Feedback Loops

How Feedback Loops Shape Systems

Every system that changes over time has stocks (accumulations — things you can count or measure) and flows (rates of change — things flowing in or out). Think of a bathtub: the water level is the stock, the faucet is an inflow, and the drain is an outflow. The water level at any moment depends on the history of all the water that flowed in minus all that flowed out.

What makes systems interesting is feedback — when the stock influences its own flows. A reinforcing loop (positive feedback) amplifies change: the more you have, the more you get. Population growth is the classic example — more people means more births, which means more people. This produces exponential growth, the "hockey stick" curve. Left unchecked, reinforcing loops produce explosive or collapsing behavior.

A balancing loop (negative feedback) seeks a goal. A thermostat measures the gap between actual and desired temperature, then adjusts heating to close that gap. The bigger the gap, the harder it pushes. This produces goal-seeking behavior — the system smoothly approaches its target.

Add a delay to a balancing loop and things get wilder. If the thermostat reacts to a delayed temperature reading, it overshoots — heating too long because it doesn't realize it's already warm. Then it overcorrects the other way. The result is oscillation. Longer delays and stronger corrections make the swings worse. This is why the shower temperature is hard to get right, and why supply chains oscillate.

Real systems usually combine both loop types. The S-curve (logistic growth) starts with reinforcing growth, but as the population approaches its carrying capacity, a balancing loop kicks in and slows things down. The result is the characteristic S-shape — fast growth, inflection point, then leveling off. Bacteria in a petri dish, adoption of new technology, fish in a lake — all follow this pattern.

Overshoot and collapse happens when a reinforcing loop drives growth past what the environment can sustain. Population grows while resources are plentiful, but consumption depletes the resource faster than it regenerates. By the time the system "notices" it's overshot, the resource base is gone and population crashes. This is the story of Easter Island, overfished stocks, and boom-bust commodity cycles.

Predator-prey dynamics (Lotka-Volterra) produce a different pattern: sustained oscillations where neither species drives the other to zero. When prey are abundant, predators thrive and multiply — which reduces prey — which starves predators — which lets prey recover. The two populations chase each other in endless out-of-phase cycles. Lynx and hare populations in Canada follow this pattern almost perfectly.

Escalation is a reinforcing loop between two actors. Each side builds up to exceed the other by some margin, which triggers the other to build more. The result is a ratchet — both sides grow even though neither gains a lasting advantage. Arms races, price wars, and social media engagement algorithms all follow this pattern. Asymmetric rates mean one side temporarily pulls ahead, but the other always responds.

The tragedy of the commons emerges when multiple actors share a regenerating resource. Each actor benefits from harvesting more, but the resource can only regenerate so fast. When total harvesting exceeds regeneration, the resource collapses — and everyone loses. Fisheries, groundwater, and shared bandwidth all suffer from this dynamic. The key insight: individual rational behavior produces collective catastrophe.

Eroding goals is a subtle trap. A balancing loop tries to close the gap between performance and a goal, but when progress is slow, a second loop quietly lowers the goal toward current performance. The system settles at a lower equilibrium than anyone intended. "Good enough" becomes the new standard. This is how quality standards slip, exercise routines degrade, and organizations slowly accept mediocrity.

These ideas come from Donella Meadows' book Thinking in Systems — one of the clearest introductions to systems thinking. The core insight: you don't need complicated models to understand system behavior. Stocks, flows, and feedback loops are enough to explain most of the dynamics you see in the world.