Control Engineering and Wargaming

A recent post on the Lost Battles email list got me thinking a bit. In essence, the post suggested two things. The first was that a wargame army, or indeed a real one, could be regarded as a system engineering problem and modelled as such. The second point was that the number of hits on an opposing army is less important than the distribution of hits. If I clobber an opponent hard at a critical point, this does more damage than distributing more clobbering over the whole army.

I might come back to the issue of the distributions of hits on an enemy force later, but for the moment I would like to focus on the idea of an army as a control system engineering problem.

We can match up the components of an army to those of a system. The troops, in their units (whatever that may be) are the parts we need to control. To do this we have a command structure, which is controlled by a general. The general wishes to make the system achieve his ends; in this case, the end is victory over the opposition.

Within this model there is also feedback, damping and noise.

Feedback is the observation or reporting of the outcome of activity. For example, the battalion you send to capture the village can report success, failure or being bogged down in a fire fight.

Noise is the plethora of unexpected or unplanned events that can knock a system out of joint, or at least impede to some extent the easy functioning of your army. So, for example, enemy actions could be noise. The unexpected arrival of a flank march could seriously disrupt the smooth stroll to victory that you were expecting.

Damping is the delay and inertia in the system responding to new orders. For example, a delay would be found in sending and receiving (and deciphering) a message. Inertia may well be found in, for example, the difficulty of the brigade you want to exploit a gap in the enemy line has in disengaging, changing front and charging.

There would also be delays in, for example, the generals responding to reports from their units and, of course, some circumstances where a unit would simply not report at all. For instance, notoriously, on the first day of the Battle of the Somme in 1916 the commanders lost contact with their troops and had no idea what was going on, where their units were or even if they still existed at all.

Sadly, many of the units no longer did exist, at least as fighting units, but even those gaps which had been created in the German lines could not be exploited due to the absence of information. This was not really solved until all units were equipped with radio; communication was still a problem the next year at Cambari, and was not really solved until World War Two.

I think, as a model of an army and its command and control structures this stands up reasonably well. We can, at least, identify the bits of the system with real world entities and decision making. I suspect, however, that there are two problems inherent within the model; the first is that the battlefield is a much more dynamic place than a control system and that which it controls, and the second is that the decision making function is distributed in an army, rather than centralised.

The first issue is that, as I think I have noted before, an army is a collection if individuals. Now, the individuals may be capable of acting as a coherent whole; indeed, we hope that they manage that. Even a modern platoon with is much greater distribution of soldiers than, say, a phalanx, has to operate as a unit in order to achieve anything. But the fact is that at some point the individuals will react as individuals, not as some nicely painted and firmly based wargame unit. 

As a slight aside on this, I am starting to suspect that what is important, at least up to modern warfare, is the morale of the officers rather than the troops themselves. After all, in a phalanx you have a large helmet on, a big shield and you are surrounded by colleagues, many of whom you are related to or have grown up with. You cannot see much of what is going on around you. What you do not know about is not going to crumble your morale.

On the other hand we hope that officers do have a bigger and wider view. They, therefore, are the ones who will spot the outflanking movement or incoming rear attack and, if they are not heroes, will probably run. Seeing the officer run would encourage the rear ranks of the phalanx to do likewise, and so the whole unit can collapse.

As I say, I digress, but the point is that army units are made up of normal, fairly rational humans who do not respond like automata, so that part of this model might need a little finessing.

Secondly, in a control engineering system, control is exhibited at one point. Now, you can argue that that is a relevant model, and in part it might be, but certainly throughout history there is a record of lesser commanders taking action – think Nelson at Copenhagen. I have raised the question before of who ordered the Greek wings to turn in at Marathon; it probably was not the Polemarch, who was the (nominal) overall commander.

So, again, we have a slight issue with our original model. Command functions are distributed throughout the system, at least down to the lowest level which we actually model. Actually, of course, it goes further than that: Private Smith can decide to run away, but if we do not choose to model Private Smith, we cannot account for his decision making in the model.

The problem with this, of course, is that it makes a nice, clean, intelligible model much more complex. On the other hand, as a model, it does throw some of the dynamics of an army into relief.