Dependent Events

Dependent events are activities linked in sequence so that a downstream step cannot start until upstream steps finish. In Goldratt’s factory example, station B works only on what A produced, and C works only on what B produced; each step inherits the timing of everything ahead of it in the chain. Goldratt introduced the term in The Goal, illustrating it with the dice-and-matchstick game where bowls feed parts forward turn by turn.

Dependency alone is harmless. The trouble appears when it combines with variance: real steps run faster or slower than their average. In a chain, a slow turn upstream starves the next step, but a fast turn cannot be banked, because the downstream step can only hold one position at a time and the extra output piles up uselessly. Negative fluctuations propagate forward and accumulate; positive ones are mostly wasted. This is exactly why a balanced plant underperforms its nominal average rate.

CF imports this as core systems intuition. The dependency-plus-variance argument is the mechanical case for why a weakest link always exists and why protecting it pays. CF generalizes the lesson beyond factories: any process of dependent steps (a project, a learning sequence, a chain of arguments) needs buffers and excess capacity at the right points, not uniform balance. The same reasoning grounds drum-buffer-rope scheduling and supports CF’s preference for locating and exploiting the constraint rather than optimizing every stage equally. The key CF emphasis: this analysis holds only for dependent events, not for independent ones whose fluctuations average out on their own.