The Unaccountability Machine

The Unaccountability Machine— Dan Davies

Part 1: The Nature of the Crisis

1.Something’s Up

““Capitalism is disappearing, but Socialism is not replacing it. What is now arising is a new kind of planned, centralised society which will be neither capitalist nor, in any accepted sense of the word, democratic. The rulers of this new society will be the people who effectively control the means of production: that is, business executives, technicians, bureaucrats and soldiers.”

George Orwell, ‘James Burnham and the Managerial Revolution’,1946

The idea was that the financial system of the developed world, from around the fall of Communism in Europe, had reached a point where the overall system of incentives in the economy was so criminogenic that banks had a natural tendency to organize themselves into fraudulent behaviour.

Then the 2008 financial crisis happened, followed by a long period of recession and austerity, and suddenly it turned out that the technocratic consensus wasn’t as competent or moderate as it had appeared. Ten to twenty per cent of the electorate suddenly realized that they might have to take an interest in politics after all. So they started paying attention again, and they didn’t have the basic assumptions of the mainstream. All they knew was that the people who used to be in charge seemed to have screwed things up mightily.

The fundamental law of accountability

It is important to be clear at this stage, exactly what an accountability sink is, and how they are constructed. It’s not just the way in which the hourly paid worker has been set up to act as a human shield. In order to make the sink effective, you need a combination *of things: that person, plus a policy that there’s no way to appeal the decision by communicating with a higher level of management. (Even if you somehow managed to get the CEO’s phone number, you would come up against the fact that the policy was in place precisely to protect them from making that decision personally.)

So the crucial thing at work here seems to be the delegation of the decision to a rule book, removing the human from the process and thereby severing the connection that’s needed in order for the concept of accountability to make sense. You could even coin a sort of law management here:

The fundamental law of accountability: the extent to which you are able to change a decision is precisely the extent to which you can be accountable for it, and vice versa

The construction of accountability sinks has damaging implications for the flow of information. For an accountability sink to function, it has to break a link; it has to prevent the feedback of the person affected by the decision from affecting the operation of the system. The decision has to be fully determined by the policy, which means that it cannot be altered by any information that wasn’t anticipated. If somebody can override the accountability sink and overrule a policy that is in danger of generating a ridiculous or disgusting outcome, then that person is potentially accountable for the outcome.

*If nothing else, you’ll have a few tips about how to set things up in your own job to divert any troublesome accountability that might be building up.

In any case, it is neither psychologically plausible nor managerial realistic to expect someone to follow orders 99 per cent of the time and then suddenly act independently on the hundredth instance.

Although dramatized by grim humour, the KLM squirrel debacle illustrates a few potentially important things about the underlying reality of management and information. A decision with no real owner had been created because it was the outcome of a process. The process worked well, until something that hadn’t been anticipated (the pet squirrel craze) showed up, and then it delivered disastrous results. There was no effective way in which information could be fed back to the people who could change the policy, so the decisions continued to get worse. And then, when something so outrageous happened that it couldn’t be kept out of newspapers, there was nobody to blame.

Academic publishing is extremely profitable. The foundations of Robert Maxwell’s media empire were built on Pergamon Press, which was one of the first companies to realize that the business model has two incredibly favourable properties. Firstly, the customer base is captive and highly vulnerable to price gouging. A university library has to have access to the best journals, without which the members of the university can’t keep up with their field or do their own research.

Secondly, although the publishers who bought the titles took over the responsibility for their administration and distribution, this is a small part of the effort involved in producing an academic journal, compared to the actual work of writing the articles and peer-reviewing them. This service is provided to the publishers by academics, for free or for a nominal payment (often paid in books or subscriptions to journals). So not only does the industry have both a captive customer base and a captive source of free labour, these two commercial assets are for the most part the same group of people.

A not-wholly-unfair analysis of academic publishing would be that it is an industry in which academics compete against one another for the privilege of providing free labour for a profit-making company, which then sells the results back to them at monopoly prices. It is, as you’d expect from that description, highly profitable – and passionately hated by the academics.

However, the model persists because somewhere along the way, the journal industry managed to insert itself into the staff promotion and recruitment function of universities all over the world. In doing so, it created an extremely useful accountability sink for senior academics and managers of universities, while also solving an awkward and unpleasant interpersonal problem for them – how to judge the quality of scholarship without offending the scholars.

If you trace back many important decisions of the last few decades, you will regularly come up against the uncomfortable sensation that the unacknowledged legislators are relatively junior civil servants who put placeholder numbers in spreadsheets, which are later adopted as fundamental constraints; to do otherwise would mean someone having to risk being ridiculed for making a decision.

For nearly all of history, there have been two kinds of authority taking the big decisions affecting people’s lives. There is a fundamental distinction between ‘kings’ and ‘priests’. A king might be more powerful, but his orders can be argued against – it might be inadvisable to do so, but if you can change the king’s mind you can change the decision. The priest, on the other hand, gains his authority from his status as the interpreter of the Word of God, so his decisions are considerably more difficult to reverse. This means that it matters a great deal which kinds of decisions are giving to which kinds of authorities, and the question of the boundary between the two spheres has often been one of the central issues of entire eras — it was the subject of the Thirty Years War in Europe. A lot of the discontent in the modern world might come from having taken decision-making structures that were designed with ‘king-like’ leaders in mind, and handing them over to managers who didn’t act in the same way.

A good friend had lent me some books several years earlier, which I had never opened because they looked difficult. He’d said at the time that there was only one really good treatment of the importance of information and feedback in social sciences, and that it was a shame that it had never caught on. I reached up to the top shelf and pulled down the first volume that my hand landed on. It was called Brain of the Firm, by Stafford Beer.

2. Stafford Beer

Later on, Beer would identify it as one of his fundamental axioms: ‘It is not necessary to enter the black box to understand the nature of the function it performs.’ The underlying idea is somewhat stronger than this: if the black box is a complex system, it’s likely to be pointless — or even dangerous — to try to understand its inner workings and use that understanding to manipulate a precise outcome. This is a matter of respecting the complexity of the problem — a genuinely complex system is one in which you cannot hope to get full or perfect information about the internal structure, and cannot have any acceptable degree of confidence that the bits of information you don’t have can be safely ignored. Rather that trying to use a mixture of partial information, preconceived theory and guesswork, you need to step back, accept that the system will keep its secrets, and observe its behaviour. And of course, this means that the different observers might have different opinions. The property of ‘being a black box’ isn’t an objective one — it’s a description of a decision taken by someone working on the system that they don’t have enough understanding to safely treat it in any other way. In some of Beer’s writings, the property of complexity itself is also dealt with in this way; rather than trying to count combinations of connections, you define what it is for something to be a complex system by saying that it’s one which has to be treated as a black box.*

*if you can control something in your factory like a machine, it’s a ‘clear box’ in this language; there are even ‘muddy boxes’, where more detailed analysis could help understand the relationship between inputs and outputs but the system keeps a few secrets because really detailed analysis would require disproportionate effort.

Economists end up solving this problem by denying it (a mechanism that Stafford Beer looked at in detail, concluding that ignorance is a kind of information processing system of last resort). Where analysis fails, ideology steps in, and the solution the economists decided on was to fantasise an equivalent system for producers — a profit-maximising firm with an understanding of its market, a style production process and a manageable set of decisions. It ended up going disastrously wrong, but the original intellectual sin may have been the failure to respect the integrity of the black box.

So, cybernetics was the study of the control of systems — or specifically, the study of the control of systems with enough internal complexity for there to exist an interesting problem of how to manage them. The building blocks of cybernetic arguments are framed in terms of relationships between elements which have to be viewed as black boxes.

Stafford Beer’s cybernetics tells us that in these cases, while people’s opinions are important, the facts of the organisational outcomes are what we need to work with. In his most pithy formulation of the principle, he expands the black box principle to a rather more uncomfortable statement.

The purpose of a system is what it does.

Throughout Beer’s written work on management, this kind of meeting is always emphasised: unstructured, informal connections between staff at different levels and performing different functions. Some of the ephemera — particularly his emphasis on the importance of having adequate supplies of cigars and whisky to facilitate the conversation — might seem comically dated, but the idea of ensuring that there are links across the organisation to spread information and build consensus is entirely modern.

Beer was trying to adapt hierarchical systems so they were fit for their purpose in a changing world. He saw a strict command-and-control approach as dangerously inflexible, while excessive delegation would destroy the organisation’s ability to act as a coherent system. The fundamental relationship between management units in the kind of structures Stafford Beer designed was a ‘resource bargain’. A unit would be allowed to operate autonomously, but only to the extent that doing so did not jeopardise the broader system, whether in terms of a financial budget, physical resources of space, managerial time and bandwidth or general goodwill.

Central planning has two big problems: it’s planning, and it’s central. The first of these problems is perhaps the most obvious: a national economy is a big and absurdly complicated thing and the idea of writing down a huge spreadsheet or computer program to make all the decisions is an impossibly difficult task. But the second is perhaps the more wicked of the two: the problem of getting the information necessary to do any planning at all, and then centralising it in one place where the decisions are made. The problem of centralisation of information has a lot of logistical aspects, of course — the Soviet Union was always running into problems because it couldn’t change plans quickly enough in response to shortages and gluts. But it also has serious conceptual problems; Friedrich Hayek won a Nobel Prize and can fairly be regarded as on of the antecedents of cybernetics* for noticing that ‘information’ itself is a tricky concept, and that it can’t necessarily be treated like any other commodity.

*Stafford Beer records in one of his essays that he had read Hayek’s work in the 1950s and thought it was excellent stuff; he was profoundly shocked in the 1970s to find that Hayek was an economic advisor to General Pinochet.

3. Aliens Among Us

Maybe this third entity — the corporate owner of YouTube — is the non-human intelligence, and the algorithm is just one of its component parts. Maybe another, higher-level algorithm is made out of policies, employee handbooks and corporate communications. And if that’s the case, the hardware that’s running this algorithm is mainly made up of people, sending emails, holding meetings, conveying information to one another but shaped by the overall structures.

So, from here on in, I will try to refer to ‘decision making systems’ rather than ‘artificial intelligences’. Corporations are systems, and the make decisions, so they’re decisions making systems. The question is whether they’re black boxes or not — whether we are able to attribute the actions of the corporation to individual human beings within it.

What this means is that when you have to think about connections between parts, rather than just counting the individual parts of a system, the number of possibilities grows very rapidly; the potential combinations multiply, rather than just adding up. Very quickly, they multiply up to astronomically huge numbers, spelling absolute death to any hope of knowing the entire state of the system.

In fact, knowing only a few of the feedback circuits can be actively misleading, if you rely too greatly on your partial information. It is a sobering thought, for example, that despite employing some of the best and brightest* analysts in the world, the advice given by the US State Department over the last fifty years could comfortably have been outperformed by a parrot that had been trained to repeat the phrase, ‘Don’t start a war.’ The repeated failures of the State Department are not the consequence of ignorance; they are the consequence of having very good and deep — but not total — knowledge of an extremely complicated situation, in which facts outside of that information set turned out to be crucial. Knowing a great deal of detail about a subset of a system has a habit of increasing your confidence in your opinions disproportionately from their reliability.

*The phrase ‘the best and brightest’ is often used by people who don’t know that its original context was ironic. It entered the language as the title of David Halberstam’s book about the policy mistakes of the Vietnam War.

Here’s the thing: working inside a corporation (or any large organisation) is the quickest way to realise that you have only a partial understanding of how it works. You find yourself involved with decisions, but you know that you make them on the basis of collective understanding in line with policies, with regard to the sensitivities of other decisions, and so on. There are amazingly few occasions in everyday business life when somebody makes a specific and important decision as a creative act.

An organisation does things, and it systematically does some things rather than others. But that’s as far as it goes. Systems don’t make mistakes — if they do something, that’s their purpose. But it also works the other way around. Systems don’t have inner desires, so they don’t do things intentionally either. There’s just a network of cause and effect. We might think they’re conspiring, but they’re working within structures that made the outcome inevitable. Or we might see everything as a terrible cock-up, but we don’t understand that the outcome was the inevitable result of the way the system works.

Part 2: Pathologies of the System

4. How to Psychoanalyse a Non-human Intelligence

The trouble with complex systems is that combinations of things tend to multiply together rather than adding up, so the number of possible states gets out of control very quickly. Even a Rubik’s Cube has more than 43 quintillion possible states; clearly a brain or an organisation has far more.

Trying to build things meant that everything had to be kept specific, but the theory of information couldn’t straightforwardly be applied to massively connected systems. You needed to find a way of describing things that was both rigorous and representative of reality. As economists will tell you, this isn’t an easy thing to do.

This law states that anything which aims to be a ‘regulator’ of a system needs to have at least as much variety as that system. In order to understand what that means, we need to get a short way into the theory of cybernetics. But for the meantime, for a sort of high-level metaphorical understanding, * think about how you steer various kinds of vehicles. A train can really only go forwards and backwards, so it only needs a single handle. A car can make turns, so its control system requires a steering wheel to represent the added dimension. And an aeroplane needs a joystick rather than a steering wheel, because it can make two kinds of turns. Let’s see where cyberneticians took this idea.

Which to be honest, is often all you really need in these contexts. The whole damn thing* in cybernetics is made of newspapers and high-level analogies. This isn’t computer science, where you can point to the states, bits and circuits on a piece of silicon.

Everything is driven by Ashby’s law of requisite variety: a given system has the potential to achieve stability only if every source of variability from the environment is matched by an equal or greater source of variety in the regulatory system.

If a manager or management team doesn’t have information-handling capacity at least as great as the complexity of the thing they’re in charge of, control is not possible and eventually, the system will become unregulated.

This sort of decision is fundamental to the cybernetic analysis of systems; you are always attenuating variety in some way or other unless you are describing a system that consists of everything in the universe.

Regulation by veto is a powerful way of solving huge and complicated search problems, because vetoing a situation as unacceptable is a very informationally ‘cheap’ action.

Further complications arise as you start to consider bigger and more realistic systems. For example, you usually need to respect the fact that the system is dynamic in time, and to ensure that the regulator can take in information faster than the system can generate variety, and respond to it quickly. You also need to build translation capacity into the system, so that every black box receives its inputs in a format that it can convert to outputs. In many large organisations, a significant proportion of the staff carry out actions that might be seen as the equivalent of taking a note on the vet’s prescription pad and ensuring that it affects what’s on the squirrel’s feeding tray.

Decision-making systems break down if the variety of their environment is not matched by the variety of their means of regulation. Implicitly, a control system is a model of its environment. If the model is missing important sources of variety in the environment, overestimating the variety of part of the regulatory system, or assuming that information is being transmitted when it isn’t, the system will gradually drift out of control without anyone necessarily understanding why.

5. Cybernetics Without Diagrams

Diagrams present you with the information ‘all at once’ and leave you to work out the flow of cause and effect for yourself, while a verbal explanation usually presents you with the story of cause and effect and leaves you to remember the connections. In the context of systems, where feedback is ubiquitous, the relationships are vital and the flow of cause and effect has no obvious start or end point, it's not hard to see why people draw diagrams. But this emphasis on connections means that diagrams are often ineffective ways of explaining something for the first time; they give you a network of relationships, but in a context that doesn’t tell you much about what the things are which the relationships are holding between.

So, I’m not going to draw a single box or arrow; I’m just going to explain, in words, the main elements of Stafford Beer’s viable system model. I’m going to leave out as much detail as I can; if you’re really interested in the system, there’s no substitute for Brain of the Firm and The Heart of Enterprise.

The first part of the model is System I, or operations, the bit of the organisation that is involved in making change in the real world. It’s a part of the system that could, in principle, be turned into an independent organisation.

Systems that aim to prevent this sort of thing happening by enforcing rules for sharing and scheduling are regulatory: System 2.

Coordinating the activities of a group of different System I operations is the first real management function in Stafford Beer’s model, and it’s called System 3 — optimization or integration. This part of the system directs the management of each individual operation in order to coordinate their activities towards a particular purpose.

System 3 is where you start to find *management jobs — those that are entirely devoted to communication and administration within the organisation. It’s also where system-level accountability is established; the key activity of the integration and optimization function is to agree the resource bargains with the System I units, and to ensure that they are being kept. And consequently, it’s the first place you might start to look if you think that unaccountability is creeping into the system. We’ll be looking at this a lot, later.

It’s easy to get confused between systems 2 and 3. Both of them look like they’re doing the same thing, in making the operations accountable to one another and to the wider organisation. The difference is that System 2 is all about preventing clashes and managing conflicts, while System 3 is concerned with achieving a purpose. On the ground, a useful way of drawing the distinction is to look at the management functions which everyone agrees to be necessary as opposed to those that they complain about.

*Stafford Beer refers to the part of the system responsible for internal organisation as the ‘metasystem’, one of very, very many newly coined jargon words which I am going to attempt to shield you from.

It’s a concept that’s recognisable to anyone who has hung around consultants or business school academics as ‘management by exception’. That’s the commonsensical principle that as much as possible, people ought to be given tasks to do and left to achieve them. An ‘exception’, in the jargon, is an ‘exception to the rule’ — something the business unit comes across that it can’t deal with and so ‘escalates’ to the level above. In a company or organisation run on these principles, the job of management consists mainly in dealing with these sorts of unusual cases, and then occasionally revising the objectives handed down to the level below.

If you believe Stafford Beer, there’s a pretty strong result here in terms of battles between competing management philosophies. Management by exception is not just common sense; it’s the right thing to do, objectively.

If the System 3 middle management tries to fight the variety generated by the System I operations single-handed, it’s going to be overwhelmed. System 3 is a central coordinating function without the resources to handle that volume of information. Everyone who’s worked in a dysfunctional organisation will recognise this as ‘micromanagement’ and will be familiar with the ways in which it breaks down. The environmental variety coming in from the suppliers and customers doesn’t get handled at the right level because System I employees aren’t allowed to make decisions. Meanwhile, the coordination, communication, integration and planning functions are neglected because System 3 is spending all its time trying to do someone else’s job. Middle management becomes bloated and overstaffed as it tries to add variety to itself, while the operations are miserable because of constant interference from people who don’t really understand what they do.

On the other hand, the central functions can be neglected by taking the opposite approach — that of excessive delegation. If everything is left to the operating-level management, there is no coordination of the resource bargains and no planning. With no adequate System 3 function, the operating units have no means to resolve their conflicts other than through internal politics, with resources allocated by grabbing and hoarding them. Nobody talks to anyone else, and the System I level becomes bloated with ‘operational’ managers who spend all their time treading on each other’s toes and fighting turf wars.

Plenty of organisations have no formally identified central planning department, but the integration and optimisation function is performed by an informal network of System I managers. That’s perfectly possible, as long as they have made the mental leap understanding that from time to time they need to adjust their thinking to perform a coordination role for the benefit of the organisation. Stafford Beer occasionally seemed to suggest that this kind of informal internal networking could be the best way to create System 3, which was why a big lounge at head office with whisky and cigars was important.

At the front is a conductor — and here we can see why Stafford Beer often said it was difficult to separate the operational systems. The conductor has one definite System 2 role; to mark the beat so everyone plays in time. That’s clearly regulatory in nature, but the conductor isn’t just a human metronome. They’re responsible for the dynamics of the piece, directing instruments to play softer or louder at the appropriate points, and to change the tempo according to the needs of the piece. System 3 integration and management role of the conductor is to provide the feedback to the musicians about the noise they’re making and how it needs to adapt.

So now we move on to the higher functions. I picked the example of an orchestra for this part of the explanation because there’s a clear boundary between decisions made during its performances (the operations) and other kinds of decision. Conducting an orchestra in a piece of music is a particular job, but a lot of other decisions need to be made to provide the context for the performance. As well as the ‘here-and-now’ functions which ensure that the music is played, the orchestra needs to be able to make higher-level choices — picking the repertoire and deciding where and when the performances will happen. The conductor might be involved in this, or there might be an artistic director or a tour manager — the point is that the function needs to exist.

This is System 4 of the model, often described as the intelligence function. Its defining characteristic is that while System 3 manages things happening ‘here-and-now’, System 4 is responsible for ‘there-and-then’. It is meant to be dealing with information from those parts of the environment that aren’t in direct contact with System I, and so which are capable of generating shocks that can’t be handled by the ‘management by exception’ system. The usual reason why this part of the environment isn’t detected by the operations is that it doesn’t exist yet; the key job of System 4 is to make sure that the resource bargains System 3 strikes with operations will remain feasible following anticipated future structural changes. *

It’s also important to note again that systems of this sort don’t necessarily match up to organisation charts, and that individuals can appear in different roles in different functions and contexts. For example, when on stage the piano player is part of System I operations, and will follow the conductor’s instructions. If, however, the orchestra is accompanying Elton John, the piano player will also be involved in higher levels of management and decision-making — because the piano player is Elton John!

*What about coping with unanticipated structural changes? We’ll get there in a minute, but the honest answer might be ‘sometimes they don’t’. A viable system is one capable of surviving indefinitely and of adapting to unanticipated shocks — which doesn’t mean that it can adapt to everything.

Matching ‘here and now’ with ‘there and then’

This brings us to the highest conceptual level of the system. We’ve emphasised at every stage that management is about variety engineering: making sure that every management function is matched, in terms of its information-handling capability, to the kinds of shocks and variety that might affect it. So System 3 has to be matched to the variety it manages (the exceptions and escalations coming out of System I), and System 4 has to be matched to its own source of variety (the uncertainty about the future and things happening outside the immediate environment).

There are two obvious failure modes here — changing too much, and changing too little. If the orchestra never changes its programme it will stagnate, but if it changes too often the musicians will be under-rehearsed and quality will suffer. The Variety transmitted from the intelligence function needs to be matched both to the change it anticipates in the environment and to the capacity of the operational function to reorganise itself.

System 5 is what Stafford Beer calls philosophy or identity. That might sound like an odd name to give the job of variety engineering to balance two competing management systems, but he does this because there’s a specific technique that’s best suited to this task, and ‘identity’ or ‘self-creation’* is a good name for it.

It matters what kind of an orchestra this is. Having a consistent identity is a great way of reducing the variety you need to deal with, because it means that there are a lot of possibilities that can simply be ignored.

Beer uses the Greek Word autopoiesis *but, really, there is no need.

In fact, understanding that identity, philosophy and purpose are tools of information management us the key to understanding the most famous slogan of management cybernetics, ‘POSIWID’ — or the ‘purpose of a system is what it does’. The identity-creating function is intrinsically linked to the variety-balancing function. In working to balance the immediate needs of the system with its response to a changing environment, System 5 is making the decisions which determine ‘what it does’ and, consequently, its purpose. POSIWID is not just a glib piece of cynicism; it’s a description of how a system retains viability and identity. It’s also, we’ll see later, the key to understanding the polycrisis of the last two decades — the failures we see in operations and management of our society have their roots in specific questions of philosophy and ideology.

The ability to translate information into action is the last piece of the puzzle, but it might have been the first piece because it’s so crucial.

Stafford Beer takes an approach to this problem that helps to mark the distinction between information theory and cybernetics. That solution is to say that information and action are one and the same; variety coming in from the environment, or being transferred from one system to another, only counts as ‘information’ if it has a causal role in decision-making. Otherwise it’s just ‘data’ — collections of facts that hang around on disk drives, waiting to be erased* or for the format to become obsolete.

*There are a lot of people is Silicon Valley who might do well to consider how much money they have invested in ‘data’ without bearing this distinction in mind.

For a piece of data to have the capacity to affect decisions, it has to arrive in time and in the right form.

The musicians also need to understand the conductor’s hand gestures and body language, so that they know when to play louder or softer. They also need to build up a memory of how the piece sounds, so that they are aware of what to expect. This building up of communication, transmission and translation capability happens in the rehearsal room; the key purpose of most forms of management training is to establish the channels and translation systems, so that people are able to handle variety in the real-life performance of their tasks.

The overlapping of different systems — and the tendency of individuals to have different roles at different levels of abstraction — is a key part of Beer’s theory, and one of the main reasons why his diagrams got so complicated. He claims that every ‘viable system’ needs to have all five of the functions described so far in order to be capable of long-term survival, but that every such system can also be seen as System I within a larger system. Similarly, since we defined System I as part of an organisation that could in principle be a viable separate organisation, the internal management of System I needs to have its own equivalents to systems 2, 3, 4, and 5; it needs internal regulation, optimisation and intelligence, and a balancing, identity-preserving function of its own.

Often, when you’re trying to diagnose why a system is failing, you need to consider both the larger system in which it’s embedded and the organisation within its operations. A great source of management problems, for example, is that organisations often fail to identify some of their operations as distinct systems, and so they lack their own internal ‘higher functions’. A division of this sort will generally be a ‘problem child’; unable to absorb its own environmental variability, it will bounce from crisis to crisis, taking up disproportionate time and effort on the part of the middle managers to which it has been assigned.

There is a solution to this kind of problem in the viable system model — Beer emphasises it in his later work, particularly in the parts of the second edition of Brain of the Firm that were written after his experience in Chile. A truly viable system needs to have communication channels that link the operations to higher-level management functions and even to higher levels of recursion. Stafford Beer called these ‘algedonic signals’ — a kind of neurological metaphor, coined from the Greek words for pain and pleasure.

It might be easier, though, to understanding them as the kind of messages sent by the red handle in a train driver’s cab. As well as a physical check on the individual vehicle, the emergency brake sends an organisational message to the whole railway, informing it that a piece of track can’t be used.

Red-handle signals are problematic for organisations. If they’re being generated frequently, this is an indicator of missing, misidentified or dysfunctional systems further down the organisation. And by their nature, they go from one level of management to another, bypassing the usual systems in between.

• There are five core functions, and if any of them are missing or under-resourced, the flow of information won’t be balanced with the capacity to process it.

• Information only counts if it’s being delivered in a form in which it can be translated into action, and this means that it needs to arrive quickly enough.

• Systems preserve their viability by dealing with problems as much as possible at the same level at which they arrive, but they also need to have communication channels that cross multiple levels of management, to deal with big shocks that require immediate change.

But there was hardly any coordinated policy reaction until the bankruptcy of Lehman Brothers on 15 September 2008.

This ought to immediately alert us — the problem was in System 4, the ‘there and then’ intelligence function. Absence or weakness in this system is one of the most common problems in organisations, and central banks have a few organisational quirks that make them particularly vulnerable.

It looks like that’s what central banks do: make forecasts of the future, publish them and update them as the situation changes. But forecasts are their version of ‘here and now’; when you hear them talking about inflation targeting, they mean inflation forecast targeting. Interest rate changes adjust the economy with a lag, so they need to be implemented on the basis of the expected path over a period of time — not just because of what’s happening on the evening news.

If something isn’t in the forecast, it isn’t part of the information set and can’t affect decisions — which is why central banks try to supplement their economic and statistical models with other data sources. But there’s a more subtle problem — if you’re doing all this forecasting, it’s hard to believe anyone who tells you that you’re not looking to the future.

A real System 4, though, is explicitly concentrated on those parts of the environment that aren’t yet relevant to what it’s doing. This capability was weak in the central banks; they were not looking for things that might have upset their policy-making framework. The information was there, but it hadn’t been organised into the decision-making process and didn’t shape the view at the management or operational levels. It remained as mere ‘data’ or was attenuated away by simply ignoring it: the ‘information-processing system of last resort’.

In the years when it looked like things were going well, the central banks developed a view of the world in which the changing structures of global finance weren’t part of their job. They didn’t pay attention to the debt bubble, and they had got rid of the communication channels that might have carried the red-alert warnings up to the highest levels of policy-making. More precisely, they had got rid of the translation systems. There was no shortage of people warning that there was a problem in 2006 and 2007, but none of these warnings was given in a form that could be recognised by the world’s central bank governors as requiring action. A failure to build System 4 and balance its variety against System 3 is itself a failure of System 5 — that’s the function that has the job of balancing ‘here and now’ against ‘there and then’. So, using the viable system model to diagnose the causes of the global financial crisis, we end up with a rather interesting conclusion. Where things went wrong was a matter of philosophy. The central banks had an identity-creating function, but it had failed; it defined their purpose in such a way that they failed to understand that particular kinds of information were relevant to them.

Intermission

If you cut the connection between a cat’s cerebellum and the rest of its brain, it still looks alive. It can walk, right itself when it falls over, and eat and drink when food and water are placed in front of it. It might even be able to groom and clean itself, but it is no longer capable of purposive action. The decerebrate cat will survive as long as its environment is compatible with this, but it can’t generate its own responses to change — it needs to be sheltered from the everyday world, using other resources. Beer made the point that like the cat, you will often see a ‘decerebrate organisation’. (His typical example was a university)

An organisation in this situation is one that has, for one reason or another, stopped paying attention to some kinds of information. It’s only aware of its immediate surroundings — this quarter’s revenue, the current staffing level, things like that — and has lost the ability to make plans. Like the cat, it can continue to survive as long as nothing major changes, but the next time it encounters a shock it will go into crisis.

Part 3: The Blind Spots

6. Economics and How It Got That Way

This has been all but eclipsed by another kind of economics. One of Friedrich Hayek’s contemporaries defined economics as ‘the science which studies human behaviour as a relationship between ends and scarce means which have alternative uses’. It’s this idea of a generalised study of human behaviour under conditions of scarcity that has been responsible for the economists’ intellectual imperialism. Because if you announce that you’re the experts on human beings when they have to choose between different priorities under conditions of scarce resources, well, when aren’t resources scarce? When don’t people have to choose between different things they want?

Because, as a decision maker considering a long-term plan, one of the basic things you need to consider is that all your competitors will be doing the same thing.

Here’s a story about past strategic decision: Nokia once made rubber out of dandelions. They were actually very good at it; a Finnish Air Force plane which crashed into a lake in Karelia during the war in 1942 was recovered in 1998 with its tyres still inflated. Because they made rubber, they made insulation for electric cables. Because of that, they got into the telecoms industry;* for a while they were the world’s biggest manufacturer of mobile phones and today they make complicated switching equipment.

*I confess — this is an oversimplification. Nokia did dozens of other things and it’s very hard to trace the timing and logic of its entry into various different industries. Basically the only principle I could extract from its corporate history is that if Finns need something, Nokia would usually have a go at making it.

This is how the cybernetic problems of the post-war industrial state have been shaped by economics. There are three major blind spots: the belief in markets as computing fabric, the flattening of time and uncertainty and the commitment to Ricardianism.

The belief in markets as computing fabric is such an obvious source of blind spots as to hardly require explanation. Faith in the power of free markets to solve problems is so great as to support the reverse inference of economists that if something appears to be a market equilibrium, then it must be an optimal solution.

7. If You’re So Rich, Why Aren’t You Smart?

Business hallucinations

Mechanising and computerising the accounting system allowed companies to grow and get more complicated, but on the basis of the same system, one based on financial reporting. This meant that the increase in complexity was masked from the people who were meant to be managing it. Their reports were largely unchanged. And consequently, companies began to hallucinate.

Often, a core skill of middle management is the ability to manipulate the financial reports to compensate for a set of numbers that aren’t giving the right answer. Everyone who has put together a business plan knows that if you can’t fudge the key assumptions to justify the decision your boss wants to make, you don’t know enough about the business.

There are two obvious ways to fail here. The translation and error-correction mechanism might be inadequate, or the managers might intentionally distort the signals in order to follow priorities of their own. The tragedy of senior management is that it can drift into either of these failure modes without realising; if either problem arises, it arises in their information and communication environment, so they won’t notice it. It’s the problem identified by Niccolò Machiavelli — a prince who is not wise cannot be well advised, and a manager who doesn’t have access to excess analytical capacity won’t be able to tell when something has gone wrong with their subordinates. But maintaining that spare management capacity is expensive.

Of course management theories change — when the foundations of the problem are constantly shifting, the answers are bound to change.

If you translate this problem into the more abstract language of cybernetics, it becomes easier to understand. In the business environment, complexity (environmental variety) will naturally increase. In any given organisational structure, the variety which management systems can bring to bear also increases, but more slowly.*

*If it was possible to measure variety in absolute terms for real-world.

And this story repeats itself through the history of management science; almost every classic of the literature seems to have described a way of adapting systems to a more complicated world, and then to have become obsolete itself. If you look past the slogans and think about what things like ‘management by objectives’, ‘focus on core competences’ and so on actually mean, they are all different ways of advising executives to restructure their businesses so that they don’t generate complexity faster than it can be managed.

When things go wrong with management consultancy, it’s more likely to be because the consultants are tackling a new problem and there isn’t anyone in the company who knows the answer. The consultants get commissioned because they advertise themselves as brains for hire; a company can cut overhead costs by having fewer middle managers performing staff functions in the ordinary course of business and buying in brain power when confronted with a difficult question. When you write this idea down in black and white, it’s pretty easy to see why it won’t work except by pure luck.

The reason is invariably a failure to respect the complexity of the problem. Management problems are complex, high-variety questions, Rubik’s Cubes rather than rows of blocks. In order to solve them, you need to make decisions about how to represent the problem in such a way that you can simplify it and solve it, without losing vital details that will blow your solution apart as soon as it’s implemented.

The generic solutions are not always bad. A lot of management problem are surprisingly generic: a missing communication channel or a failure to notice that the environment has changed. If the problem is understood correctly, the solution is usually quite simple — it’s likely to be a variation on the same theme of ‘the world has got more complicated and you need to take steps to reconcile that complexity with your capacity to manage it’ that has been the message of management science since Alfred D. Chandler.

But because the solutions are often simple, the work is surprisingly unpleasant. An effective consultant is likely to spend most of their time telling people obvious things that they don’t want to hear. That’s a difficult combination; while not particularly intellectually stimulating, it’s emotionally taxing. It’s not surprising that so many people find doing this intolerable, and consequently let their ethics slip. Telling your client what they want to hear is a better way to get repeat business; the problem won’t go away and the person commissioning the work will still like you.

Intermission

Meanwhile, in Chile

The great misconception about the Chilean experiment is that it was an attempt to apply computer technology to the task of organizing a command economy. This is a handy myth for those who want to use it as an example for whatever argument they’re trying to make. Neoliberals and apologists for Pinochet and Friedman can pretend that it was a quixotic project with no more hope of working than the Soviet system; socialists and other apologists can pretend that it nearly worked. Neither is the case, and by now it ought to be obvious why — management cybernetics demonstrates quite clearly why a centrally planned economy doesn’t respect the law of requisite variety and can’t get the information it needs to function.

*The publication of Eden Medina’s Cybernetic Revolutionaries, a very good history of the project, has helped in recent years.

Even in the early stages when Beer was first retained as a consultant, people had been warning the Allende government that it was nationalising things too fast, without regard for the capacity to manage them, and seemingly without much in the way of an overall industrial plan. They had introduced a system of ‘intervenors’ who sat in the factories and passed on orders from central government; this was unpopular, and contributed to the impression of central technocratic control rather than devolved autonomous production.

Part 4: What Happened Next?

1. Enter Friedman

Only a crisis — actual or perceived — produces real change. When that crisis occurs, the actions that are taken depend on the ideas that are lying around.

Milton Friedman, Capitalism and Freedom (3rd edn), 2002

In the early day of the post-war corporation, the question of purpose hardly arose. Companies like DuPont had been given an obvious reason to invest in the production of smokeless gunpowder by the Second World War, and at its end they found themselves in possession of a lot of capital equipment and chemical know-how. They went out looking for new things to do with cellulose and petrochemicals because it would have been strange not to. As they grew more complicated, they had to reorganise their corporate forms and management structures; academic writing in management was really just catching up with the things that engineers and accountants were inventing out of necessity.

J.K. Galbraith, possibly the only other economist of the 1970s with anything like Milton Friedman’s public profile, once said that the essence of leadership was ‘the willingness to confront unequivocally the major anxiety of their people in their time’.

This is the whole theory of the firm, according to the Friedman doctrine. There is no analysis of the company as a decision-making system, just individuals making decisions. While modern HR people might talk about ‘bringing your whole self to work’, Milton Friedman explicitly tells the (assumed to be male) modern executive that he has to divide his moral self into judgements ‘as a person in his own right’ and decisions ‘in his capacity as a businessman’.

The Friedman doctrine forestalls this cacophony of complexity by substituting an abstract ‘representative shareholder’. Rather than thinking about actual human beings working in association, we’re invited to replace them with a black box that only cares about profits, the pretend that we can enter into the same sorts of relationships with that black box as we could with the individuals.

This is what makes the Friedman doctrine, as far as I can see, a lie; it’s an exhortation to executives to reverse the truth. They are meant to ignore the reality of the company and act as if they are directly employed by human beings, but then ignore the reality of human beings, and act as if they are employed by a theoretical construct.

It’s an attractive lie, though, thanks to the combination of the accountability sink with the shift in perspective. The great anxiety of the managerial class was that they were losing their individuality as corporations became more complex, but that they were still subject to criticism. The Friedman doctrine invited them to disassociate themselves from their roles; to attribute all the bad consequences and all the frustrating lack of independence to a separate work-self, which was under an obligation to a simple principle.

But they key move was the first one made, and it’s no wonder that it caught on. It was so terribly attractive to, it solved so many problems for and soothed the anxieties of, so many important people.

The intellectual backing for the leveraged buyout movement was explicitly disciplinary, and the use of debt as both as technology of control and as a way of serving the interests of capital-owners is clear as the moral of the case studies. Companies would pay out huge dividends and take on self-consciously risky Antoine’s of debt in order to ‘create a sense of urgency’ among their management, or to communicate management’s confidence that their accounting policies weren’t as aggressive as they looked.

Business schools and consultancies as we mentioned earlier, are a key part of the internal communication system of management, and journals like this are how that communication system aligns itself. The JACF and its ideological peers were read by the people who went into the consultancies, banks and investment firms — they all shared the same language and the same assumptions.

That’s part of what made them so fearsome and effective, like any well-drilled cadre. Across the economy, corporate managers were left pleading about their history and community, the trust of their customers, the value of their technology. And time after time, they found that unless it could be quantified in an accounting system and placed into exactly the right kind of spreadsheet, this kind of information no longer mattered. So they borrowed more money, paid bigger dividends and hoped.

From a cybernetic point of view, it’s interesting as an example of how the systems and structures mattered so much more than the individuals involved. The development of the Friedman doctrine into the intellectual backing for the leveraged buyout boom and the private equity industry are best seen as a conflict between two comprehensive systems of interest, both of which might have regarded the other as a threat. The great unremarked class struggle that happened in the 1970s and 1980s was that between capitalism and managerialism.

The managers lost this struggle, pretty comprehensively. And as we’ve seen, the combination of the blind spots in management and the blind spots in economics came together to produce an ideology which was bound to remove management capacity. And that created further blind spots, and further reduced the system’s ability to cope with shocks. The story of how we got to where we are is a story of the attempts of the system to cope with this, and to search for short-term equilibrium.

1. The Morbid Symptoms

The crisis consists precisely in the fact that the old is dying and the new cannot be born; in this interregnum a great variety of morbid symptoms appear.

Antonio Gramsci, Prison Notebooks,

Trans. Quintin Hoare, 1971

I wanted to write a sort of cybernetic political thriller, but it didn’t quite work out that way. It seems that in order to get to the point where the story begins, you need to write eight chapters explaining the construction of the murder weapon. But here we are. To recapitulate, the basic problem is that systems in general need to have mechanisms to recognise themselves when the complexity of their environment gets too much to bear. But the high-level governing systems of the industrial world — economic policy and business management — had some defects and blind spots which prevented this from happening. How did that turn into an economic crisis, and how did it mature into an ongoing political polycrisis? How did we get from the Friedman doctrine and leveraged buyouts to Trump, Brexit and populism?

The first part of the story involves expanding on the conclusion of the last chapter. A key part of recent history has been a story of class struggle, but not the usual kind. There’s been a class war between the capital-owning class and the managerial class.

We’ve already seen how accounting systems can act as an information-reducing filter. It follows from this that the greater the emphasis placed on accounting-based targets set by the CEO, the greater the filtering effect.

If you consistently demand the impossible, you will inevitably get the unethical.

Outsourcing is a contractual relationship, and contracts (like debts) are typically all-or-nothing affairs. Either a requirement has been fulfilled or it hasn’t. The typical outsourcing contract narrows the bandwidth of the communication channel to either ‘everything is going more or less as anticipated’ or ‘it’s stopped working and we need to find out why’. If it’s the first of these two cases, everything’s fine. If it’s the second, the continued stability of the system will depend on how much information-handling capacity can be brought to bear to address whatever problem has arisen.

Since the outsourcing communication channel is designed to spend most of its time transmitting ‘everything’s OK’, it’s difficult to guess how much additional bandwidth needs to be allocated to carry messages like ‘but the following conditions are changing which might affect things in the near future’ — let alone how much might need to be allocated at short notice when it starts to say ‘everything’s no longer OK’. If you have targets to make, it will always be tempting to cut out a bit of spare capacity.

And this is of course what happened; in an accounting system which targeted overhead costs, combined with an emphasis on generating cash, it looked like a no-brainer to thin out the ranks of middle managers who didn’t appear to do anything. But unfortunately, a ‘no-brainer’ was exactly what it turned out to be. Over the course of Jack Welch’s career, the industrial world’s productive system — the corporations — set about the equivalent of amateur brain surgery, hacking away bits of their regulatory and information-handling system, to see if they could do without them.

You used to be able to buy a copy of the book Jerome wrote in 1943, Economics Is an Exact Science, from them; I got mine in about 2002.

In the introduction to that book, Levy sets out his view of the purpose of capitalism:

The working class is the original and fundamentalist economic class…The function of the investing class is to serve the members of the working class by insuring them against loss and by providing them with desired goods. The justification for the existence of the investing class is the service it renders the working class, measured in terms of wages and desired goods. The contrary is not true. The working class does not exist to serve the investing class. The working class has the right to insure itself through organizations composed of its members or through government, thereby eliminating the investing class.

Although unemployment is, in one sense, a waste, the alternative possibility is for managers and capital-owners to lose control of the wage bargain. For that reason, the capitalist class is often prepared to tolerate lower output as the price of retaining control of wages and prices. The secret to the Great Moderation of the neoliberal era — the sharp decline in the volatility of output and inflation — was that economic policy began to use the unemployment rate as a variety attenuator; external shocks could be accommodated or passed on, so long as wage pressure and price inflation did not build up.

Governments were also subject to the same cognitive pressures of a world that was becoming more complicated, so they also needed solutions to attenuate information.

The imposition of Friedmanism and the debt burden in the corporate sector gave a clear signal about the new priorities, but the equivalent processes in the public sector just added a lot of noise. The system was calling for ‘reform’ and revolution, but not giving any clues as what specifically ought to be done.

And so the progress of outsourcing and decerebration ended up being even more intense, and the public sector became as big a fee pool as the private sector for the consultancy industry. This had the effect of closing off the other possible way in which the working class might have insured themselves against volatility — through the policies of governments that they elected.

This had a frightful effect on public sector management. The coordination function was impaired; the difficulty of ‘joined-up government’ and making policy for problems that crossed the boundaries of different agencies was repeatedly remarked on. And the operational delivery functions started to suffer severe cognitive loss, too. A company that sells goods and services for profit can never completely sever the connection which takes information from its customers; the people who buy the thing have the ability to refuse to do so. In many cases, people who interact with the state don’t even have the ability to transmit that single bit of information because they can’t shop elsewhere; they can complain if they like, but they interact with the sermon representative, the paradigmatic accountability sink.

Things got worse over a long period of time, but this was initially hard to notice. Recall that in Jerome Levy’s high-level view of the economy, the investing class has two purposes — providing insurance against the business cycle to the working class, and providing them with consumer goods. While the first of these services had been abandoned, this was not immediately obvious — the business cycle itself had been temporarily calmed down. And although many of the purchases were funded by debt, the second still seemed to be functioning. Over the course of a few decades, the risk transfer was completed.

In general, hardly any effort is expended on considering what kinds of communication channels should be maintained to allow the population to express its views to the government, or at least, not from an information engineering point of view. It seems quite clear that different arrangements might have different characteristics — a proportional election system should be capable of carrying slightly more information than a first-past-the-post one, a monthly opinion poll has a shorter lag than an annual one, and so on, but this isn’t how they’re thought of; elections are simply horse races with executive power as the prize, and opinion polls are rarely used as more than a sort of racing form to predict the winners of the next race. The channels seem to be designed to carry very few bits of information.

The only kind of communication that such a constrained channel can carry is a scream: the signal passes through the levels of control and announces that something has gone wrong which threatens the integrity of the system itself. This is why there was a family resemblance between the ‘populist’ movements that sprang up in the 2010s. Narendra Modi in India, Beppe Grillo in Italy, Donald Trump in America, Nigel Farage in Britain or Recep Tayyip Erdoğan in Turkey instinctively realised that they were on the same side; each of them, in their own culturally specific context, was acting as a communication channel for a population which wanted to convey a single bit of information: the message that translates as, ‘HELP! THE CURRENT STATE OF AFFAIRS IS INTOLERABLE TO ME.

One notable thing that happened during the early months of the Covid-19 pandemic is that the suicide rate in England and the USA went down, despite the fact that many people had confidently predicted that the opposite would happen. After all, the rate had been rising for many years, and there were plenty of reasons to believe that an event which worsened many of the normal indicia of suicide — economic loss, loneliness, mental health problems — would make things worse. But somehow, although the virus and the lockdown were awful for many, they weren’t quite as unbearable as everyday life over the previous decade.

You can’t say how much information a human being is taking in and reacting to at any given time, but you can easily observe the difference between a human being that is coping and one that is overloaded. That’s my diagnosis of what led to the series of connected political eruptions between the financial crisis and the pandemic. The hypothesis set out as early as 1970 by Alvin Toffler in his book Future Shock turned out to be correct: the number of people who were no longer able to cope with the modern world reached a critical mass.

Two of the longest-running pieces of research in medicine are the Whitehall studies, the first of which began before Stafford Beer ever set foot in Chile. Whitehall I, from 1967 to 1977, recorded the general health outcomes of 17,500 male British civil servants; Whitehall II, ongoing since 1985, widened the remit to include women; both cohorts have been followed up and re-examined over the years.

Marmot ended up concluding that the psychic feeling of being in control of your life is extremely important as a source of well-being, and that conversely, being out of control is physiologically harmful as wells s emotionally intolerable.

At various points in this book, we’ve noted that you can tell when a cybernetic system is overloaded because it breaks down. Marmot’s main conclusion from his research was that inequality in society was a major driver of public health risks, but it could be givens cybernetic interpretation too. The connection that he found looks like the result of a variety mismatch; people are, increasingly, unable to regulate the input from their immediate environment, and they correctly perceive this as a threat to health and life.

I think this is what explains the thread between MAGA, Brexit, the Five Star Movement, Hindutva and all the rest. The populist movements of the 2010s all promised a simpler world; they were, in the words of J. K. Galbraith, taking on the great anxiety of their people and addressing it. They were also promising to restore the broken communication channels — to make voices heard, to force the managerial class to listen.

1. What Is to Be Done?

Should we all stand by complaining, and wait for someone malevolent to take it over and enslave us? An electronic mafia lurks around that corner.

Stafford Beer, Designing Freedom, 1974

Record producers have a saying: ‘If you can hear the problem, you can hear the solution.' Identifying what’s wrong with a mix is the same thing as identifying what needs to be changed. Variety engineering is, unfortunately, not quite the same. Identifying which channels are broken, which systems are missing or which translation mechanisms don’t work is a significant step forward, but there isn’t a big desk in front of you with a slider that will add more capacity wherever it’s needed.

There’s one sense in which the recording engineers’ maxim is more applicable, though. These are problems of human organisation, and the overwhelming majority of environmental variety of created by humans. The knowledge that something is an organisational problem means that the source of its complexity is the human beings in the organisation. That, in turn, means that the human beings in the organisation have sufficient variety and capability to match the complexity of the problem; the problem can’t be bigger than them, because their own misdirected efforts created it. Only problems that aren’t cybernetic can genuinely be insoluble.*

*One implication of this proposition is that it gives us a rigorous deduction of what would constitute environmental Armageddon — it’s the singularity point at which the problem of climate change ceases to be soluble purely by changing human behaviour and organisation.

This is the paradigm shift that might be required — that organisations and systems can be like people, having purposes without a single goal. An artist doesn’t have a successful career by maximising their art; they do it by repeatedly producing work that they are proud of.

That’s what the world could look like if we got rid of the blind spots. Business ought to be like artists, not paperclip maximisers. The economic concept of optimisation, and the institutions of management and government which enforce its use, effectively act as a brutal information reducing filter. By taking away the pressure to maximise a single metric (and therefore to throw away information that doesn’t relate to it), organisations could apply their decision-making capabilities much more effectively. They could innovate more, design more sustainable solutions and build less adversarial, longer-term relationships with their people.

There are a number of models, most of them ignored for decades, in which the corporate sector provides a stabilising function, insuring the working class against fluctuations in the business cycle, rather than expecting them to soak up the volatility.

The intriguing thing is that Simon and Galbraith didn’t write polemics to the effect that this was how corporations should behave — they just described what was in front of them at the time. Before Milton Friedman’s essay, lots of people assumed that this was just naturally the way things would tend. Without the Friedman fiction by, without very great re-engineering of the systems of corporate finance, the industrial economy might have just gone on and developed into a technostructure.

Maybe they were right? It would certainly be good if they were, because that might indicate a much easier path to defuse the immediate source of crisis. If the problem with the modern corporation is the result of the capitalist counter-revolution against the managerial class, we just need to change the terms of the battle.

Dismantling the leveraged buyout industry would get rid of an overhanging threat across the entire managerial class; it would open up a huge space for different models of corporate governance.

But how would you go about doing that? If someone thinks that they could run a company better than the existing management, they’re allowed to spend their money buying it; preventing that from happening would involve a great disruption to the system. You can’t stop the management of a company from taking on debt, either. A lot of the time, taking on debt — even risking bankruptcy — is the correct and necessary thing for a company to do. As soon as you start trying to design regulatory regimes to prevent ‘excessive’ corporate borrowing, or distinguishing between companies on the basis of their owners, you quickly start to realise that this is a case where the law of requisite variety is telling you the project is doomed* —the variety of financial situations that a company can be in is much greater than anything that could be written into a rule.

*Buried here, in this footnote towards the end of the book, is probably the most useful and valuable piece of advice in all its pages: checking whether Ashby’s law of requisite variety has been respected is a great way of spotting a doomed project. It will even tell you if the project is going to fail because of insufficient resources or fail because it’s an impossible thing to achieve.

Corporations are decision-making systems, not ‘intelligences’. They have homeostatic forces which aim to maintain their equilibrium, and higher-order decision-making systems which mean they are able to reorganise themselves in order to respond to shocks beyond the scope of anything anticipated when they were designed. To attribute motivation to them is to make assumptions about the internal workings of the black box — the original intellectual sin of cybernetic analysis.

One of its signals has been so amplified that it drowns out the others. The ‘profit motive’ isn’t something that can be ascribed to corporations — they don’t have motives. What they have is an imbalance between the two key higher-functions — here-and-now versus there-and-then. They aren’t capable of responding to signals from the long-term planning and intelligence function, because the short-term planning function has to operate under the constraints of the financial market disciplinary system. Either a corporation has a survival condition based on needing to make a monthly interest bill, or there’s an implicit threat from the financial environment that if it fails to behave in a particular way, it will be taken over by an outside entity.

If you take away that pressure, it’s quite likely that the natural equilibrium of corporate decision-making systems will be less hostile to human life. Viable systems fundamentally seek stability, not maximisation.

A maximising system involves defining an objective function, and throwing away all the other information. Sooner or later, the environment is going to change, and something which isn’t in the information set any more is going to lead the system into destruction.

Consequently,

Every decision-making system set up as a maximiser needs to have a higher-

Level system watching over it.

If every maximising system has to have a higher-level system governing it (to make sure it doesn’t go bonkers), then that logically implies that the top level of any decision-making system that’s meant to operate autonomously can’t be a maximiser. And so, the governing philosophy of the overall economic system can’t be based on the constrained optimisation methodology that’s currently dominant in the subject of economics. Otherwise there’s a risk that the system will go bonkers, and that it will start pushing maximising objectives, oblivious to the danger that it’s on course for making human life impossible. Like it actually has done.

Management cybernetics doesn’t give any clues as to how such a profound social change might be achieved, unfortunately. I could make something up, but it would feel like a shabby way to treat you after we’ve come so far together.* My only guess is that it might be that what’s really intolerable about unaccountability is the broken feedback link, and that if we can solve the problem of communicating with the system — pay more attention to the ‘red-handle alert’ mechanisms that indicate an unbearable outcome — people might not be so furious about the death of personal responsibility.

*Particularly Chapter 5, which I promise was a lot harder to write than it was to read.

Further Reading

Two collections of shorter pieces and lectures are How Many Grapes Went into the Wine, edited by Roger Harnden and Allenna Leonard (John Wiley, 1994) and Think Before You Think, edited by David Whittaker (Wavestone Press, 2009). These are probably the best place to start, because you can dip into them without than needing to clear time to hold the whole structure in your head, and you’ll almost always find something amusing and insightful when you do.

Micheal Jackson’s Creative Problem Solving (John Wiley, 1991) is probably the one read if you really fancy having a go at applying management cybernetics to a real-world consulting assignment, although many friends swear by The Fractal Organization by Patrick Hoverstadt (John Wiley, 2011). Of course, Eden Medina’s Cybernetic Revolutionaries (MIT Press, 2011) is the definitive account of what really happened in Chile.

Among many works detailing the state of management science, I’d recommend Management Studies in Crisis: Fraud, Deception and Meaningless Research by Dennis Tourish (Cambridge University Press, 2019) and Nothing Succeeds like Failure: The Sad History of American Business Schools by Steven Conn (Cornell University Press, 2019). It is a bit of an indictment of things that H. Thomas Johnson and Robert S. Kaplan’s Relevance Lost: The Rise and Fall of Management Accounting (Harvard Business School Press, 1987) is still itself every bit as relevant today as when it was published.

The Big Con by Marian Mazzucato and Rosie Collington (Allen Lane, 2023) came out just after I had finished writing about the decerebration of the public sector and covers that topic thoroughly and excitingly.

Flying Blind: The 737 MAX Tragedy and the Fall of Boeing by Peter Robinson (Penguin Business, 2021) is also an enthralling history of a massive accountability sink. I learned a huge amount from Gill Kernick’s Catastrophe and Systematic Change: Learning from Grenfell (London Publishing Partnership, 2021) about another case study in which interlocking systems combined to create a tragedy that couldn’t systematically be pinned on any single organisation, let alone an individual.

If you’re only familiar with the modern form of capitalism and management, it’s very worth reading As I See It (Prentice-Hall, 1976), the autobiography of J. Paul Getty, to see that things really did used to be different.