Excess Capacity

Excess capacity is the slack a factor carries above the level the goal actually requires — a workstation fast enough to keep the next one busy with room to spare, a table strong enough to hold far more than you ever put on it. In Goldratt’s factory model, every non-constraint should have excess capacity, while the constraint instead gets a buffer. This is necessary, not wasteful: because variance combined with dependent events is unavoidable, only spare capacity lets a station recover after a bad run, which is why a balanced plant (100% utilization everywhere) actually performs badly.

CF generalizes this into the empirical backbone of its binary epistemology. Temple argues that any system which actually works, repeatedly and stably, must have excess capacity on most of its factors — otherwise too many factors would sit near the borderline of failure and the system would routinely break. So when you evaluate options, most factors comfortably pass; they are good enough with a margin of error, and small changes within that margin should not move your judgment at all.

This grounds CF’s rejection of weighted-factor analysis and Bayesian credences. Adding up weighted factors rewards tiny gains in excess capacity that produce no real improvement, optimizing local optima with no global benefit. Worse, no single linear weight can capture a factor that has huge slack now (small changes irrelevant) yet could cause total failure if it ever crossed its breakpoint. CF’s answer: give abundant factors a pass/fail binary evaluation and spend attention only on the rare constraints.