Audit every model diagram for missing feedback loops — do effects circle back to influence causes?
After drawing a mental model, audit it for missing feedback loops by tracing whether any effects circle back to influence their own causes, because circular causation governs most complex systems but is invisible to linear thinking.
Why This Is a Rule
Linear thinking is the default: A causes B causes C. But most complex systems operate through circular causation: A causes B, which causes C, which feeds back to amplify or dampen A. Teams that ship faster (A) build confidence (B), which increases ambition (C), which increases scope, which slows shipping (back to A — a balancing loop). Without the feedback loop, the model predicts unlimited acceleration. With it, the model predicts oscillation — a fundamentally different behavior.
The feedback loop audit asks one question for every effect in your diagram: "Does this effect circle back to influence any upstream cause?" If the answer is yes for even one effect, you have circular causation — and your model must account for it or it will produce incorrect predictions.
Most initial mental model diagrams are linear because linear thinking is faster and feels complete. The audit is a deliberate second pass that catches the circular dynamics the linear pass missed. Senge's "Fifth Discipline" identifies feedback loop blindness as the primary obstacle to systems thinking — this rule addresses it directly.
When This Fires
- After drawing any mental model diagram (Draw mental models as boxes and labeled arrows within 10 minutes — spatial layout reveals hidden gaps)
- When a model produces predictions that seem too simple for a complex system
- During design reviews when evaluating system behavior predictions
- Any time you've modeled a complex system and want to check for missing dynamics
Common Failure Mode
Finding no feedback loops and concluding the system is purely linear. Most complex systems have feedback loops — if your model has none, the model is almost certainly incomplete, not the system. Zero feedback loops in a model of team dynamics, market behavior, or organizational change is a model failure, not a system property.
The Protocol
After completing a mental model diagram: (1) For each effect in the model, trace forward: does this effect eventually circle back to influence any of its own causes? (2) Check for reinforcing loops (A amplifies B amplifies A — exponential growth/decline) and balancing loops (A causes B which dampens A — self-correction). (3) If feedback loops are found → add them to the diagram and evaluate how they change the system's predicted behavior. (4) If no feedback loops are found in a complex system model → the model is likely incomplete. Ask: "What could the effects of [end node] do to [beginning node]?"