A feedback loop is a closed chain of causal influence: a change travels through connected variables and eventually affects the original variable. Feedback makes system behavior dynamic because consequences become conditions for later action.
A reinforcing loop amplifies movement; a balancing loop counteracts movement or seeks a target. Neither type is inherently good or bad. Real systems usually contain many interacting loops whose relative dominance changes over time.
Example
More users can create more recommendations, which attract more users. This reinforcing loop supports growth until a constraint—such as declining service quality—strengthens a balancing response.
See Reinforcing and Balancing Feedback Loops for a detailed explanation.