In praise of hierarchies

In which your humble correspondent has a minor disagreement with Chris Dillow.

Blogging titan Chris Dillow does not like managers.

As you would expect from such a luminary, Chris’s reasons for disliking the Boss Class are astute and well-thought-out:

The pursuit of efficiency – managerialism’s main goal – cannot be a value-free exercise. “Efficiency” has many conflicting meanings. Does it mean increasing GDP per head, Pareto-optimality, utilitarianism, maximizing Rawlsian primary goods or maximizing capabilities, in Amartya Sen’s sense? How do we choose a meaning of efficiency from this menu? And if we choose utilitarianism (as managerialism often does) what is the moral justification for imposing costs on some – up to and including death, Mr Hoon – so that others can gain?

Chris makes the important distinction between managerialism and management. Managerialism is the ideology of the managerial class, and as such is the ruling ideology of our age. Management, or technocracy, is the application of science to the problem of achieving certain ends given certain resources, and is therefore a branch of engineering*.

I disagree with Chris that all top-down management is harmful. My reasons for thinking this are as follows.

An industrial society requires specialisation. Our world is too complex for one individual intellect to encompass any more than a tiny fraction of it in any depth. In order to function, our industrial society requires that people specialise and develop expertise in particular areas. We need teachers, nurses, paramedics, electrical engineers, baristas, taxi drivers, geologists, science fiction writers, booksellers, economists, entertainers, and economics bloggers. No one (contra Heinlein) can learn how to excel in every one of these areas.

Specialisation requires informational transfer. Individual people, with their individual specialisations, need a way of communicating. It is only through communication that these individual experts will be able to co-ordinate, and it is only through co-ordination that they will be able to deliver the products, projects, and services that we desire.

How is this co-ordination to be achieved? There are two broad answers to this question. One is hierarchy, and the other is… Something. I’m not going to say ‘markets’ because I’m fed up with that word. Call it decentralised co-ordination.

DIGRESSION

I know what a hierarchy is. I understand what a hierarchy is. It is a situation in which one person is vested with authority, and this authority entitles them to tell other people what to do, and it requires that those people do actually do what the first person tells them to do (within various limits and subject to various contextual caveats yaddah yaddah…).

But what is a market?

Seriously.

At this stage a market might as well be an invisible unicorn that grants magical wishes, or a pair of intersecting lines on a page of a textbook, or a place you go to at weekends to buy overpriced honey with bits of honeycomb left in it. I honestly don’t know any more. “The market” means so many things to say many different people that I give up on trying to understand it.

So instead of market, I will use the term decentralised co-ordination. This encompasses a much broader set of behaviours than just whatever markets are taken to be. When motorists are traversing a roundabout they are engaging in decentralised co-ordination. The crowds who walk down Oxford Street are engaging in decentralised co-ordination when they avoid bumping into each other (most of the time).

END OF DIGRESSION

De-centralised co-ordination has a problem. This problem is best illustrated by an example. Suppose we want to build an aeroplane. This is a complex task and therefore requires specialisation. Specialisation requires co-ordination. Co-ordination requires communication. The problem is that within a completely  decentralised system of co-ordination each individual would have to communicate with each other individual. This results in a combinatorial explosion in the amount of communication that has to go on.

If you have, say, 300 engineers working on the design, manufacture, and commissioning of your aeroplane then, in a completely decentralised situation, each engineer would have to engage in 300! interactions (that’s factorial 300) in order to co-ordinate her individual efforts with her fellow engineers.

How much is 300 factorial? If we assume that each interaction will be a “brief chat”, how many “brief chats” will our engineers have to engage in so as to design, manufacture, and commission our aeroplane?

According to Wolfram Alpha, about this many:

30605751221644063603537046129726862938858880417357699941677674125947

65331767168674655152914224775733499391478887017263688642639077590031

54226842927906974559841225476930271954604008012215776252176854255965

35690350678872526432189626429936520457644883038890975394348962543605

32259807765212708224376394491201286786753683057122936819436499564604

98166450227716500185176546469340112226034729724066333258583506870150

169794168850353752137554910289126407157154830282284937952636580145235

23315693648223343679925459409527682060806223281238738388081704960000

00000000000000000000000000000000000000000000000000000000000000

00000000

Wow. OK.

So, what do we do?

A far more sensible option is to divide the engineers into functional groupings and have each group sub-divided into smaller groups, where each group reports to one particular engineer who digests their information and then passes it up the chain, whilst receiving instruction from those above based on what she has told them; in other words, a hierarchy.

If we have one engineer at the top of the chain, and if she has seven engineers reporting to her, and they have seven engineers reporting to each of them, and they have seven engineers reporting to each of them… we end up with a potential capacity of 1 + 7 + 49 + 343 engineers. What are the total number of “brief chats” required now?

Well let’s see, we have one to seven, then each one of those seven to seven, then each one of those seven to (substantially less than) 343.

So it’s 7 to the 4 right? That’s how many brief chats are going to have to take place. Which according to Wolfram Alpha is:

2401

Which is a lot, yeah? But a lot less than we’d be dealing with if we expected our engineers to communicate on a completely non-hierarchical basis. Two thousand four hundred and ten is a tractable number. You can do business with that kind of number.

This is why hierarchies exist. They are more informationally efficient than non-hierarchies. We have managers because if we didn’t, we couldn’t have cheap flights, we couldn’t have a nuclear deterrent, we couldn’t have the Internet, we couldn’t have electricity. We couldn’t have almost all of the goods and services afforded by our wonderful industrial society.

Bosses exist for a reason; if we didn’t have them, we’d have to create them. Or live in mud huts and grow all our own food.

This may sound strange. Many people think of hierarchies as barriers to communication. We see them as controlling, fusty, bureaucratic, and inefficient.

But really they’re not. Like democracy, hierarchies are the least bad solution to the problem they are intended to solve.

LEFT OUTSIDE SUGGESTS KAIZEN

Some guy on Twitter disagreed with my disagreement with Chris. So what about Kaizen? Well Kaizen is all very well. But the other pillar of the Toyota production system is something called Just In Time. JIT is not a system that spontaneously arose because of the self-interested actions of atomistic individuals working as equals. JIT requires enormous amounts of discipline, morale, and managerial gumption to implement in the first place, and then requires that workers maintain that level of discipline consistently for a long period of time.

So how does one ensure that the workers maintain that level of discipline? Whose responsibility is it to ensure the workers are kept in line?

Well, the bosses, of course.

Once again, the question devolves to an issue of who benefits. Are workers better off for being corralled and controlled? Are the fruits of hierarchical, specialised industrialisation such that workers are – on the whole – better off for being managed? I honestly don’t know.

The real problem here is…

THE POLITICAL ELEMENT

Chris has a point. Left Outside has a point. Hierarchies lend themselves to rent-seeking and power-grubbing. Here’s a points from Left Outside:

I’m thinking more along the lines of avoiding bad news, important people overriding the more knowledgeable.

This happens. But I submit that this will happen in basically any situation involving human beings ever. This is a problem of power, not hierarchy, per se. The problem, as ever, is politics. It is about the morality of the right of some people to dictate to others, as opposed to the effectiveness of the right of some people to dictate to others.

Kaizen is a good thing to have, but it only works if you treat your employees with respect. I don’t know what ‘respect’ means in this context. Some people would argue that demanding that people work for a capitalistic enterprise is inherently disrespectful, but there you are.

Bosses are always going to exist. I’d rather they were all elected by their peers, rather than being appointed by our capitalist overlords; but the fact remains they are necessary. Any socialist utopia worth living in will have bosses, because any socialist utopia worth living in will need the kinds of products, projects, and services that can only be provided by hierarchical organisations.

IN SUMMARY

Like any technology, hierarchy can be used for malicious and anti-social purposes. But that doesn’t mean the technology is intrinsically harmful or that it isn’t worth using in the first place.

*Chris calls technocracy a practice, in the sense used by Alasdair MacIntyre.