Constraint

A constraint is anything that limits a system's ability to generate its goal (usually throughput, often money). ToC uses constraint and bottleneck interchangeably.

Types of constraints

• Physical. A machine, a person, a square-footage limit.
• Policy. A rule that limits throughput (e.g. "no overtime").
• Market. Demand-side limit on what you can sell.
• Variance. Random fluctuations that the system can't handle.
• Buffer size. A small buffer starves the next step.

Why "constraint" not "problem"

ToC avoids calling constraints "problems" because:

• Problems have many causes. Constraints are singular.
• Problems need solving. Constraints need managing (exploit, subordinate, elevate).
• Problems are negative. Constraints are often neutral or positive — they're where to focus.

CF and constraints

CF adopts constraints as the central concept for multi-factor decisions:

• Most factors are not constraints.
• The constraint is the borderline factor.
• Focus on the constraint; ignore the rest.

"The key to focusing improvements in the right place, and making them globally effective, is to find bottlenecks. Bottlenecks are also called 'constraints' or 'limiting factors'." — criticalfallibilism.com