Training cadre bottleneck

Why does doubling an army's size take far longer than doubling its equipment?

A state decides to double its army. The money is voted, the factories are told, and within a couple of years the rifles, vehicles and shells genuinely have doubled. The soldiers have not. The units exist on paper; the second half of the army is somehow not the same thing as the first.

The obvious explanation is that people are harder to recruit than steel is to make. But that is usually false — in a national emergency, volunteers and conscripts are the abundant input, not the scarce one. So the shortage is somewhere else. Where?

Compare the two production systems, because they have different shapes and that is the entire answer.

To double equipment output, you build a second production line. What does that consume? Capital, machine tools, floor space, material. What it does not consume is the first line. The process is parallel: additions do not come out of the existing stock, so it is limited by money and industrial capacity, both of which a state can throw at the problem.

Now ask how a soldier is produced. Competence is not manufactured; it is transferred, person to person, from someone who already has it — which means the input to the training system is drawn from its own output. That is a completely different structure. The rate at which new soldiers can be made is proportional to how many competent instructors you currently have, and every one of them was made the same way.

A process whose rate is proportional to its own stock grows geometrically, and geometric growth sounds fast. So why is this slow? Two reasons, and they compound.

The first is a maturation delay. A recruit who completes training cannot thereby train anyone. What makes an instructor is not the certificate but the years afterwards — judgement about what actually goes wrong, which is precisely what cannot be transmitted by someone who has not seen it. So the pool of instructors grows with a lag of years behind each graduating cohort. Insert that lag into the geometric process and the doubling time is set by the maturation period rather than the length of a course. More barracks and rifle ranges do not touch it.

The second is that the withdrawal is not free. Every experienced sergeant taken out of a company to run a training school is one that company no longer has, so expansion is paid for by degrading the existing force at exactly the moment the emergency is under way. The more urgently you need to expand, the more you need the existing units intact — and the same people are the answer to both.

A third consequence explains a strange kind of failure. Suppose a state ignores all this and reaches its headcount target anyway. What has it built? A force in which the ratio of experienced to inexperienced people inside each unit has fallen, perhaps far. But a unit's effectiveness is not headcount — it is closer to headcount multiplied by how well experience is distributed through it. So it is entirely possible to double the soldiers without doubling the army, and to discover the difference only under contact.

Can we test this? If the binding constraint is the stock of experienced people rather than facilities, three things should follow. Money, buildings and ranges should not shorten expansion time much. An influx of experience from elsewhere — recalled reservists, returning veterans, instructors lent by an ally — should shorten it a great deal, disproportionately to their numbers. And expansion should be accompanied by a visible decline in the units the instructors came from. The refuting observation would be an army whose expansion rate tracked its training budget rather than its pool of experienced non-commissioned officers, with the parent units showing no measurable degradation. That would mean the bottleneck is infrastructure after all.

Three refinements, one of which is a concession.

First, the distinction is not really "people versus machines". It is parallel production against autocatalytic production — a process that needs some of its own output as an input. Hence the same pattern wherever expertise passes person to person: pilots, surgeons, welders certified to a code.

The concession is that the industrial side is not as parallel as I made it sound: a second production line requires machine tools, and machine tools require toolmakers, who are trained by toolmakers. Industrial mobilisation has its own cadre bottleneck one level down. The contrast is real but a matter of degree — equipment production has fewer autocatalytic stages, and more of its constraints yield to money.

Second, none of this says expansion is impossible — only that its rate is set by a quantity that cannot be bought. States that expanded quickly did so by finding stocks of experience already in existence: reservists, veterans of an earlier war, allied training missions, or long-service regulars kept in peacetime for exactly this purpose. That last is the real function of a small professional army in a state that expects to need a large one, and it is why cutting such a force looks cheap and is not.

Third, the timing is unforgiving. Because the cadre must mature before it can multiply, the decision to expand must be taken years before the force is needed; a state that waits until the threat is obvious is starting a process whose early years produce nothing but a weakened existing army.

• Why peacetime professional armies are sometimes described as a training cadre with a force attached.
• How the same maturation lag governs the reconstitution of a force that has taken heavy losses among its experienced ranks.

Every term the collection defines is gathered in the glossary.