An institutional architecture of procurement

01 · A graph, not a chain

Decisions to procure weapons platforms are usually described as one-way top-down derivations. In such schemes the starting point is the political bodies, either the government or the general staffs as they set doctrine. These models are correct as anatomies but insufficient as mechanics, since they do not reflect the dynamics of the relations between actors or their consequences for procurement processes.

In general terms, processes are modelled as a set of steps or phases chained sequentially, with a review at the end of the process and no way of representing alterations to that flow. Reviews, taking place at the last step of the cycle, influence the next one, but are incapable of absorbing and reacting in real time to the surprises that arise during the process.

The idea that doctrine and the technological development of weapons are intertwined, or directly condition one another, is nothing new, and it is this relation that turns this structure into a scheme with mechanics rather than a simple list. What these models lack is not pieces, since those are already accounted for; what they lack is the dynamic that joins and conditions them. In none of the frameworks reviewed has a solution been found to a very simple problem: what happens when a decision travels through the system and meets resistance?

That is why we have modelled the object as a circuit of relations running in both directions, with five nodes and a relation of mutual inhibition, because, for instance, integrating into a supplier bloc and developing capability independently inhibit one another.

02 · The two pulses

It is normal to see the governments of states adopt different lines of reasoning when deciding to buy weapons systems or programmes, and although the final recipients of those programmes are the armed services of those countries, it is not uncommon for the decisions taken to run contrary to those that would follow if the general staffs were the ones deciding as they set doctrine. Two cases serve as examples: a shift in international relations with allied countries can redirect the effort of industry and the programmes it is able to develop; another clear case is that in difficult budgetary situations it is the most expensive programmes that are first delayed or cancelled. The problem does not end with the presence or absence of particular weapons systems in the inventory; it leads to that situation conditioning doctrine, according to whether certain capabilities are available or not. We should not ignore, however, that between decisions being taken and their effects there is a delay that until now no formal procedure takes into account.

On the other hand, it is not the first time that a civilian technology has found its way in and settled, one way or another, into the units, so that as its use consolidates it endows the government with opportunities it did not have before. The impulse comes from below and has, at first, no doctrinal backing; the units respond to a problem or a need with what they have to hand, and this happens particularly in open conflict. This first use by the units can nonetheless remain a mere anecdote if there is no support from higher levels, whether general staffs or government bodies. For this tech-push to prevail there must be an attitude of active support and assimilation into doctrine on the part of the general staffs, and at the same time no reluctance or outright opposition on the government side.

03 · The budget is the constraint

An unavoidable consideration is that the pulses of armament programmes require a budget in order to stay alive and reach the final phase. Given the idiosyncrasy of the structure of armed forces, it is inescapable that the executive and legislative bodies should control the budget lines assigned to those programmes. Since the budget line is limited, the model that emerges is one in which the pulses compete for a share of the allocation, from which a rivalry between pulses follows.

The outcome of that rivalry does not show which programme is better than another; it reflects which programme has been more successful within this dynamic, where actors, relations and decisions generate a structure through which programmes advance or fall back, driven forward or held up.

04 · The clash of pulses

It is perfectly possible, and indeed it is what happens, for two pulses to coexist in the system. The evolution of these pulses may bring them to meet at a node, at a body or an institution, and produce a clash. Each node is occupied by a government team, a military leadership, ministry staff or the technological and industrial base. Those occupants are the ones who facilitate or obstruct the passage of the pulses. Their position is sustained by the current state of the graph and they will offer opposition to change, the arrival of a pulse or the clash of two pulses being a manifestation of that change.

Resistance is neither declared nor explicit, which is why it is not known how much a node can take, and it can only be discovered by putting it under pressure. That is how the resistance it generates is revealed. In our model the point of clash changes with the phase. In Need and Study it is the General Staff, where the demand rising from the units meets doctrinal formalisation. From Approval onwards it is the Ministry of Defence, which on one side receives the political pressure of government expressed in domestic and foreign policy, and on the other takes on the action of the military establishment as regards doctrinal needs that condition operational capabilities. In both cases the clash is resolved by seeing which of the two thrusts prevails, and whichever prevails, the disputed capability ends up with its use restricted. The restriction of use is the trace of the dispute, not its result. It marks that there has been a clash over that capability, regardless of which side prevailed.

05 · The decision is not a parameter

Once a structure and an anchor are fixed, a seed is fixed as well, so that the world about which decisions are made is entirely determined before any decision is taken. Decisions are neither an initial parameter nor a parameter settled halfway through the process. The parameters are the anchor and the seed; the anchor fixes the world, its structure and its propensities, and the seed selects one realisation of that world. Decisions are what change what is done within that world. Deciding the queue or sequencing of the pulses, or the budgetary allocation, is what produces differentiated consequences over that same world, with anchor and seed held fixed.

This separation has a consequence that the model makes visible and that is worth declaring, because it is counter-intuitive. The fitness of a pulse, what it is, plus the affinity of its links plus the circumstances it falls into, predicts well who wins the allocation, but not how much of it reaches service. A pulse can lose almost every contest for funding and still enter service more often than any other, because it started close to the end of the route. When those two things come apart, what governs is not the strength of the pulse but where it set out from. It is the starting position, and not fitness, that weighs most on the outcome, and this is not corrected by refining the model, because it is not a defect of calibration but a property of the structure.

06 · What the model makes examinable

What our model leaves examinable is principally what each capability is exposed to in a given situation and what gives way first depending on where the pressure enters. In addition, anchoring adds a third thing: how far the starting situation bounds the outcome without deciding it.

In order to test what has been set out, what follows is the implementation of the algorithm in an interactive application. The user can anchor a situation and obtain a measure, observing the shape of the distribution. By then changing the anchor and measuring again, and comparing the two shapes, he will see that the starting situation bounds what is possible without going as far as deciding it, and will also see that sometimes the centre barely moves while the spread opens or closes, which is the difference between two countries that on average arrive the same and in risk are nothing alike.

07 · What cannot be read

It is important to stress that the absolute figures the model handles are not defensible, because they depend on propensities that we fix at the outset. What is legitimate to use is the difference observed between two clearly distinct situations and the shape of the distribution of those results. Two similar situations show no clear difference, and the model says so explicitly.

What cannot be seen, however many times the operation is repeated, is which of the weapons platforms prevails, and that is not obtained by refining the model. It is not a calibration problem that could eventually be solved with better data or better-adjusted propensities. The structure admits several outcomes because the situation it represents admits them. That is precisely the key distinction. The procedure can be repeated, discussed and redone, and even so the result is not guaranteed. Tractable is not predictable, because the space is not explored: exploration proceeds from an anchor.