A feedback loop is a system structure in which a process’s output is routed back as an input, either reinforcing the trend (positive/reinforcing feedback) or counteracting it (negative/balancing feedback).
The idea
In a reinforcing loop, more output produces more of the same effect, driving exponential growth or collapse (compound interest, viral growth, arms races). In a balancing loop, more output triggers a corrective response that pulls the system back toward equilibrium (a thermostat, supply and demand, homeostasis). Most real systems combine both loop types, and their interaction — not either loop alone — determines long-run behavior.
When to use it
- Diagnosing why a system is growing, stalling, or oscillating
- Designing incentive or measurement systems, where fast, accurate feedback enables course correction
- Understanding delayed effects, where a correction arrives too late and causes overshoot
How to apply it
- Map what output feeds back into the system as an input, and through what delay.
- Classify the loop as reinforcing (amplifies) or balancing (dampens).
- Look for the dominant loop at each point in time — dominance often shifts as a system evolves.
Watch out for
- Long delays between action and feedback make loops hard to see and easy to over- or under-correct
- Reinforcing loops without a counteracting balancing loop tend toward runaway growth or collapse
- Metrics that create their own reinforcing loop (optimizing for a proxy) can distort behavior
Related models
- Compound Interest — a canonical reinforcing feedback loop.
- Second-Order Effect — feedback loops are a common source of second-order effects.
- Bottleneck — balancing loops often act through bottlenecks.
Sources
Donella Meadows, Thinking in Systems: A Primer (2008); Jay Forrester’s foundational work in system dynamics.