Constraint

A constraint (also bottleneck or limiting factor) is the part of a system that caps its throughput toward a goal. Goldratt’s Theory of Constraints uses the chain metaphor: a chain has exactly one weakest link, and its strength is set wholly by that weakest link. Strengthening any other link is wasted, because the chain still fails at the same load. In a factory the constraint is the slowest step everything else waits on; improving a faster step just produces parts that pile up before the bottleneck. Hence Goldratt’s rule: optimization away from the constraint is wasted. The constraint is found and managed through the five focusing steps — find it, exploit it, subordinate everything else to it, elevate it, then repeat once it moves.

CF generalizes this beyond business. Because real dependent systems have one or a few constraints, most parts carry excess capacity — they are already more than good enough, so changes to them don’t affect the outcome. This grounds CF’s decision-making: a constraint is the decisive consideration, the single factor not good enough, and binary evaluation amounts to locating it rather than weighting and adding fifty qualitatively different factors. It also explains why optimizing local optima gives no global gain, and why genuine progress comes from simple, high-leverage changes at the few degrees of freedom that govern the system. CF further extends the idea to thinking and learning via constraint applied to epistemology.