Other Mental Models · OM-10
A system's output can circle back and become its own input — amplifying itself further (a reinforcing loop) or correcting itself back toward stability (a balancing loop). Nearly every self-sustaining or self-correcting system runs on one of these two mechanisms.
A structure in which a system's output is fed back into the system as a new input, creating a circular causal chain rather than a one-directional cause-and-effect relationship. Reinforcing (positive) feedback loops amplify a change further in the same direction; balancing (negative) feedback loops counteract a change and push the system back toward a stable state.
A foundational concept in cybernetics, formalized by Norbert Wiener in his 1948 book Cybernetics, and subsequently central to systems theory, engineering control systems, ecology, and economics.
The Mechanism
Reinforcing loops amplify; balancing loops stabilize — most real systems contain both
A thermostat is the textbook balancing loop: as room temperature rises above the target, the system responds by reducing heating, pulling the temperature back down — the further from target the temperature gets, the stronger the corrective response, producing stability. Compound interest is the textbook reinforcing loop: a larger balance earns more interest, which enlarges the balance further, which earns still more interest — with no internal counterforce, the effect compounds without bound.
01 · REINFORCING LOOPS PRODUCE EXPONENTIAL, ACCELERATING CHANGE UNTIL SOMETHING EXTERNAL CONSTRAINS THEM
Left unchecked, a reinforcing loop doesn't self-limit
Because each cycle of a reinforcing loop amplifies the prior output further in the same direction, the resulting growth or decline is inherently exponential rather than linear — real-world reinforcing loops eventually hit some external constraint (a market saturating, a resource depleting, a system failing) since nothing in the loop's own internal logic causes it to level off on its own.
02 · BALANCING LOOPS CAN OVERSHOOT OR OSCILLATE IF THE CORRECTIVE RESPONSE IS DELAYED
A delay between deviation and correction can turn stabilization into instability
When there's a meaningful time lag between a system deviating from its target and the corrective response actually taking effect, a balancing loop can overshoot the target and oscillate rather than smoothly stabilizing — a well-documented failure mode in engineering control systems, economic policy responses, and organizational course-correction, where the delay itself becomes the source of instability.
03 · MOST REAL SYSTEMS CONTAIN MULTIPLE INTERACTING LOOPS OF BOTH TYPES SIMULTANEOUSLY
Overall system behavior emerges from the interaction and relative strength of many loops at once
Complex real-world systems (an economy, an ecosystem, an organization) typically contain numerous reinforcing and balancing loops operating simultaneously, and the system's actual observed behavior emerges from the relative strength and timing of these interacting loops — understanding a single loop in isolation often gives an incomplete or misleading picture of overall system behavior.
Where It Fails / Inversion
Where it fails / inversion
Not every observed pattern of change reflects a genuine feedback loop — some trends are driven by a one-directional external cause without any circular causal structure at all, and mistaking a simple linear cause-and-effect relationship for a feedback loop (or vice versa) leads to fundamentally mis-designed interventions.
How To Use It
Worked example · diagnosing runaway growth or persistent decline in a business metric
A team observing a metric that seems to be accelerating in one direction (viral user growth, or a persistent revenue decline) should look for the specific reinforcing loop driving it (more users generating more word-of-mouth generating more users; declining morale driving departures driving further declining morale for those remaining) — identifying the actual loop, rather than treating each cycle's output as an independent, unrelated event, reveals where an intervention could realistically break the cycle or introduce a needed balancing force.
How to use it
When a metric or system behavior seems to be accelerating in one direction without limit, look for the specific reinforcing loop generating that acceleration, and consider what balancing force (a natural constraint, or a deliberately introduced one) would be needed to stabilize it — reinforcing loops rarely self-limit without either an external constraint or a deliberately designed counterforce.
See Also