Delay is the enemy of every control system
Why lag causes instability in machines and organisations alike.
Delay is the enemy of every control system, mechanical or organisational, because any system correcting itself based on measurements that lag behind reality is always steering by where things used to be rather than where they actually are now, and that gap between the true current state and the stale information a correction is actually based on is what turns steady, well-behaved correction into overshoot, oscillation, or outright instability, in a factory's supply chain exactly as readily as in a thermostat.
The physics of delay across systems
A control loop, whatever it is actually controlling, corrects based on a measurement, and every measurement describes the system as it was at the moment it was taken rather than as it is at the moment the resulting correction is finally applied. When that gap is tiny relative to how fast the underlying system changes, the correction still lands close enough to the true, current state that no real harm is done. When the gap grows large relative to how fast the system changes, the correction is increasingly acting on a picture of reality that has already moved on, and every one of the specific failures already covered in this collection, overshoot from correcting a moment too late, oscillation once gain and delay combine past a critical point, follows directly from that same widening gap between the measurement's age and the system's own pace of change. Nothing about this mechanism depends on the system in question being mechanical at all, an organisation reacting to information that is weeks old is fighting the exact same gap a thermostat fights when its sensor is running a few seconds slow, only stretched out to a timescale that makes the underlying cause far easier to miss entirely.
The supertanker comparison
A fully loaded supertanker, ordered to reverse engines or turn hard, takes a genuinely enormous distance, often several kilometres, to actually respond in any meaningful way, its own momentum carrying it forward largely unchanged for a long stretch after the command was given. A ship's officer who forgot this and steered exactly the way they would steer a small boat, correcting hard the instant a problem became visible and expecting an immediate response, would overcorrect constantly, since by the time any correction actually took effect the situation would already have moved on to something the correction no longer suited. Experienced officers instead act on where the ship will be well before it gets there, correcting early and gently rather than waiting for a problem to become obvious and then reacting sharply, precisely because they have internalised the delay between an order given and its effect finally showing up in the ship's actual heading.
Why organisations inherit the same problem
A large organisation reacting to a change in its market, a shift in customer behaviour or a competitor's sudden move, faces exactly the supertanker's problem stretched out across weeks or months rather than kilometres of open water. Information about the change has to travel up through layers of people before it reaches whoever can actually decide what to do about it, a decision then has to travel back down through those same layers before it becomes real action, and by the time that action is finally visible in the outside world, the market has often already moved on from the situation the decision was actually responding to. An organisation reacting sharply to news that is already months stale ends up doing to itself exactly what an over-aggressive control loop does to a mechanical system, overcorrecting to a picture of the world that is no longer the current one, and swinging from one overcorrection to the next rather than settling calmly toward whatever the market actually needs.
The one number worth remembering
A decision that takes several months to travel from being noticed to being acted on is, in control terms, responding to information that is already several months out of date by the time it finally changes anything at all, a delay easily long enough for the underlying situation to have shifted meaningfully, sometimes reversed entirely, in the time it took the response to arrive and take effect.
Where shortening the delay is worth more than reacting harder
Just as a control loop's usual instinct when a system misbehaves is to turn the gain down and correct more gently, an organisation's usual instinct when it feels slow to react is very often to demand sharper, faster-feeling decisions from the people already at the end of the existing chain, which addresses the symptom rather than the cause. The more durable fix, harder to arrange but far more effective, is shortening the chain itself, letting whoever is closest to noticing a change also be closer to acting on it, rather than asking people further down an unchanged chain to simply move faster once their turn in the sequence finally arrives. A faster reaction to stale information is still, in the end, a reaction to stale information, no matter how quickly it is delivered once the decision is actually made.
What this changes in practice
Recognising delay as the shared root of instability across mechanical and organisational systems alike points toward the same fix in both settings, shrinking the gap between something changing and someone or something noticing and reacting to it, whether that means a faster sensor on a machine or a shorter, flatter chain of communication inside an organisation. The theory of constraints, already touched on earlier in this set through the factory example in Goldratt's The Goal, rests on exactly this same recognition, that a system's real performance is limited by whatever is slowest to respond, whether that slowness lives in a machine's own physics or in the layers of people a piece of information has to pass through before anything about the system actually changes in response to it.