale. A healthy one carries many species and overlapping food webs, moving against each other in response to weather and disturbance. Underneath is a slower layer, genetic variation, that lets the system acquire responses to conditions its current members have never seen. Evolution is that loop.
What resilience looks like, when it is working, is this. Many loops. Different mechanisms. Different time scales. Redundancy where it matters.
Meta-resilience, and why you don’t see it until it’s gone
Here is where it gets interesting.
The loops we have been talking about are the everyday workers of a system. They do the jobs we notice. Underneath them, though, there is another layer: mechanisms whose only purpose is to keep the everyday loops in working order. DNA repair enzymes that catch transcription errors before they accumulate. Stem cells that replenish tissues as they wear down. Cellular machinery that clears out damaged cells before they cause problems. These loops do not regulate any function you would think to name. They regulate the regulators.
Meadows gives this layer a name. She calls it meta-resilience, and she is pointing at something specific. A resilient system can absorb a shock. A system with meta-resilience can also rebuild the parts of itself that did the absorbing, so that when the next shock arrives, the machinery is still there.
The trouble is that you cannot see any of this while it is working. Meta-resilience is invisible by design. The loops that maintain the loops operate quietly in the background, their output is the continued ordinary functioning of the system, and their failure shows up only when the everyday loops finally give out. At that point it looks like the everyday loops were the problem.
Many chronic diseases are diseases of this second layer. Cancer, cardiovascular disease, autoimmune conditions, in many cases these are not failures of a vital function so much as failures of the systems that were supposed to keep the function working. DNA repair stops keeping up with damage. Inflammation loses its off-switch. The outward function looks fine for years, and then it does not. What broke was not the visible loop. It was the machinery that was holding the loop together.
Ecosystems show the same shape. A forest with many species has meta-resilience built in: if one is lost, others fill the role through seed dispersal, succession, and genetic variation, which are themselves loops. Strip out half of the species, and for years the forest may still look the same. But the machinery that would rebuild it after a fire or a drought is gone. The next shock arrives at a forest that can no longer rebuild itself.
This is the shape behind coral bleaching, biodiversity loss, and climate change at a planetary scale. Kodak is the same pattern inside a company. It did not collapse the moment digital photography arrived. It collapsed because the mechanisms that would have let it act on what it noticed had been hollowed out years before.
The pattern is always the same. The system functions. It looks stable. And then it does not. What failed was never the visible loop. It was the loop behind it.
The plateau, and what we traded for it
Meadows has an image for all of this that I keep coming back to. She asks us to picture a system as moving around on a flat surface, a plateau, where it can go through its normal ups and downs in safety. A resilient system has a wide plateau. It has a lot of room to wander, and if it drifts toward an edge, soft elastic walls ease it back toward the middle. As the system loses resilience, the plateau narrows. The walls harden. Eventually the system finds itself standing on something like a ridge, where any movement risks sending it over a side it cannot climb back from.
The disturbing part of this picture is that a system standing on a ridge does not look, on an ordinary day, very different from a system on a wide plateau. You cannot tell by looking. The week-by-week numbers come back the same. Productivity is fine. Output is fine. Stability, as commonly measured, is fine. The difference between the two systems is entirely in how they would respond to a shock and you only find that out when the shock arrives.
Which is how resilience gets traded away. It is hard to see, and the things it costs us to maintain it are not. Redundancy shows up on a balance sheet as waste. Buffer stock ties up capital. A second supplier adds complication. An underused reserve capacity looks like inefficiency. When a system has multiple mechanisms doing overlapping work, the reasonable, defensible move is to remove some of them. The remaining mechanism still does the job. The system still functions. On a normal day nothing looks different at all. The only thing that has changed is what the system could survive.
The Texas electrical grid in February 2021 is one public version of this trade. For decades the state had kept its grid separate from the rest of the country, a choice that made it cheaper to run and also removed the possibility of drawing power from neighbors in an emergency. Under normal weather, the trade looked free. Then a cold snap came through, demand surged, generation froze, and the system had nowhere to reach. What failed was not generation or transmission. It was the meta-loop that had been removed years earlier in the name of cost.
Seen this way, practices that look inefficient in the short term start to make structural sense. Holistic medicine, at its non-commercial core, tries to do more than treat the immediate illness. It tries to rebuild the body's capacity to handle the next one on its own. Development aid that moves past distributing food or money, and works to repair the conditions keeping people from feeding themselves, is the same move at a different scale. Both spend resources on the loops behind the loops. Both are, if you wait long enough, dramatically cheaper.
What resilience asks of us is uncomfortable. It asks us to manage systems for a property we cannot measure directly, to defend redundancies that look like waste, to protect mechanisms whose job is to protect other mechanisms. It asks us to value a plateau whose edges we cannot see. To be honset, nobody gets promoted for doing this. And yet the difference between a system that lasts and a system that surprises you with its collapse is almost always here: in the loops behind the loops, in the walls of the plateau.
🧩 What’s Coming Next
This foundations series will build your systems thinking toolkit step by step:
2 | Stop! Let’s Talk Stocks: Not Wall Street, Just Bathtubs ✔️
3 | Go With the Flow: Pipes, Currents, and Traffic Jams (A Love Story) ✔️
4 | Causal Loop Diagrams 101: Stop Talking, Start Drawing ✔️
7 | Reinforcing Feedback Loops: Congratulations, You Made It Worse ✔️
10 | One-Stock System Dynamics: Choose Your Own Catastrophe ✔️
14 | Exponential Growth & Collapse: Fine, Fine, Fine, Oh No ✔️
15 | Resilience: You Won't Notice Until It's Gone ✔️
16 | Self-Organization: Assembly Not Required
📚 Main Resources
Meadows, D. H. (2015). Thinking in Systems. Chelsea Green Publishing.
Sterman, J.D. (2000) Business Dynamics: Systems Thinking and Modeling for a Complex World. Irwin McGraw-Hill, Boston.
My lecture notes from “System Dynamics” and “Simulation” classes :)
Some explanations and phrasings closely follow or directly quote these sources. The text was refined for coherence and citation accuracy with the assistance of large language models.