Feedback Loop Identification¶

Why It Matters¶

Linear causal reasoning ("A causes B") works for simple chains but fails when effects circle back. Most interesting systems have multiple interacting loops. Identifying which loops dominate at any moment reveals why systems resist intervention, overshoot, or suddenly flip behavior.

The Key Move¶

When tracing a causal chain, always ask: does this effect eventually influence its own cause? Draw the loop. Determine polarity: count the negative links in the loop. Odd number = balancing; even number (including zero) = reinforcing.

Classic Application¶

Population dynamics. More rabbits -> more births -> more rabbits (reinforcing). More rabbits -> less food per rabbit -> more deaths -> fewer rabbits (balancing). The interaction of these loops produces oscillation, equilibrium, or collapse depending on parameters and delays.

Surprising Application¶

Skill acquisition plateaus. Practice -> skill -> success -> confidence -> more practice (reinforcing). But also: skill -> raised standards -> perceived gap -> frustration -> less practice (balancing). Plateaus occur when the balancing loop gains dominance. Intervention: detach self-assessment from external standards temporarily to let the reinforcing loop run.

Failure Modes¶

• Seeing feedback everywhere: not all circularity is feedback. Two things correlating because of a common cause isn't a loop.

• Polarity confusion: the sign of the loop isn't the sign of any single link. A "negative" link in a reinforcing loop doesn't make the loop stabilizing.

• Ignoring loop strength: identifying a loop tells you nothing about whether it matters. A real but weak feedback can be safely ignored.

Go Deeper¶

• Richardson, Feedback Thought in Social Science and Systems Theory - intellectual history of the concept
• Meadows, "Leverage Points" essay - how loop structure determines where intervention works