by Andy Beckett / 8 September 2003
During the early 70s, in the wealthy commuter backwater of West Byfleet in Surrey, a small but rather remarkable experiment took place. In the potting shed of a house called Firkins, a teenager named Simon Beer, using bits of radios and pieces of pink and green cardboard, built a series of electrical meters for measuring public opinion. His concept – users of his meters would turn a dial to indicate how happy or unhappy they were with any political proposal – was strange and ambitious enough. And it worked. Yet what was even more jolting was his intended market: not Britain, but Chile.
Unlike West Byfleet, Chile was in revolutionary ferment. In the capital Santiago, the beleaguered but radical marxist government of Salvador Allende, hungry for innovations of all kinds, was employing Simon Beer’s father, Stafford, to conduct a much larger technological experiment of which the meters were only a part. This was known as Project Cybersyn, and nothing like it had been tried before, or has been tried since.
Stafford Beer attempted, in his words, to “implant” an electronic “nervous system” in Chilean society. Voters, workplaces and the government were to be linked together by a new, interactive national communications network, which would transform their relationship into something profoundly more equal and responsive than before – a sort of socialist internet, decades ahead of its time.
When the Allende administration was deposed in a military coup, the 30th anniversary of which falls this Thursday, exactly how far Beer and his British and Chilean collaborators had got in constructing their hi-tech utopia was soon forgotten. In the many histories of the endlessly debated, frequently mythologised Allende period, Project Cybersyn hardly gets a footnote. Yet the personalities involved, the amount they achieved, the scheme’s optimism and ambition and perhaps, in the end, its impracticality, contain important truths about the most tantalising leftwing government of the late 20th century.
Stafford Beer, who died last year, was a restless and idealistic British adventurer who had long been drawn to Chile. Part scientist, part management guru, part social and political theorist, he had grown rich but increasingly frustrated in Britain during the 50s and 60s. His ideas about the similarities between biological and man-made systems, most famously expressed in his later book, The Brain of the Firm, made him an in-demand consultant with British businesses and politicians. Yet these clients did not adopt the solutions he recommended as often as he would have liked, so Beer began taking more contracts abroad.
In the early 60s, his company did some work for the Chilean railways. Beer did not go there himself, but one of the Chileans involved, an engineering student called Fernando Flores, began reading Beer’s books and was captivated by their originality and energy. By the time the Allende government was elected in 1970, a group of Beer disciples had formed in Chile. Flores became a minister in the new administration, with responsibility for nationalising great swathes of industry. As in many areas, the Allende government wanted to do things differently from traditional marxist regimes. “I was very much against the Soviet model of centralisation,” says Raul Espejo, one of Flores’s senior advisers and another Beer disciple. “My gut feeling was that it was unviable.”
But how should the Chilean economy be run instead? By 1971, the initial euphoria of Allende’s democratic, non-authoritarian revolution was beginning to fade; Flores and Espejo realised that their ministry had acquired a disorganised empire of mines and factories, some occupied by their employees, others still controlled by their original managers, few of them operating with complete efficiency. In July, they wrote to Beer for help.
They knew that he had leftwing sympathies, but also that he was very busy. “Our expectation was to hire someone from his team,” says Espejo. But after getting the letter, Beer quickly grew fascinated by the Chilean situation. He decided to drop his other contracts and fly there. In West Byfleet, the reaction was mixed: “We thought, ‘Stafford’s going mad again,’ ” says Simon Beer.
When Stafford arrived in Santiago, the Chileans were more impressed. “He was huge,” Espejo remembers, “and extraordinarily exuberant. From every pore of his skin you knew he was thinking big.” Beer asked for a daily fee of $500 – less than he usually charged, but an enormous sum for a government being starved of US dollars by its enemies in Washington – and a constant supply of chocolate, wine and cigars.
For the next two years, as subordinates searched for these amid the food shortages, and the local press compared him to Orson Welles and Socrates, Beer worked in Chile in frenetic bursts, returning every few months to England, where a British team was also labouring over Cybersyn. What this collaboration produced was startling: a new communications system reaching the whole spindly length of Chile, from the deserts of the north to the icy grasslands of the south, carrying daily information about the output of individual factories, about the flow of important raw materials, about rates of absenteeism and other economic problems.
Until now, obtaining and processing such valuable information – even in richer, more stable countries – had taken governments at least six months. But Project Cybersyn found ways round the technical obstacles. In a forgotten warehouse, 500 telex machines were discovered which had been bought by the previous Chilean government but left unused because nobody knew what to do with them. These were distributed to factories, and linked to two control rooms in Santiago. There a small staff gathered the economic statistics as they arrived, officially at five o’clock every afternoon, and boiled them down using a single precious computer into a briefing that was dropped off daily at La Moneda, the presidential palace.
Allende himself was enthusiastic about the scheme. Beer explained it to him on scraps of paper. Allende had once been a doctor and, Beer felt, instinctively understood his notions about the biological characteristics of networks and institutions. Just as significantly, the two men shared a belief that Cybersyn was not about the government spying on and controlling people. On the contrary, it was hoped that the system would allow workers to manage, or at least take part in the management of their workplaces, and that the daily exchange of information between the shop floor and Santiago would create trust and genuine cooperation – and the combination of individual freedom and collective achievement that had always been the political holy grail for many leftwing thinkers.
It did not always work out like that. “Some people I’ve talked to,” says Eden Miller, an American who is writing a PhD thesis partly about Cybersyn, “said it was like pulling teeth getting the factories to send these statistics.” In the feverish Chile of 1972 and 1973, with its shortages and strikes and jostling government initiatives, there were often other priorities. And often the workers were not willing or able to run their plants: “The people Beer’s scientists dealt with,” says Miller, “were primarily management.”
But there were successes. In many factories, Espejo says, “Workers started to allocate a space on their own shop floor to have the same kind of graphics that we had in Santiago.” Factories used their telexes to send requests and complaints back to the government, as well as vice versa. And in October 1972, when Allende faced his biggest crisis so far, Beer’s invention became vital.
Across Chile, with secret support from the CIA, conservative small businessmen went on strike. Food and fuel supplies threatened to run out. Then the government realised that Cybersyn offered a way of outflanking the strikers. The telexes could be used to obtain intelligence about where scarcities were worst, and where people were still working who could alleviate them. The control rooms in Santiago were staffed day and night. People slept in them – even government ministers. “The rooms came alive in the most extraordinary way,” says Espejo. “We felt that we were in the centre of the universe.” The strike failed to bring down Allende.
In some ways, this was the high point for Cybersyn. The following year, like the government in general, it began to encounter insoluble problems. By 1973, the sheer size of the project, involving somewhere between a quarter and half of the entire nationalised economy, meant that Beer’s original band of disciples had been diluted by other, less idealistic scientists. There was constant friction between the two groups. Meanwhile, Beer himself started to focus on other schemes: using painters and folk singers to publicise the principles of high-tech socialism; testing his son’s electrical public-opinion meters, which never actually saw service; and even organising anchovy-fishing expeditions to earn the government some desperately needed foreign currency.
All the while, the rightwing plotting against Allende grew more blatant and the economy began to suffocate as other countries, encouraged by the Americans, cut off aid and investment. Beer was accused in parts of the international press of creating a Big Brother-style system of administration in South America. “There was plenty of stress in Chile,” he wrote afterwards. “I could have pulled out at any time, and often considered doing so.”
In June 1973, after being advised to leave Santiago, he rented an anonymous house on the coast from a relative of Espejo. For a few weeks, he wrote and stared at the sea and travelled to government meetings under cover of darkness. On September 10, a room was measured in La Moneda for the installation of an updated Cybersyn control centre, complete with futuristic control panels in the arms of chairs and walls of winking screens. The next day, the palace was bombed by the coup’s plotters. Beer was in London, lobbying for the Chilean government, when he left his final meeting before intending to fly back to Santiago and saw a newspaper billboard that read, “Allende assassinated.”
The Chilean military found the Cybersyn network intact, and called in Espejo and others to explain it to them. But they found the open, egalitarian aspects of the system unattractive and destroyed it. Espejo fled. Some of his colleagues were not so lucky. Soon after the coup, Beer left West Byfleet, his wife, and most of his possessions to live in a cottage in Wales. “He had survivor guilt, unquestionably,” says Simon.
Cybersyn and Stafford’s subsequent, more esoteric inventions live on in obscure socialist websites and, more surprisingly, modern business school teachings about the importance of economic information and informal working practices. David Bowie, Brian Eno and Tony Blair’s new head of policy, Geoff Mulgan, have all cited Beer as an influence.
But perhaps more importantly, his work in Chile affected those who participated. Espejo has made a good career since as an inter- national management consultant. He has been settled in Britain for decades. He chuckles urbanely at the mention of Pinochet’s arrest in London five years ago. Yet when, after a long lunch in a pub near his home in Lincoln, I ask whether Cybersyn changed him, his playful, slightly professorial gaze turns quite serious. “Oh yes,” he says. “Completely.”
Project Cybersyn: Chile 2.0 in 1973
by Patrick Philippe Meier / February 21, 2009
We were both stunned by what was possibly one of the coolest tech presentations we’ve been to at Berkman. Assistant Professor Eden Medina from Indiana University’s School of Informatics presented her absolutely fascinating research on Project Cybsersyn. This project ties together cybernetics, political transitions, organizational theory, complex systems and the history of technology.
I had never heard of this project but Eden’s talk made we want to cancel all my weekend plans and read her dissertation from MIT, which I’m literally downloading as I type this. If you’d like an abridged version, I’d recommend reading her peer-reviewed article which won the 2007 IEEE Life Member’s Prize in Electrical History: “Designing Freedom, Regulating a Nation: Socialist Cybernetics in Allende’s Chile”.
Project Cybersyn is an early computer network developed in Chile during the socialist presidency of Salvador Allende (1970–1973) to regulate the growing social property area and manage the transition of Chile’s economy from capitalism to socialism.
Under the guidance of British cybernetician Stafford Beer, often lauded as the ‘father of management cybernetics’, an interdisciplinary Chilean team designed cybernetic models of factories within the nationalized sector and created a network for the rapid transmission of economic data between the government and the factory ﬂoor. The article describes the construction of this unorthodox system, examines how its structure reﬂected the socialist ideology of the Allende government, and documents the contributions of this technology to the Allende administration.
The purpose of Cybersyn was to “network every ﬁrm in the expanding nationalized sector of the economy to a central computer in Santiago, enabling the government to grasp the status of production quickly and respond to economic crises in real time.”
Stafford is considered the ‘Father of Management Cybernetics” and at the heart of Stafford’s genius is the “Viable System Model” (VSM). Eden explains that “Cybersyn’s design cannot be understood without a basic grasp of this model, which played a pivotal role in merging the politics of the Allende government with the design of this technological system.”
VSM is a model of the organizational structure of any viable or autonomous system. A viable system is any system organised in such a way as to meet the demands of surviving in the changing environment. One of the prime features of systems that survive is that they are adaptable.
Beer believed that this five-tier, recursive model existed in all stable organizations—biological, mechanical and social.
Based on this model, Stafford’s team sought ways to enable communications among factories, state enterprises, sector committees, the management of the country’s development agency and the central mainframe housed at the agency’s headquarters. Eventually, they settled on an existing telex network previously used to track satellites. Unlike the heterogeneous networked computer systems in use today, telex networks mandate the use of speciﬁc terminals and can only transmit ASCII characters. However, like the Internet of today, this early network of telex machines was driven by the idea of creating a high-speed web of information exchange.
Eden writes that Project Cybersyn eventually consisted of four sub-projects: Cybernet, Cyberstride, Checo and Opsroom.
▪ Cybernet: This component “expanded the existing telex network to include every ﬁrm in nationalized sector, thereby helping to create a national network of communication throughout Chile’s three-thousand-mile-long territory. Cybersyn team members occasionally used the promise of free telex installation to cajole factory managers into lending their support to the project. Stafford Beer’s early reports describe the system as a tool for real-time economic control, but in actuality each ﬁrm could only transmit data once per day.”
▪ Cyberstride: This component “encompassed the suite of computer programmes written to collect, process, and distribute data to and from each of the state enterprises. Members of the Cyberstride team created ‘ quantitative ﬂow charts of activities within each enterprise that would highlight all important activities ’, including a parameter for ‘ social unease ’[…]. The software used statistical methods to detect production trends based on historical data, theoretically allowing [headquarters] to prevent problems before they began. If a particular variable fell outside of the range speciﬁed by Cyberstride, the system emitted a warning […]. Only the interventor from the affected enterprise would receive the algedonic warning initially and would have the freedom, within a given time frame, to deal with the problem as he saw ﬁt. However, if the enterprise failed to correct the irregularity within this timeframe, members of the Cyberstride team alerted the next level management […].”
▪ CHECO: This stood for CHilean ECOnomy, a component of Cybersyn which “constituted an ambitious effort to model the Chilean economy and provide simulations of future economic behaviour. Appropriately, it was sometimes referred to as ‘Futuro’. The simulator would serve as the ‘government’s experimental laboratory ’ – an instrumental equivalent to Allende’s frequent likening of Chile to a ‘social laboratory’. […] The simulation programme used the DYNAMO compiler developed by MIT Professor Jay Forrester […]. The CHECO team initially used national statistics to test the accuracy of the simulation program. When these results failed, Beer and his fellow team members faulted the time differential in the generation of statistical inputs, an observation that re-emphasized the perceived necessity for real-time data.
▪ Opsroom: The fourth component “created a new environment for decision making, one modeled after a British WWII war room. It consisted of seven chairs arranged in an inward facing circle ﬂanked by a series of projection screens, each displaying the data collected from the nationalized enterprises. In the Opsroom, all industries were homogenized by a uniform system of iconic representation, meant to facilitate the maximum extraction of information by an individual with a minimal amount of scientific training. […] Although [the Opsroom] never became operational, it quickly captured the imagination of all who viewed it, including members of the military, and became the symbolic heart of the project.
Cybersyn never really took off. Stafford had hoped to install “algedonic meters” or early warning public opinion meters in “a representative sample of Chilean homes that would allow Chilean citizens to transmit their pleasure or displeasure with televised political speeches to the government or television studio in real time.”
[Stafford] dubbed this undertaking ‘ The People’s Project ’ and ‘ Project Cyberfolk ’ because he believed the meters would enable the government to respond rapidly to public demands, rather than repress opposing views.
As Cybersyn expanded beyond the initial goals of economic regulation to political-structural transformation, Stafford grew concerned that Cybersyn could prove dangerous if the system wasn’t fully completed and only individual components of the project adopted. He feared this could result in “result in ‘ an old system of government with some new tools … For if the invention is dismantled, and the tools used are not the tools we made, they could become instruments of oppression.” In fact, Stafford soon “received invitations from the repressive governments in Brazil and South Africa to build comparable systems.”
Back in Chile, the Cybernet component of Cybersyn “proved vital to the government during the opposition-led strike of October 1972 (Paro de Octubre).” The strike threatened the government’s survival so high-ranking government officials used Cybernet to monitor “the two thousand telexes sent per day that covered activities from the northern to the southern ends of the country.” In fact, “the rapid ﬂow of messages over the telex lines enabled the government to react quickly to the strike activity […].”
The project’s telex network was subsequently—albeit briefly—used foreconomic mapping:
[The] telex network permitted a new form of economic mapping that enabled the government to collapse the data sent from all over the country into a single report, written daily at [headquarters], and hand delivered to [the presidential palace]. The detailed charts and graphs ﬁlling its pages provided the government with an overview of national production, transportation, and points of crisis in an easily understood format, using data generated several days earlier. The introduction of this form of reporting represented a considerable advance over the previous six-month lag required to collect statistics on the Chilean economy […].
Ultimately, according to Stafford, Cybersyn did not succeed because it wasn’t accepted as a network of people as well as machines, a revolution in behavior as well as in instrumental capability. In 1973, Allende was overthrown by the military and the Cybersyn project all but vanished from Chilean memory.
email : firstname.lastname@example.org
STAFFORD BEER PAPERS COLLECTION
FANFARE for EFFECTIVE FREEDOM
Cybernetic Praxis in Government
by Stafford Beer / 14th February 1973
The Third Richard Goodman Memorial Lecture
Delivered at Brighton Polytechnic, Moulsecoomb, Brighton
This is the first memorial lecture I have given for a man I knew personally – a man whom I also loved. He was a tenacious cybernetician, the pioneer of that work here in Brighton, but one whose name at least was known throughout the cybernetic world. More than this and more importantly than this, he had a dedication to humanity. It may not be well known, but I knew, that he was as interested in the cybernetics of society as he was in the more recondite mathematics of the science. And I also know very well that he would have been captivated by the unfinished story I am telling here formally for the first time. If I could have had his advice while the project was unfolding, it might have been a better story. But I still hope that it is worthy of his memory.
In November 1970 Dr Salvador Allende became President of the Republic of Chile. In November 1971 after some letters had passed, a meeting held in London, and some homework done. I arrived in Santiago. There I first met the prepared group of a dozen men who formed the nucleus of a team which is now much larger, and with whom I am still working-for I have been commuting the 8000 miles between London and Santiago ever since. The charge was daunting indeed: how should cybernetics be used in the exercise of national Government? You will note that the question whether cybernetics had any relevance to the problems of society and of government had already been answered affirmatively.
What was and is the situation? The answer as I have intimately known it for these last eighteen months. is immensely complicated. Let me paint my own crude picture for you with a rapid brush. First, more than half the total population lives an urban life in the small central region of this long thin country–a region that perfectly balances the arid North and the wet South in a superb climate. Here the people are highly literate, and constitutionally minded their men are frank and friendly their women gorgeous and gay. There is as great a spirit of freedom in the air as I have sensed anywhere in the world-and decreasingly sense in so much of it today. Yet, as you must surely know. Chile is in the middle of a Marxist revolution that has so far been constitutional, so far legal, so far bloodless.
On the land the previous government had begun a process of agrarian reform and that policy had general agreement. Landowners would no longer control estates larger than eighty hectares -say about 200 acres. The residual land was split up and handed to worker’s co-operatives who have the support of government agencies. In the six years of that previous government about 20º of the programme was implemented. But the people were impatient especially in the South and a deeply embedded bureaucracy slowly moves. New forms of expression were given to agrarian reform and the programme was completed not always in good order. in the first two years of the government of Popular Unity. This rate of change has surely contributed to the current food shortage: not so much perhaps because the new arrangements are inefficient in themselves. but because the remaining landowners-disrupted by these events and fearful of further change-are eating their seed corn rather than investing it in production.
In industry too the new government’s policies of nationalisation and worker participation have been implemented so rapidly that the control of that process was-and remains-extremely difficult. Foreign managers of expropriated firms have mostly left the country, and the problem of finding men to take temporary charge (these are the interventors) was-and remains-severe. It has been exacerbated by a brain drain of native Chileans: too many qualified professionals have left the country. That they should do so was surely implicit in their upbringing and their expectations, but their problem was much aggravated by the psychological panic induced by Opposition campaigns to spread rumours of terrors to come. As to industrial investment, we should note that all the banks were nationalised and those banks hold the internal assets of the landed classes.
Politically the government’s problems have been huge, all along. In the Presidential election that put Dr Allende in power, he obtained only 36% of the vote. The coalition he leads itself contains factions, which struggle for influence between themselves. Throughout he has faced a hostile Congress and Senate, capable of blocking any government initiative by the Oppositions’ majority of 60 % to 40%. On the other hand, the government is empowered to block the majority vote of Congress-so long as its own support is at least a third. Hence the political stalemate, hence the tension of the marginal vote, hence the importance of the Congressional Election next month.
All of this is easily recognised especially in cybernetic terms as a grossly unstable situation. And its explosive economic tendencies were perfectly predictable when 1 first became involved. There had been a very large and very sudden increase in the purchasing power of the rank and file. Wages rose fast for the land-workers in particular – who were put on the same footing as the blue-collar workers.
Social security benefits were much increased for everyone with young, old, or incapacitated dependants. Then clearly there would be a run on stocks: clearly there would be a run on reserves. Indeed this was well understood: on my very first visit a Minister took several hours to explain the risks being run, and the political determination with which those risks were accepted as the price of rapid social progress. The question was whether the government could get a sufficient grip on the situation in time – before this inflationary time-bomb blew up in its face.
In the event it did not, and the state of’ the country is very precarious. It is superficial to think of’ this in terms of food shortages and “housewives marches” tiresome as the food problem certainly is for the middle class. The more important fact is that Chile suffers from the effects of’ an economic blockade. There has been a blockade of spare parts, which has made it even harder to keep agriculture going, industry productive, and transportation moving. There has been a blockade on exports, by which I refer especially to copper – which used to earn more than eighty percent of the country’s foreign exchange. The attempt is being made to close world markets to Chilean copper. and the world price has fallen. Above all, there has been a blockade on foreign credit. And since Chile’s natural resources will one day make it a rich country when those resources are properly deployed, it follows that the stranglehold on credit is not a solely economic matter.
It appears to me that the government did not anticipate the full vindictiveness with which the rich world would react to its actions, which I emphasise have – so far – been perfectly legal. At any rate, a true resolution of the very potent conflicts in Chilean society is not discernible within the mounting instability, and may be long postponed. But I consider that this is largely a phenomenon of’ the cybernetics of international power: you could say that the Chilean people have not been given a chance. They are being systematically isolated behind those beautiful Andes Mountains, and are in a state of siege. The mass media have not helped much-especially inside the country itself, where freedom of speech has been respected in very testing circumstances. Because of its ownership, this freedom is largely employed to oppose the government. Because of its prestige, the anti-government press is widely copied – embroidered even – across the world.
It says a lot for the good intentions of the Government that the work I shall describe been going on in the midst of such obvious turmoil. It wanted scientific tools to help the country’s problems, and it knew that their provision would take time-perhaps long. So it may be proved. The government has so far had to work with the tools governments have used without success. It also wanted to work out the between science and the people, and that too ought to interest us all. We have moved an epoch in which the misuse of science has created a society that is already close to technocracy. The very language – the dehumanised jargon-in which powerful talk about the wars they wage, or powerful companies talk about the people they frankly makes me vomit.
I am a scientist, but to be a technocrat would put me out of business as a man. Yet I was eighteen months ago, intent on creating a scientific way of governing. And here today, proud of the tools we have made. Why? Because I believe that cybernetics can do the job better than bureaucracy – and more humanely too. We must learn how to expunge technocracy, without rejecting science – because the proper use of science is really the world’s brightest hope for stable government. Some people in Chile share that view; and they reject technocracy as strongly as do I. All of us have already been misrepresented in that respect, just as the scientific work we have done has already been misrepresented as analogous to other management control systems that have failed. Both comments miss out the cybernetics, to discuss which we are here – and a subject, which for government in general, is not at all understood.
What is cybernetics that a government should not understand it? It is, as Wiener (1) originally called it twenty-five years ago “the science of communication and control in the animal and the machine”. He was pointing in that second phrase to laws of complex systems that are invariant to transformations of their fabric. It does not matter whether the system be realised in the flesh or in the metal.
What is cybernetics that government should need it? It is, as I should prefer to define it today. “the science of effective organisation”. In this definition I am pointing to laws of complex systems that are invariant not only to transformations of their fabric, but also of their content. It does not matter whether the system’s content is neurophysiological, automotive, social or economic.
This is not to argue that all complex systems are really the same, nor yet that they are all in some way “analogous”. It is to argue that there are fundamental rules which, disobeyed, lead to instability, or to explosion, or to a failure to learn, adapt and evolve, in any complex system. And those pathological states do indeed belong to all complex systems – whatever their fabric, whatever their content – not by analogy, but as a matter of fact.
With cybernetics we seek to lift the problems of organisational structure out of the ruck of prejudice-by studying them scientifically. People wonder whether to centralise or to decentralise the economy – they are answered by dogmas. People ask whether planning is inimical to freedom – they are answered with doctrines. People demand an end to bureaucracy and muddle—they are answered with a so-called expertise which from its record has no effect. If dogma, doctrine and expertise fall to give effective answers, then what criterion of effectiveness shall cybernetics use? My answer to this question is: the criterion of viability. Whatever makes a system survival-worthy is necessary to it.
Necessary, yes, one might reply, but surely not also sufficient? The more I consider that criticism, the less I see its force. Suppose one was to say for example (pleading necessity), that since a particular anarchic society is failing apart, a high degree of autocracy will be needed to ensure its survival.
Then the critic might say “but this way lies totalitarianism and the loss of human freedom”. Not so, if we adhere to our viability criterion. Because that society would be unstable also: sooner or later would come a revolution-it always does. Suppose one were to say (pleading necessity) that a particular repressive society must throw over all constraint. Then the critic might say “then you will have chaos, and no one will be safe”. But that situation would not conduce to survival either. and the pendulum would swing the other way-it always does. The point is that a truly viable system does not oscillate to those extremes because it is under horneostatic control in every dimension that is important to its survival. Then when it comes to designing systems of government, we need to understand the cybernetic laws of homeostasis. Fortunately, and thanks mainly to Ross Ashby (2), we do understand.
Let me briefly explain. Homeostasis is the tendency of a complex system to run towards an equilibrial state. This happens because the many parts of’ the complex system absorb each other’s capacity to disrupt the whole. Now the ultimately stable state to which a viable system may run (that state where its entropy is unity) is finally rigid – and we call that death. If the system is to remain viable, if it is not to die, then we need the extra concept of an equilibrium that is not fixed, but on the move. What causes the incipiently stable point to move is the total system’s response to environmental change and this kind of adjustment we call adaptation. The third notion that we need to understand homeostasis is the idea of a physiological limit. It is necessary for a viable system to keep moving its stable point but it cannot afford to move it so far or so fast that the system itself is blown apart. It must keep its degree and its rate of change within a tolerance fixed by its own physiology. Revolutions, violent or not, do blow societies apart – because they deliberately take the inherited system outside its physiological limits. Then the system has to be redefined, and the new definition must again adhere to the cybernetic criteria of viability. Then it is useless for whoever has lost his privileges to complain about his bad luck so long as he uses a language appropriate to the system that has been replaced. He must talk the new language or get out. This fact is the fact that is polarising Chilean society now.
By the same token, a society that does not have a revolution violent or not, inevitably goes on talking the inherited system’s language even though the rate of change has made it irrelevant to the problems which that society faces. Perhaps this fact is the fact that begins to polarise British society now.
At any rate, cybernetic analysis – I have tried to give you merely its flavour – enables us to study the problems of a particular society in terms of its viability. In general I have only this to say about societary homeostasis in the nineteen-seventies:
• A homeostat works (and we know all the cybernetic rules) by moving its stable point in a very complicated response to the shocks it receives to its total system.
• Any homeostat takes a finite time to re-establish its new stable point. This is called the relaxation time of the system.
• Today it is typical of social institutions that the mean interval between shocks (thanks to the rate of change) is shorter than the relaxation time. That is because the institutions were originally designed to accept a much longer interval between shocks.
• From this it follows that societary institutions will either go into a state of oscillation, or Plunge into that terminal equilibrium called death.
The cybernetician will expect the politician to adopt one of two basic postures in the face of these systemic troubles.
The first is to ignore the cybernetic facts and to pretend that the oscillations are due to some kind of’ wickedness which can be stamped out. The second is to undertake some kind of revolution, violent or not, to redesign the faulty instruments of government. I do not have to relate the polarisation throughout the entire world to which this cybernetic expectation is the key. But it seems very clear to me as a matter of management science that if’ in these typical circumstances you do not like violence, then you should quickly embark on a pacific revolution in government. If you do not, then violence you will certainly get.
Outstandingly it was Chile that embarked on this recommended course of pacific revolution. But, as I have already argued, the process has strained Chile’s internal homeostatic faculties to the breaking point. Let me restate the reasons I gave before in cybernetic terms. Firstly it is because its minority government has been frustrated in fully restructuring the system according to the criteria of viability. Secondly it is because in the wider world system Chile’s experiment was observed as an oscillation to be stamped out. How this will end I do not know. Meanwhile, however, we had set out to redefine the internal homeostasis.
I went to Chile armed with a model of any viable system, which I very well understood. It had taken twenty years to develop, in modelling, testing, and applying to all manner of organisations. The book expounding it (3) was already in the press when this story started.
One of the key ideas the general theory embodies is the principle of recursion. This says that all viable systems contain viable systems and are contained within viable systems. Then if we have a model of any viable system, it must be recursive. That is to say, at whatever level of aggregation we start, then the whole model is rewritten in each element of the original model, and so on indefinitely.
If we model the state when one element is the economic system: if we model the economic system, then one element is an industrial sector: if we model that industrial sector, then one element is a firm. The model itself’ is invariant. See what happens if we go on with this recursion. An element of the firm is a plant, an element of the plant is a particular shop; an element of’ the shop Is a section: an element of’ the section Is a man. And the man is assuredly a viable system – as a matter fact, the model started from the cybernetic study of’ man’s effective neurophysiological organisation in the first place.
A second key idea was that by using the viability criterion all alone – for the reasons I gave earlier, one might succeed in identifying regions of policy in the total organisational space that represent homeostatically stable points for long term survival. I am pointing now to a possibility that it is open to mankind at last to compute a set of organisational structures that would suit the needs of actual men-as being at once themselves independent viable system with a right of individual choice, and also members of a coherent society which in turn has a right of collective choice. Now one of the main issues identified was the issue of autonomy, or participation (these are catchwords), or perhaps I mean just liberty for whatever element within whatever viable system. Then this means that there ought to be a computable function setting the degree of centralisation consistent with effectiveness and with freedom at every level of recursion. This I now believe. It is a bold claim. Let me try to give it verisimilitude.
Government and management control systems range over a fairly wide spectrum on the autocratic-permissive scale and still remain viable. What is happening in cybernetic terms is that the homeostat connecting “the boss” to the people’s homeostat is either in high or low gear – while still operating within physiological limits. In an autocratic system the people’s homeostat is robbed of flexibility: in a permissive system, it is deprived of guidance and help. As long as oppression and freedom are seen solely as normative values, the outcome is determined by self-interest. Then we get polarisation, and people will fight to the death for a prospect, which is in either case ultimately not viable. But if we raise our eyes to the higher level of the total system in designing government controls, and use the viability criterion to determine the balance point, liberty must be a computable function of effectiveness for any total system whose objectives are known.
For example, when winning a war is the accepted objective – either for a nation or a guerrilla force – personal freedoms are acceptably sacrificed. But when a society fails to define its objectives, its consequent self-indulgence in freedom is met by a running tide of authoritarianism. And this is the explosive situation that so much of the world faces today, whatever its political colour, and at whatever level of recursion. Using the analysis I made a little earlier, the threat is that our world may not be viable much longer. Hence my plea for a cybernetic understanding of what is going on. I do not believe it has anything to do with genuine ethics: it is all about power.
Above all, the polarity between centralisation and decentralisation – one masquerading as oppression and the other as freedom – is a myth. Even if the homeostatic balance point turns out not to be always computable. it surely exists. The poles are two absurdities for any viable system. as our own bodies will tell us. And yet government and business continue the great debate. to the advantage only of those politicians and consultants who find the system in one state and promptly recommend a switch to the other.
These notions are central to the work I shall next describe. In Chile I know that I am making the maximum effort towards the devolution of power. The government made their revolution about it. I find it good cybernetics. But the tools of science are not anywhere regarded as the people’s tools: and people everywhere become alienated from that very science which is their own. Hence we are studying all these matters with the workers. Hence the systems I have to tell you about so far are designed for workers as well as ministers to use. Hence we are working on feedback systems to link the people to their government.
The enemy in all this is the image of exploitation that high science and the electronic computer by now represents. We are fighting that enemy and its ally technocracy. And so it must be only in Chile that you will find a famous folklore singer declaiming: “Seize the benefits that science gives the people in their quest” and “Let us heap all science together, before we reach the end of our tether”.
I am proud to have worked with Angel Parra on that song, which is called Litany for a Computer and a Baby About to be Born. Contrast that title with the headline given to the first public mention of this work which was leaked in a British newspaper last month, and has since been copied all over the world. It said: “Chile run by Computer” – Woe to the sub-editor who wrote that.
All that I have so far said is a very necessary preliminary to a right understanding of the economic control system I shall describe, which in any other terms would be a nightmare. But as society becomes differently understood – cybernetically restructured, politically redefined, differently lived by our children – yesterday’s nightmares may become tomorrow’s dreams. That is true for the whole of’ technological development. Without the re-structuring and the redefinition the nightmare remains as we who live in the polluted wake of ‘the industrial revolution ought very well to know. The thinking begins with one very clear idea. If things are changing very fast, then government needs instantaneous information. If its information is out of date, then its decisions are worse than irrelevant. Please consider this point very closely.
In 1956, Mr Harold Macmillan (who was at the time Chancellor of the Exchequer) complained that controlling the economy was like trying to catch a train using last year’s Bradshaw (time-table). It was true: the vital statistics of the nation were twelve months out of’ date. Sixteen years later, Mr Harold Wilson (at the time Immediate Past Premier and the newly elected President of the Royal Statistical Society) has recently explained things are better, and maybe many key national statistics are now only six or eight months out of date. And of’ course lags of either magnitude are commonplace in governments throughout the world. It will not do. This is not only because decisions taken cannot the benefit of the latest information, there is a far more ominous reason given in cybernetics.
It is a familiar notion that economic movements operate in cycles. Then out-of-date information is not merely ‘late’ it is precisely incorrect – because it represents some cyclical trend that has since been superseded, but this is not recognised. If economic cycles were regular in periodicity and amplitude there would be no problem and the delay could be easily corrected. The decision-taker would discount the time lag, and extrapolate. Indeed he tries to do this. Please look at Figure 1. By the time we discover either of the crises depicted, those crises are actually over. But we take action without knowing that, and therefore decide on exactly the wrong action each time. Now doing this actually causes instability.
To put the point in proper scientific terms: an unstable oscillation will occur at precisely the frequency for which the time lags cause a phase shift of 180º. The negative feedback signal reinforces – instead of correcting – the original error. It happens that the time it takes to implement a new government economic policy is of similar order to the statistical delay in acquiring facts, and so it is very possible to have the control system completely out of phase.
Lest this explanation should sound absurdly naive, let me add two reasons why the difficulty is not as perfectly obvious as I have made it appear. In the first place, neither of the lines I have drawn in Figure 1 is clear: both are fuzzy. That is, there is a tremendous amount of ‘noise’ present in the system – much of it deliberately injected by economic participants who stand to gain by causing this confusion. The second point is more difficult. The controller of in economic system is not a straightforward servomechanism with a known transfer function. It is itself a complex system, with its own time lags, which are separate from the time lags in the economy. It too may begin to oscillate, and in my experience, it does. Then there is a distinct likelihood that there will be a resonance effect between the two loops. If so, the oscillation in the controller would actually force a new oscillation onto the already oscillating system.
No wonder then that no one can disentangle all these effects, and no wonder that we do not perceive anything as simple as Figure 1 proposes. But in the absence of a complete explanation there is something that we can do. Instead of solving the problem, we can dissolve it. Let us get rid of all the time lags. Indeed we ought to break with the very idea of arbitrarily quantized managerial time. Just as lags in reporting the past produce a bogus periodicity, so quite clearly do the lags fed forward in planning the future. A year’s forward projection, or five-year plan, predetermine the cycle of expenditure and investment, and betray the capability of a viable system to adapt to environmental change. We cannot afford to await ‘the next quinquennial review’ when someone is standing on our foot.
What is the alternative to these inherited systems of lagged quantized reporting on what has happened and lagged, quantized response to projected change? The answer from the mid-sixties onward has been and remains real time control. We have the technology to do it. This concept was fundamental to the plan we drew up for Chile in late 1971. We would abandon the hare-and-tortoise race to make relevant statistics overtake the lag in data, capture, and analysis, and implant a real-time nervous system in the economy instead. We would forget about the bureaucratic planning systems that talk in terms of months and years, norms and targets, and implant a continuously adaptive decision-taking, in which human foresight would be permanently stretched as far in any context as this real-time input of information could take it. Above all, we would use our cybernetic understanding of filtration to deploy computers properly as quasi-intelligent machines instead of using them as giant data banks of dead information. That use of computers taken on its own as it usually is, in my opinion, represents the biggest waste of a magnificent invention that mankind has ever perpetrated. It is like seeking out the greatest human intellects of’ the day, asking them to memorise the telephone book, and then telling them to man ‘Directory Enquiries’ at the telephone exchange.
Having advocated all these policies for many years in Britain and elsewhere before going to Santiago I was alert to the potential objections. I knew very well what is the standard response of economists, of managers, of civil servants, of ministers, and of ‘established’ science to these ideas. Let me list seven of them and give you the answers in brief, since some (though I trust not all) of these worries may be in your minds already.
• First Objection: The boss will be overwhelmed with data.
Answer: Not so. This is what happens now, as any manager who has had a foot-high file of computer read-out slapped in front of him can attest. The idea is to create a capability in the computer to recognise what is important, and to present only that very little information – as you shall see.
• Second Objection: The management machine will over-react to such speedy signals, which may not be representative.
Answer: Not so. This also happens now, as shown embryonically in Figure 1. The objection disregards cybernetic knowledge of filtration, and damping servo-mechanics.
• Third Objection: Such a system would be too vulnerable to corrupt inputs.
Answer: Not so, again. Present inputs are corrupt and go undetected because they are aggregated and because the time has passed when they could be spotted. Clever computer programmes can make all sorts of checks on a real-time input to see if it is plausible.
• Fourth Objection: ‘Intelligent’ computer programmes to do all this are still in the science-fiction stage.
Answer: This is woolly thinking. People do not really think out what is involved because they conceive to the computer as a fast adding machine processing a databank – instead of seeing in the computer, quite correctly, the logical engine that Leibniz first conceived. The computer can do anything that we can precisely specify: and that includes testing hypotheses by calculating probabilities – as again you shall see.
• Fifth Objection: Even so, such programmes would take hundreds of man-years to write and be debugged.
Answer : I am sorry, but they did not. That is because the people involved in both London and Santiago were first rate programmers who understood what they were doing. Let me be brutal about this: how many managers are aware of the research done into the relative effectiveness of programmers? They should be. The best are anything from ten to twenty times as good as the worst; and when it comes to cybernetic programming, only the very best can even understand what is going on.
• Sixth Objection: A real-time system with on-line inputs? It is Big Brother; it is 1984 already.
Answer: Stop panicking and work out the notion of autonomy. I have still more so say about this later. All technology can be, and usually is, abused. When people turn their backs on the problem, crying touch-me-not, the abuse is the worse.
• Seventh Objection: Only the United States has the money and the knowledge to do this kind of thing: let them get on with it.
Answer: “I find that slightly boring”
Note: This objection was voiced to me in one of’ the highest level scientific committees in this land. The answer came from the Chairman and I was glad not to be in his withering line of fire at the time. But he did not prevail, and neither did I.
In Chile it took just four months to link up the key industrial centres to computers lit the capital city – using a mixture of Telex lines and microwave connections (Figure 2). Purists may well point out that this does not constitute a real-time teleprocessing network and they will be right. However, we have used the real-time philosophy and have simulated an on-line system. The programs are written for that: and if someone will kindly donate the teleprocessing equipment, it will soon be in action. (I have mentioned the problem of’ foreign exchange already.) Meanwhile, we have to use too many human interfaces. But I am not going to apologise much about that. The fact is that we can cope with daily input and that is-relatively-very close to real-time: in normal government terms, you cannot tell the difference.
This communications network was in itself a fairly simple technological manoeuvre; but even so it constitutes a big advance for government cybernetics. During the October crisis of 1972, some of the most senior people in Chilean government came fully to understand in practice what Wiener had expounded theoretically long before – communication is indeed control.
Well: to know today what was the state of the industrial economy yesterday is a considerable advance on knowing what it was six months or a year ago. But we were trying to do more than merely get up to date. Frankly, there is not much point in knowing what happened even yesterday – because even yesterday is the purest history. Nothing can be done about it any longer. But if we can get hold of a close idea of what is going to happen next week, then we have at least a chance of doing something about that. And certainly knowing what has been happening over the last few days is the best basis for estimating what likely to happen over the next few days….
The question is how? One may call for data, but he has to meet the problems I listed just now (- the ‘fatal’ British Objections -) if he is to make effective use of them. One may know all about yesterday, but he has to be fairly ingenious to say the right things about next week. The initial four-month plan of action, which had included setting up the communications network, tackled these problems too: and it successfully defeated them.
Interdisciplinary operational research teams set out to make (crude. but effective) models of all the major enterprises in the social economy. These were not to be the vast, static, historical, and essentially out of date and non-stochastic input-output matrices beloved of so many state planners. We wanted to get at the dynamic systems which made the enterprises tick; and we wanted them in a form that managers and ministers could immediately grasp. Therefore we used a visible and visualizable type of model, called a “quantified flow-chart”. Start with production (a Marxist government has no illusions about the source of the creation of wealth). If we list the production operations of the firm and their productive capability, we can make a map of production flow – in which the flow lines are proportional to the relative amounts of flow, using some convenient measure, and the operations themselves are boxes at the confluences – also shown in relative sizes according to their productive capability.
Now of course if that kind of presentation can be made for the flow of production it can be made for any other kind of dynamic system in which management may be interested: cash flow, for example, or the deployment and movement of people and of goods. And although we started out on this task under the aegis of Operational Research, I am hopeful that as people become accustomed to the idea we can use a better approach. Do we really need objective scientific enquiry to understand what the structure of the system is and how it should best be quantified? Actually not. The people who best understand what these systems are really like are the people who operate them. You do not need a string of degrees to understand how to make a quantified flow chart of the activity that surrounds your daily life. So here I hope will be the start of ‘participation’ in the future: and OR expertise will be used merely in teaching and in guidance.
With this simple device we start on the road leading to the answer to those objections about overload. In cybernetics we have an actual measure of complexity which we call variety. By devising systems in which homeostats are set up between management and whatever is managed, we embark on the process that I have labelled ‘variety engineering’.
The quantified flock, chart is in itself a variety attenuating filter. In the first place, it can select its own degree of optical resolution. For example, it can show a box called simply steel production: or it can show three boxes identifying kinds of steel production – by open-hearth, electric arc, and converter. say: or it can show every individual furnace. By the same token, it can lump together all the materials that go into a steel-making furnace charge, or it can distinguish between them. This variety engineering concerns the account of the operation that has meaning for a particular management group, and the degree of optical resolution chosen depends on the level of recursion at which this operation is being considered. In the second place, iconic representation is also a variety attenuator in the suppression of words and numerical data: it is a product of Gestalt psychology, in which pattern is relied upon to convey information.
The next variety attenuator involved in this representation is the concept of capability. The real-time variation in actual flows and outputs is killed in the iconic quantified flow chart, and referred instead to a relatively static idea of ‘what can be done’. You might think that this would be difficult to define, but in practice it is fairly easy. Capability is a systems concept: what outputs is the total system capable of generating in each part, given the limitations imposed on any part of the system by other parts? Then ‘capability’ is not to be confused with ‘capacity’ which is not a systems concept – because it alleges that some part of the system can in theory do something that may be rendered impossible by other parts. This variety attenuator is valuable because it reflects reality for the whole system concerned, and that has meaning for the recipient of the iconic representation.
However, we could – given a breakthrough of some kind – do better than the results of which we are currently capable. After all: if capacity exceeds capability in some parts of the system, there must be other parts of the system (called bottlenecks) that are actively restricting capability. These bottlenecks may have to do with low local capacities, or they may have to do with technological constraints. For example: a mill’s engine might be perfectly adequate to drive its rolls at twice their current speed – if only we had a better lubricant. Then these considerations define potentiality, which is something better than capability. Potentiality is the performance of which the system would be capable, “if only. . . . .” That does not mean that we look for pie in the sky: it means that we look for investment – in new equipment to cure the bottlenecks, or in research to cure technological shortcomings. It is not very difficult keeping one’s feet firmly on the ground, to define a system’s potentiality.
But if potentiality is better than capability, there is something worse – and that is actuality. The performance of systems cannot rise to their potentiality without investment of some kind, it cannot even rise to their capability unless activity is perfectly well organised. It never is. In consequence, what actually happens falls short of the capability expressed before. Moreover, actuality expresses that very reality of which I spoke earlier – the day-to-day vicissitudes of life. It was this continuous variation which drove our thinking down the road to real-time control. Somehow we have ended up with three versions of systemic truth: actuality, continuously fluctuating; capability, a much steadier variable; potentiality which is absolute until the system itself is structurally changed. And it is capability which the iconic representations represent. To make them show potentiality would for the moment, be unrealistic, to make them show actuality would, at all times, result in their dancing in perpetual fandango before our eyes. So this capability attenuator is a powerful but sensible reducer of operational variety.
So be it, in so far as iconic flow charts are concerned. But what about continuous reporting and the problems of real-time control? Whatever information we collect, it is due to be hurled round dozens of homeostatic loops – those loops that make up the total systems design. That information has very high variety: and the analysis we have just made multiplies it by a factor of three -or so it seems, if we want a measure not only of actuality, but of capability and potentiality as well ….. But rescue is in sight. Both capability and potentiality are relatively static measures. If we take their ratio the resulting index will also be relatively static. Moreover, such a ratio will be a massive variety attenuator – because it ill be a pure number, varying between nought and one. So instead of trying to consider all-in-one-breath that we have a capability of 800,000 tons and a potentiality of 1,000,000 tons, we shall think of a ratio of 0.8. While the capability to use 110 men contrasted with a potentiality to use only 22 yields a ratio of 0.2, and the capability cost of an item of product at 120 escudos compared with a potential cost of 60 escudos indicates a ratio of 0.5. Well what is potential in current capability is a latent resource: and it could be freed by investment in some form. So I call the ratio between capability and potentiality the Latency Index. Looking at a new iconic diagram (Figure 4) we can see how potent a variety attenuator has been devised.
There is no need any longer to try and assimilate the numbers that characterise the units measured. That is the strength of an index -it is a pure number, varying over a fixed range. Hundreds of thousands of tons: hundreds and tens of men, units of money, there is no need to wrestle with them. Nor, if we stick to our ideas about iconic diagrams is there any real need to use digits at all. We can distinguish very clearly, using our eyes, between the levels represented in the iconic diagram. It might satisfy an accountant, but it would make no difference to a manager, to declare that a Latency Index had changed from 0.71 to 0.73. Who cares? The computers behind the manager’s eyes will undertake whatever process of discrimination has meaning for his Judgmental brain. Then this was the first though massive piece of variety engineering we set out to achieve in Chile, on those initial (crude, but effective) models contrived at an appropriate level of optical resolution of all the firms. As I said, the Latency Index is all about investment, and we shall certainly return to it later.
Meanwhile we must consider actuality, the real-time variable in the entire system. For if a Latency difference between 0.71 and 0.73 means nothing because both potentiality and capability are fairly static, such a difference in a fast-moving index could mean something very important. It might be part of a trend. I have already explained the arrangements by which the data representing actuality come into Santiago every day. They are used to form a second ratio, comparing actuality (the newly arrived figure) with capability (selected from the computer store). This is the Productivity Index. It is in a continual state of oscillation, which destroys the variety that is of no concern. In the next diagram (Figure 5), we can see how the three concepts of actuality, capability and potentiality are combined as two ratios to form the Latency and Productivity Indices, and how these in turn create an overall Performance Index. The reason for this iconic representation, in place of the familiar mathematical notation. lies in the fact that which part of the ratio is the numerator and which the denominator depends on what is being measured. For instance, capability is always better than actuality, but in numerical terms it may be more (e.g. output) or less (e.g. man-hours per unit). Naturally enough, the smaller number in the ratio is the numerator, since the index will be less than 1.0.
The indices procure an enormous variety reduction: even so, we still have problems in conforming to Ashby’s Law of Requisity Variety when it comes to managing the economy. The company production models for instance generate on average about ten triple-indices per plant: these always include raw material and finished stocks, the output of key production processes, and labour absenteeism. This degree of resolution is minimal, and managements have free rein to install whatever extra indicators they like. This honours the argument for autonomy, and it makes an insignificant difference to the workload of the computers because all the numbers inside the computational system are diurnal time series of indices varying between nought and one. The programmes are therefore infinitely extensible in application. Even so, with the system in full operation, many thousands of actuality inputs will arrive daily, generating three times as many indices; and the total number could easily rise by two orders of magnitude as the autonomy criterion is understood by managements, the operational research goes deeper, and worker participation becomes real. And so we reach the more subtle notions of variety engineering.
If a particular indicator. say the rate of crushing limestone in a cement factory in Northern Chile. is generating a new Productivity Index every day. what ought to be done with it’? Should we lay the new figure, each day, on the desk of the Minister of Economics? Surely not, this variety must also be filtered. There are two statistical notions involved, and the first is very simple. A population of (say) a hundred such figures generates a probability distribution. This may turn out to be oddly-shaped. rather than straightforwardly Gaussian. and especially it may be skewed to the right (since the index has a finite limit of one). It is a simple matter, however. to correct for this statistical aberration, by using a trigonometrical transformation. Then we may establish the mean and variance of this population of indices. These two statistics, all alone, characterise the stochastic behaviour of each index over time. Then if we take a running sample of the indexical figures as they are computed. it is easy to establish whether a significant change in the mean or variance of the statistical population has occurred. The statistical population characterising each indicator is known as the taxonomic index, because it classifies every measured activity within every operation according to its mean productivity. There is a standard computer programme that looks for changes in the taxonomic index; if such a change is found, that is notified to the management concerned. and the iconic graph is changed. Further, the history of the index over time is updated (Figure 6). These are relatively rare events. but the procedure mentioned absorbs the variety engendered perfectly well.
The more difficult problem and the more sophisticated statistical notion concerns the possible trend that each new daily figure may betray. If the economy is to be under real time control, the government cannot wait to know that a significant change has been registered for a particular taxonomic index. Although this is already much to be preferred to the orthodox system of routinely quantizing statistics where the recognition of significant change is left first to the alertness and next to the judgement of whoever is supposed to be watching the results. No, it is much more than this: we approach the problem of breaking the time barrier. Can we tell from yesterday’s figure and the short-term run in which it participates, what will happen (unless we intervene) tomorrow and next week? It is the problem of short-term forecasting with which a great deal of progress has been made in recent years.
Allow me once more to return to the facts of the Chilean work. Before the end of 1971 I had designed a specification for the computer programme to deal with taxonomic indices having daily actuality inputs, and it was in the hands of a team of operational research consultants in London, who had been commissioned to write the programs. We were discussing she short-term forecasting problem, when the London team discovered a brand new paper in the Operational Research Quarterly, – hot off the press. The authors were Harrison and Stevens and they had clearly made a major advance in the field of short-term forecasting (4). We had been talking in terms of Cusum (cumulative sum) techniques to this point- as representing the best available practice. Cusum itself was associated with the first author. who had been pressing its virtues for many years, so we were naturally impressed that this novel development came from him. The obvious power of the method (always supposing it worked), and the elegance of the mathematical demonstration behind the approach convinced us to take the plunge. It was a noteworthy decision. The London team wrote a temporary suite of programs, which included the Harrison-Stevens approach and incredibly had it working in Santiago by the March 1972 deadline of the first phase of the operation mentioned already. Meanwhile they began work on the permanent version, creating a specification that was handed over to the Chilean scientists. In the meantime, as the system was growing, experience was gained in the actual use of these complicated program suites. and they grew in sophistication all the time. But these developments, vitally important though they are, must await presentation by the men made them possible in more technical papers than this.
This suite of computer programs, called CYBERSTRIDE, is the essential feature of the filtration system that achieves the variety attenuation demanded, and which breaks the time barrier of’ which I was speaking. It takes as input the actuality figures every day: makes various checks on their integrity: it computes the triple-indices: it makes statistical judgements about the taxonomic indices as I have already described. After that, using Harrison-Stevens techniques, it really gets clever.
When a new value for any index is computed, Cyberstride looks at it in the context of the recent history of that index (Figure 7). The new point might stand for any one of the four outcomes depicted. It stands for no change or for a transient (neither of which maters to the manager), or it stands for a change of slope, or for a step function, (both of which possibilities matter very much). Using Bayesian statistical theory, the program calculates the posterior probability of’ each of these four events – for every index, every day. The programme is incredibly sensitive to these changes recognising them long before the human brain would dare to make a judgement. Cybernetically speaking, the system (as Harrison and Stevens claimed) is self-adaptive: its sensitivity increases whenever uncertainty increases – which happens whenever an apparently unusual index value is thrown up. Moreover, instead of producing merely single-figure forecasts (and who can foretell the future with that kind of precision?) it produces a joint parameter distribution that expresses the inherent uncertainty of all forecasting.
So this is what I meant in speaking of computers as quasi-intelligent machines. Cyberstride throws away the huge component of variety that has no meaning because it represents a chance fluctuation. It is at once alert to significant changes focussing on them an analytical eye, and capable of estimating on the strength of that analysis what will happen next. The only problem we had with Cyberstride, and it was very severe, was its calibration in terms of these posterior probabilities: how sensitive should it be made? Obviously, it could discard too much or become overexcited about too little. The ‘tuning’ subroutine that fixes these limits of excitation, so analogous to the so-called physiological limits of variation in any homeostat, was the big achievement of the Chilean scientists working on Cyberstride.
The variety engineering is complete-for the lowest level of recursion, the enterprise itself. If it would have been ludicrous to confront the Minister of Economics with the whole variety of fluctuating indices. it would still be absurd to inform him of even highly significant movements in the limestone-crushing activity of a cement plant in Northern Chile. Absurd yes, but also ominous I am sure you recall the argument about autonomy and over-centralisation. What happens in Chile is this. The results of applying Cyberstride daily to the new inputs which quantify the iconic flowcharts, are fed straight back to the managements concerned. It is their responsibility to do something about the warnings that are generated in this way by quasi-intelligent machines. No other human being than the responsible manager receives any information about this extremely elaborate piece of computation and I attach very weighty importance to this fact.
Then, you will ask, what about the other levels of recursion? The manager of the enterprise is very well served by all of this especially so since he can pump any indexical series he cares to contemplate into the routine – and receive the alerting advice whenever it is available. Meanwhile he may feel perfectly confident that an absence of alerting advice means that whatever operations or activities are being monitored for him by Cyberstride are fluctuating within the physiological range of chance variation. But what about the Sector Committee, the Industrial Branch, the Minister of Economics himself? These are higher levels of recursion: how are they to be informed?
Here is the coup de grace of the cybernetician, in his role as variety engineer. All viable systems are contained within viable systems. It is the principle of recursion; the model is the same. So it is easy to see what next to do. The iconic representations called quantified flow-charts are to be aggregated at sector level aggregated again at the industrial branch level, and aggregated finally at the level of total Industry. The quantifiers (those actualities, capabilities, and potentialities) are to be aggregated too-not. as is orthodox practice, in terms of averages, but in terms of new operational research models (crude, but effective) of the level of recursion concerned. In that case, raw data-heavily processed through atomic indices and through Cyberstride which produces exceptions known only to the manager concerned-bypass that atomic level of recursion, and become raw material for a molecular level of aggregation higher up. Here they lose their identity, they merge (not by averaging but by modelling) into new molecular indices.
But the new indices, although they have lost a great deal of variety in the process of molecular aggregation, have acquired variety by the sheer amalgamation of so many enterprises. How shall we deal requisitely with this new variety? Well, it is represented by triple-indices, all operating between nought and one. So although the level of recursion changes and although the atomic index changes to a molecular index, the Cyberstride suite of programmes is invariant. The whole process I have described starts again. This time and again exceptional information is fed back to its proper level of’ recursion: the sector or the branch or the minister.
Return with me now for the last time to the vexed issue of ‘autonomy. I regard the whole of this work as a fanfare for freedom – but for effective freedom. The claim was made that the degree of autonomy, and its complement the degree of centralisation, are computable functions of viability. I stick to that. By separating the levels of recursion, and within those levels by preserving freedom for each separately designed interlocking homeostat, the maximum autonomy consistent with effective organisation is assured. A problem remains. What happens when, for whatever reason, the appropriate homeostat at the appropriate level of recursion FAILS TO ACT?
Many a freedom must have been lost from the fear of those in power that subservient systems down the line would not do their jobs. And if not, it makes a good excuse for the tyrant. This is a classic and intransigent problem, but we can now deal with it easily – if we keep our cybernetic heads. An autonomous unit is supposed to react to any adverse exception reports that it receives from Cyberstride. How long will that take.and how much does it matter? The answer to both questions will vary widely. In our work we have included in the operational research modelling a requirement to assess the possible rate of reaction to change and the relative importance to the system modelled of such a change, for every indicator. When the computer sends an exception report to a manager, at whatever level of recursion, it computes for the message an acceptable delay time, which is a function of both the possible reaction time and the importance, and it starts a clock. If our quasi-intelligent machine fails to detect an improvement within this allotted time, it breaks with the autonomy and notifies the next level of recursion (as well as telling the responsible manager that it has done so).
These special signals are different in kind from the routine management signals. We call them “algedonic”. The word means pain-and-pleasure and it was work in neurocybernetics that taught me this answer. We rely on our bodily organs to do their jobs, but if they should fail we get a special signal – transmitted by specially adapted neural pathways – that bring the facts to our conscious attention. The mechanism is precautionary. Clearly it involves a threat to autonomy but the body politic cannot sustain the risk of autonomic inaction any more than we can as human beings. And remember that there is nothing covert about this. The delay factors are discussed with the managers concerned, and they are informed if the algedonic signal is transmitted. Indeed they may be very relieved – if the problem is seen as beyond their control – to know that the signal has automatically gone.
In this way, just as in the body, a sign of special distress automatically breaks through to whatever level is required to deal with it (Figure 8). For if the management group which receives the signal falls to act within its appropriate time delay, the signal will go up to the level next above. Thus the signal makes it possible for a problem concerning that limestone crusher in the cement factory to reach the President’s Economic Control, let us hope that never happens: but it would be surprising if signals of distress were never received there from the Sector level.
The real-time control system I have so briefly described is founded on the following elements:
A cybernetic model of any viable system,
A cybernetic analysis of the real-life systems appropriate to each level of recursion and their iconic representation;
A design of a large number of interlocking homeostats, the provision of a national communications network capable of operating now on a daily basis and eventually on the basis of continuous input; variety engineering throughout the system to incorporate filtration on the human brain’s scale; And the Cyberstride computer program suite capable of monitoring inputs, indexical calculations, taxonomic regulation, short-term forecasting by Bayesian probability theory, autonomic exception reporting, and algedonic feedback. It makes quite a package, and it exists. It represents a system of here-and-now management of the economy that is not based on historical records, but on an immediate awareness of the state of affairs and the protection of that awareness into the short-term future.
Let us call this whole thing the NOW system. Then clearly we also need the FUTURES system. What are we doing all this for? If government is not to be merely the management of perpetual crisis it needs to look ahead. Party-political programmes are supposed to be all about the kind of society the people want and the government is supposed to be dedicated to achieving that. In practice, perpetual crisis drives out mandated intentions. It even happens (dare I say?) that entire sets of electoral policies become reversed when power is granted. This can be only because government has no arrangements for realistic normative planning. It has a political theory but it does not understand the system it is manipulating. It is just laughable to say for example: “the theory is all right but the trade unions (or the City, or the banks, or the people themselves) will not operate the theory”. The unions, the City, the banks, and the people are all elements of the total system that the government claims to be able to govern.
Thus I introduce what I have to say about long-range planning in terms of understanding systems and how they respond: and I do so in deliberate contrast to the many schools of thought that base their conception of inventing the future simply on forecasting it. My objection to that approach is twofold. In the first place I do not believe that we can forecast the future – and that is a fairly strong objection. The future I reckon, is known only to God, and it seems to me that the class of men who have always come nearest to perceiving his intentions are the science fiction writers. They have usually been very close to scientific reality. The people who run society who are famous for being ‘realistic’ and ‘responsible’ turn out to be outrageously irresponsible just because they are so unrealistic. Their unrealism consists in a refusal to notice what science is actually doing and a refusal to think through the inevitable systemic consequences of the politics they underwrite.
These were the reasons why I was determined to provide the Chilean government with an instrument for investigating the systemic consequences of alternative courses of action. For there really are choices to be made. When you read that car prices in Chile have gone up by 900% in a single year, what is your response? Is this the inevitable result of Marxist dogma, is it just what you expect from nationalisation, is it a measure of inflation or what? To whom does it occur that it may be the result of a deliberate choice between economic desiderata? Thus, are we all brainwashed by the consumer society in which the motor car is an absolute good.
The second reason why I object to the forecasting approach to long-range planning is that it assumes that there is a “future” out there lying in wait for us. This is not true, except in so far as larger systems beyond our own – and in which we acquiesce – take a stranglehold on us. I have already suggested that this may doom the Chilean experiment. The real freedom we have is to change our structures and our policies so that the future is different from the future we should have encountered had we not made those changes. And this is where understanding dynamic systems becomes the task. The fact is that we need not to forecast but to experiment.
Experimentation is not easily or perhaps justifiably done when we are talking of social institutions. Scientists undertake social experiments on animal populations, which they try to use as models of human populations-but the discrepancies, may be very wide. Probably the best experimental tool available is the computer simulation. According to this approach, one programmes a computer to represent the dynamic social situation, and then experiments on that. If one asks how such a model could possibly be validated-he learns that the model can be fed with historic data – on the basis of which it ought to simulate the appropriate historic outcome. That is at least a start in a demonstration of validity.
I introduce the topic of dynamic systems simulation in this way, calling it an experimental tool, because I consider there to be a great deal of misunderstanding on the subject. If we experiment on a model putting in possible policies and reading off possible outcomes, then of course we appear to be making predictions. Some people have been causing a great deal of public disquiet with some such predictions about the ecosystemic future of the planet. Personally, I do not mind their doing so – because I believe the public ought to be thoroughly disquieted on this score. But we must make methodological distinctions here. In so far as these models make predictions, it is vital that projections of the input variables be correctly made. There is the rub because specialists disagree quite fundamentally about the trends that have been built into some of those models. Clearly if it is taken as input that fossil fuel will run out by a certain date, then predictions for the ecosystem incorporating this input will be falsified if that date turns out to be wrong. But suppose our objective is not to make predictions but to make experiments to find out how the ecosystem works. That is a different matter. We should put in a whole range of possible dates for the exhaustion of fossil fuel, and find out what difference they made to total performance and by when. After that, we should have a good idea what policy to adopt towards research into novel sources of energy. And that policy would not be the fruit of predictions that might well be falsified, it would be the embodiment of our understanding as to where the system’s vulnerability lies.
My belief is that government planning should be based on this same idea. If we make a dynamic model of the economy, concentrating our power of resolution on the areas in which our decisions appear most unsure or most frightening, then we shall learn how the system operates. The first task is to identify the crucial parameters, which (because complex systems are richly interactive and internally reverberating) are not always the parameters assumed to be critical. It is quite characteristic of cybernetic studies to obtain results that are counter-intuitive. Therein lies their value. The next task is to discover how these parameters may best be manipulated which (because political dealing is a complicated business too) may be in roundabout ways rather than by direct intervention.
What matters about a dynamic system if you want to understand how it behaves, is not so much noticing the sore points themselves, nor resolving the apparently insoluble politics of applying remedies to those sore points – all of which turn out to be unacceptable remedies for some segment of the population. What matters is to change the structure of the system so that homeostatic equilibrium is restored and the sore points disappear. That involves variety engineering: it is likely to mean the redesign of institutions, the addition of informational feedbacks, and the calculated change of time lags in various rates of flow. Economists, perhaps, would not recognise those three cybernetic prescriptions as counting towards the solution of what are regarded as economic problems. But are all our problems economic? I think there is a prior set of problems about the regulation of society (which it falls to governments to solve), which may well have economic causes and consequences, but which are themselves about effective organisation.
Returning to the Chilean story then, we wanted to create a facility for normative planning, suitable for all levels of recursion, embodying dynamic system simulation. Now the task of inventing a fresh computer compiler for this purpose was outside our time-scale. A number of compilers exist, and we chose to use the Dynamo compiler in its latest version (5). The choice was made on the grounds of its elegance and its relatively long existence- meaning that it is very well debugged. The choice has been criticised and will be again – because this is the very compiler used in the work that I referred to as making predictions about the planet using inputs that many ecologists regard as insecure. To me, that is like blaming the pornographic content of a book on the English language in which it is written. My defence of the compiler says nothing about my concurrence or otherwise with ecological predictions any more than hearing my defence of English would tell you my views on pornography.
For the record then, we again formed two teams – one in Santiago and the other in London. These teams were organised differently from the two Cyberstride teams and had no members in common, and this simulation pair operated in a different way. Instead of members of the London team taking program suites out to Santiago to be developed, as happened with Cyberstride, a member of the Chilean team came to London to learn a cybernetic skill. Moreover, whereas all the Cyberstride runs using actual data were undertaken on the Santiago computers the simulation runs for a long time were undertaken on computers in London. In this way dynamic systems models for the Chilean futures system were originally developed, but by now, the whole of the work is being done in Santiago.
There is much that is new about these models. but for obvious reasons I shall not discuss their content. What is worth remarking upon is the status of the information fed into them. As I said just now models of this type are often criticised on the grounds that their inputs are suspect. Now this is not surprising: because, as I also said before, economic information at the national level is usually about a year out of date. But Cyberstride produces information that is immediate. Then there is a question about the interface between the real-time control system and the futures system. If absolutely current information can be used continuously to update our models of the world, a new era dawns in national planning. Well, at any rate it happens just like that in the brain. We should indeed be foolish to choose between the alternatives open to us as men on the strength of knowing what our circumstances were like last year.
And now we reach the final question, How do we ‘get it all together’. How is so much sophisticated science to be made available to those who bear the brunt? In most countries this is a function for the civil service. Those people constitute the ultimate filter. The ministerial briefing stands, however, responsibly, between the minister and all those urgent facts of the NOW system, all those experiments in foresight of the FUTURES system.
I wanted ministers to have a direct experience, an immediate experience, an experimental experience. And what goes for ministers goes also at another level of recursion for managers -whether the managers of the social economy or (at yet other levels of recursion) of enterprises or of plants. Above all, if “participation” has any meaning, no one must be disbarred because of an inadequate grasp of jargon, of figure-work, of high-level rituals. As I have told you before, the workers themselves must have access to the whole of this. Let me put the point before you in two contrasting ways.
When I first expounded the cybernetic model of any viable system (which I have not expounded today) to President Allende, I did so on a piece of paper lying between us the table. I drew for him the entire apparatus of interlocking homeostats in terms of’ neurophysiological version of the model – since he is by profession a medical man. It consists of a five-tier hierarchy of systems. I worked up through the first, second, third and fourth levels. When I got to the fifth I drew an historionic breath – all ready to say “And this, companero presidente, is you”. He forestalled me “Ah” he said with a broad smile as I drew the topmost box: “at last – the people”. I do not care what political outlook any of us may have: that story ought to convey a profound message. It deeply affected me, and it affects this work. The second perception of the same point that I give you comes from that Litany written by the folklore singer Angel Parra, which I quoted at the outset. This is what his song says on the subject (my translation):
Equal I say to the Minister
Selling promises forlorn
Since all of us are hostage
For that baby to be born.
Society can no more afford the alienation of the people from the processes of government than it can afford their alienation from science. And is this really a political question any more once we say that all of us are men? The fact is that no man, worker, or minister has more neurological jelly-ware than anyone else – although he may make marginally better use of his endowment. We have seen how that man, minister, or worker, can be saved from drowning in an inundation of statistics and reports – through variety engineering and the deployment of computers as quasi-intelligent machines. But how does the filtered information get into his head? The answer to this lies in the operations room. If the connotation of that phrase reminds some people of a wartime headquarters, the allusion is quite deliberate. For in the ops room real-time information is laid out quite graphically for immediate decision, and in the opsroom a synoptic view of the whole battle is made plain so that the total system can be encompassed by human powers of foresight. We used every scrap of relevant scientific knowledge in designing the place – neurocybernetic knowledge of brain processes, knowledge from applied and group psychology, knowledge from ergonomics. The opsroom looks like a film set for a futuristic film. But it is not science fiction it is science fact. It exists, and it works, it exists and it works for the worker as well as the minister. (Photographs of this opsroom are reproduced on the inside of the dust jacket of this book.) There are seven chairs in there, because seven is the maximum creative group. There are various screens in there, all using iconic representations of information, because those are the sort the human brain can best handle.
The central screen-central in that all the others are referred to it – is a picture of the viable system (Photograph 3). It is eight feet high and four feet wide, and it is set up according to the recursion theorem for whatever group happens to be using the room. The operations involved are marked in the circles, the current molecular indexical levels for the taxonomic triple-index in each element are shown by iconic descriptions in the square boxes. The total square in each case stands for potentiality: then the green level is actuality, and the red level is capability. You can see how easy it is if you remember the explanation about their ratios to get an immediate grasp of the relative levels of Productivity and Latency in each case.
There are a great many interlocking homeostats operating here which can be discussed as people come to understand the cybernetic laws that govern their behaviour. This is not readily understood from a still photograph and in fact this screen is animated. There are no arrows to be seen, therefore: just moving lines. Scientists often suppose that to mark a line with an arrow makes it clear that the total system so encumbered with arrows is actually dynamic. Not so, people read the arrows as indicating directional, but still static relationships. Besides, the most critical loops here operate at differential speeds (they can be changed) – which tells the brain a great deal about the relative lags in the system. In the top third of the diagram, three boxes may be noted. The lower of these controls the NOW system, and the central one controls the FUTURES system. The top box (the boss or ‘the people’) monitors their interaction. to which attention is drawn in the animation by a constant movement in the big yellow circle. The Cyberstride exception reports flow on the horizontal red lines and when they exist, an Exception Screen is lit giving details. An algedonic signal is indicated by flashing red arrows on the vertical axis. In the photograph, such a signal may be seen emanating from the middle element: and any algedonic activity lights the Algedonic Screen.
Shown in Photograph 4 are the two screens I have just mentioned. On the left is the Exception Screen, showing two alerting signals from Cyberstride, together with a first indication of the kind of warning coming through. On the right is the Algedonic Screen, showing signals from two different levels of recursion below – each marked by a red light flashing at a different speed. As I said, attention veers to these two real-time inputs because of the clues given on the main screen first described. Obviously both these screens are currently set up by hand, whereas they could be set by direct electronic output from the computer. But I would like to repeat that this is simply an annoyance due to component shortages it does not represent a gap in the total cybernetic system.
This then is the real-time input to the opsroom – its sensing devices spreading out over three thousand miles of country, and its quasi-intelligent filtration continuously reducing an immense informational variety to human proportions. Then what will our seven-man team of creative thinkers want to do next? For make no mistake, the opsroom is a decision machine, in which men and equipment act in symbiotic relationship to amplify their respective powers in one new synergy of enhanced intelligence. They have to start talking and deciding on their actions. For this purpose, they will need background information, and I need hardly explain that there are no files, libraries of reports, or minutes of the last meeting here. Paper is banned from this place. The answer is Data-feed. (Photograph 5).
It consists of three data screens, as you can see, and a huge index screen. Each of the data screens in supported by five carousel projectors, each carrying eighty slides of iconic information. So we can choose three out of twelve hundred presentations – one out of four hundred on each screen. But it is obvious that twelve hundred slides cannot be listed on the index screen above, however huge. How shall we get at our treasure-trove of supportive information? It is again a problem in variety engineering: select three from twelve hundred. Of course, one could have a catalogue and a decimal keyboard. That would have requisite variety. However, experience teaches that unskilled people will not usually agree to operate such devices. They see them as calling for a typing skill, and want to insinuate a girl between themselves and the machinery. Indeed, this fact has held up the development of on-line conversational computing very seriously indeed. We are faced with an ergonomic problem. It is vital that the occupants interact directly with the machine, and with each other.
In a creative conversation, men become very animated. They seize pieces of paper and draw on them, they snatch the pencil and change the drawing: “No! No! It’s like that!”. The solution to the ergonomic problem takes note of all these things. We produced special chairs which swivel through 270 degrees of arc, in the arms of which are mounted panels containing large knobs of different shapes (clearly visible in Photograph 1). By thumping one of the three knobs in the top row, a screen is selected – and an index automatically appears on the control screen. The index is catalogued by the use of five symbols, which are repeated in the second row of knobs. By pressing the appropriate combination of knobs the item selected from the first index appears on the control screen, in the form of a second index which lists the actual slides. So a second combination procures the required presentation.
The variety engineering says there are 25 = 32 ways of combining five knobs, and if that is done twice, 210 = 32 x 32 = 1024 alternatives are made available. That is enough selection power to handle 400 slides on each screen with plenty to spare for control engineering purposes. (Four buttons would yield only 28 = 256 alternatives.) One of the two knobs in the bottom row allows one man out of the seven to seize control of Data-feed with a thump and the other releases the control when he says thump – that’s what I mean. There is no finicky skill involved in working this apparatus – and people seem to take to it very quickly indeed. As to all that thumping: I wanted to make the dramatic act of using he equipment an effective part of’ the creative conversations just like seizing the pencil, or banging the table. Thus it is that when real-time inputs indicate the need for supportive information the decision-takers may select on the three screens (for example as in Photograph 5) the iconic flowchart that contains the relevant input, a photograph of the plant concerned, and some indexical information. If an expansion or explanation of that information is available, for example the history of a Latency. Index may well be supported by an investment plan for realising that latency, then a direct clue is given on the screen as to how to key that new slide into place (such a clue is visible on Screen C). In the close-up of Data-feed (Photograph 6), the picture of the plant has been replaced with a list of products.
All this supportive information is semi-permanent. It must in principle be updated, but not too often. As to its adequacy, remember that all sixteen-carousel magazines can easily be changed so we have a new set (of 1200 slide capacity) for each level of recursion. It is enough. In any case, there are two more back-projection screens in the opsroom to allow special presentations to be made. So far I have spoken about the NOW system, but certainly Data-feed supports the FUTURES system too. The relationship between the two is very clear on the huge main screen where its coruscating homeostat is a constant reminder of the need to balance investment between what is and what will be. And that FUTURES system, with its simulation capability, has its own screen (Photograph 7). This is the flow-chart of a typical Dynamo simulation – though not a very complicated one. The two points I want to make about it are unfortunately not communicated in this static picture.
The whole raison d’être for simulations is to work with them. They do not sit there ‘Making forecasts’ as I said. The output of the model shown is a projection made by a computer of how the major variables will vary over the next ten years – if nothing changes – and that projection is illuminated on one of the spare screens. To understand how the economic system works the people in the room need two facilities, neither of which is available on an ordinary flowchart. First, they must be able to alter the structure in front of them. That is easily done in the computer: attendant scientists can change a few equations on request, and produce a new read-out in a few minutes. But how do you alter the flowchart? The answer to that is to use flexible magnets, and we did. However, to decide how to alter the flowchart you must understand the flows – and therefore we wanted to animate this screen. The problem was how to animate a flowchart that you wish continually to re-construct. The British suppliers of the animated equipment solved that problem, and I wish I could show you the flow-lines on this model moving and how readily its structure can be changed. Indeed we could spend all day in the opsroom together without exhausting its meaning as a new tool of management, and a new route to worker participation. This is the first one ever built on these cybernetic principles and it is only a beginning. The room and its furnishings were designed and made in Chile. The optical system and control logic for Data-feed were designed and built in England, and the animated screens were created by another British manufacturer. I have described such a room as this over many years and once wrote: “It is not the operational research, technology or experience that is lacking to produce the first (such) control centre. It is the managerial acceptance of the idea, plus the will to see it realised (3). I finally found both the acceptance and the will – on the other side of the world.
This has been a very long lecture, but it deals with a very large subject: how the science of effective organisation, which we call cybernetics, joins hands with the pursuit of elective freedom, which we call politics. What a new – and what a vital issue those words betoken. Where have I heard them before?
“The cybernetics of men
As you, Socrates,
Often call politics. . .”
You can tell from that name that I am quoting, and we seem to be up against a time lag of two thousand years. But now we are doing something about it. Now we have some cybernetic tools.
What I have been able to tell you today however is plainly incomplete; please bear in mind that this whole thing began just sixteen months ago. Therefore, although the system exists, it is – in a proper academic sense-unproven. I expect that it like any other infant will be slapped on the wrist (if not worse) and told to toe the line – if not worse.
For during that period of sixteen months various attempts have been made to overthrow the Chilean democracy I have seen that, from fairly deep inside. Scientifically too, during that period I have been told a hundred times that it would take more than twenty years to do what has now been done – during that period.
We have to take note that innovation, whether political or scientific, does not favour those who hold the real power. And if either kind of innovation stands to favour ordinary folk and both these do, then it will be opposed.
For this reason, I am not naming here my many colleagues and collaborators. They know my feelings of esteem and affection for their ability, their dedication and their friendship. What any of them asks of me that I can do, he should consider done.
For this reason also I commend my compatriots here today to watch, more avidly than many doubtless have, what happens next in Chile. There will be lessons there for Britain I believe: and for humanity.
So now good-bye
I remember Richard Goodman in this very place.
Requiescat in pace
1. Norbert Wiener, Cybernetics, John Wiley, New York, 1948.
2. W. Ross Ashby, Design for a Brain, Chapman and Hall, London, 1954.
3. Stafford Beer, Brain of the Firm, Allen Lane, The Penguin Press, London, 1972.
4. P.J. Harrison and C.R. Stevens, ‘A Bayesian Approach to Short-term Forecasting’. Operational Research Quarterly, Vol. 22, No.4, December, 1971.
5. Jay W. Forrester, World Dynamics, Wright-Allen Press, Cambridge, Mass., 1971.
The cybernetics of Occupy: an anarchist perspective
by Thomas Swann / June 20, 2014
In 1963, in the British journal Anarchy, a short debate took place on the relationship between anarchist forms of organization and organizational cybernetics. Organizational cybernetics, for those unfamiliar with the term, is often defined as the science of communication and control in organic, mechanical and social systems. While the term might suggest images of high technology and even cyborgs, etymologically it is derived from the Ancient Greek word κυβερνήτης (kyvernítis, or kybernetes), which means ‘steersman’ or ‘pilot’, and referred to the steering of a ship. The contemporary usage draws on this analogy in the sense that cybernetics is involved in identifying and studying the ways in which systems and organizations regulate (or steer) themselves through mechanisms of feedback and emphasizes the importance of lines of communication through which this feedback is received and actions are taken as a result. The element of control that cybernetics attempts to explain in systems is one of self-organization: a system regulates and manages itself and doesn’t require external influence in doing so. The debate in Anarchy, between British cyberneticians William Grey Walter and John McEwan, focused on the application of early developments in this science to explain how self-organization works in the context of social systems. In other words, can the study of feedback, communication and self-regulation in biological and mechanical systems by applied to human social systems? One writer who took up this idea was the noted anarchist thinker Colin Ward.
Ward, the editor of the journal Anarchy, wrote an essay in 1966 titled ‘Anarchism as a Theory of Organization’ in which he elaborates on the dynamics of anarchist organization and emphasizes the connections between anarchism and organizational cybernetics. “Cybernetic theory,” he writes, “with its emphasis on self-organizing systems, and speculation about the ultimate social effects of automation, leads in a similar revolutionary direction” as anarchism (John Duda presents a brilliant analysis of the history of cybernetics and anarchism in an article published last year in the journal Anarchist Studies). Ward draws on the work of classical anarchist geographer Peter Kropotkin as well as on examples of non-hierarchical anarchist organization that take place in everyday life.
Some of the core features of anarchist organization, according to Ward, are (1) that collectives and groupings should be based on voluntary participation, (2) that they should be aimed at performing a specific function and (3) should temporarily exist only for as long as they perform that function, and (4) that they should be small and based on face-to-face contact between participants. For Ward, these elements would go at least some of the way towards ensuring the non-hierarchical nature of these forms of organization. Functions requiring larger organizational efforts ought to be met not by fixed structures but by federations in which “large-scale functions can be broken down into functions capable of being organized by small functional groups.” In addition to this, Ward is keen to highlight the importance of autonomy within organizational forms, and it is in this that his account of organization comes closest to that of organizational cybernetics. While research on cybernetics began during the Second World War and was properly formalized by Norbert Wiener (in his book Cybernetics: Or the Control and Communication in the Animal and the Machine, published in 1948) it is to a later cybernetician that I want to turn in identifying the similarities between organizational cybernetics and anarchism. Stafford Beer developed his account of cybernetics while working in the steel industry in the UK in the 1950s. Despite little formal education in engineering, mathematics or other sciences, Beer made a name for himself internationally with the publication of Cybernetics and Management in 1959.
It was on the basis of this book that he was invited in 1970 to assist Salvador Allende’s newly-elected socialist government modernize and rationalize the Chilean economy. Thus followed arguably the most important period in the development of cybernetics as Project Cybersyn, as it became known, brought together the political principles of Allende’s government with Beer’s approach to effective organization. Central to the whole process was autonomy: the autonomy of different parts of the economy (factories and other enterprises), networked together using telex machines and a single central computer used for processing information received from these individual parts. While the project was cut short by the US-backed coup in September 1973, the application of organizational cybernetics to the problems of communication and self-regulation in the Chilean economy had some remarkable successes during its short life (Eden Medina’s book Cybernetic Revolutionaries provides the whole story of the Chilean experience with cybernetics and is perhaps the most accessible and enjoyable books on cybernetics available).
The period Beer spent in Chile radicalized him politically, leading to him virtually renouncing the materialism of the upper-middle class lifestyle he had become accustomed to and eventually moving to a remote cottage in Wales that lacked electricity and running water. As Ward and other anarchists correctly identified, a potentially radical conception of autonomy and self-organization does lie at the center of organizational cybernetics and its account of effective organization. Beer’s early work, prior to his experience in Chile and written as it was in the 1950s and ’60s, focuses on industrial production, and he shows that in organizations where different units (i.e. those concerned with producing separate components or those involved in different tasks like sales and manufacture) have the autonomy to work according to their own directives — based on their unique knowledge of what is required within their niches — a level of internal stability and effectiveness is reached which is difficult if not impossible in organizations where a strict, top-down hierarchy permeates every action of the workers at the bottom of the chain of command.
Coordination is achieved by communication between units working in different niches rather than through centralized control. Beer basically opposes Taylorist scientific management with a call for granting autonomy to the individual parts of an organization. To be sure, the autonomy is limited within the overall plan of the organization, which is decided at a senior management level (in the Chilean case this relied on a social democratic account of parliamentary authority), but the role of autonomy, and the potential of what Beer called the Viable Systems Model holds for genuinely democratic and anarchist organization, is nonetheless fascinating.
The Viable Systems Model is based on a tiered conception of an organization in which different tiers or levels are responsible for different functions ranging from ground-level operations to facilitating communication between different operations to decision-making about the strategic goals of the organization. Beer’s classic model, presented in his 1972 book The Brain of the Firm, assumes that while those at the bottom of the organization, the individual operational units, should be granted autonomy to work in their niches as they see fit. Higher level activities such as determining strategy should however be undertaken by a group of trained managers, separate from the workers on the shop floor. Following the success and ultimate untimely demise of Project Cybersyn in Chile, Beer pushed the democratic character of the Viable Systems Model further by arguing that there should be mechanisms in place through which those at the bottom of an organization are able to influence policy at the top. Indeed his recommendations are not dissimilar to those proposed by proponents of E-democracy today. It is in the radically democratic character of anarchist organization, however, that the Viable Systems Model can be pushed to its furthest logical conclusion in terms of politics. As anarchists like Ward and McEwan highlighted, by insisting that the different levels in the organization be seen as functions into which different individuals can step at different times, rather than as fixed offices, those involved in the day-to-day operations of the organization can be the very same people involved in strategic decision-making and other levels of the organization. The autonomy in this more radical version of the Viable Systems Model is extended to include not just operational issues but all aspects of the organization.
In this way Beer’s politicization of organizational cybernetics can be taken in an even more radical direction and can form the basis for understanding the dynamics of effective and stable democratic organization. Crucially, the autonomy embedded in the Viable Systems Model takes on a dual role as an essential part of an effective system and as a practical outworking of anarchist political ideals. Turning to more recent examples of radical and anarchist organization, the general structure of this type of Viable Systems Model can be clearly seen. To elaborate on this, I want to take the example of the Occupy movement, and while what I’m presenting is an ideal account of how Occupy camps were organized, it serves to show the potential in this organizational form (which was common to much of the uprisings that occurred in 2011 and after) for a genuinely non-hierarchical, anarchist politics.
The Cybernetics of Occupy
The Viable Systems Model is composed of five layers or systems. In addition to these five layers there are the operating units of the organization (in the case of the Occupy example, working groups) and the environments in which they operate, some overlapping and others not, as well as the whole environment in which the organization or system exists. The overall goal of organizational cybernetics is to show how an organization can achieve its aims while remaining internally stable.
The blocks labelled 1A, 1B, etc. in the diagram above represent the members of working groups in communication with one another. While in Beer’s account of the firm these are middle managers or foremen who both formally and informally shout across the void, as he puts it, in this modelling of the ideal Occupy camp these can be any member of a working group in communication with any other. There aren’t any specific roles at this level of the organization. The importance of this level of communication is basically to make working groups aware of what the others are doing, so that autonomous activities can be coordinated in an informal way.
System 2 in the diagram envelops systems 1 within it. This is to represent the formal space in which the communications between working groups take place. It could be either the physical space of the camp itself or the online spaces created on various social networking platforms.
The third level of the model aims to allow members of the organization to reflect on the activities of the working groups in relation to the overall strategy of the organization as a whole. Discussions at this level would take place during the General Assemblies that came to symbolize Occupy’s decision-making structure. This allows the members of the working groups to consider their activities and adjust them if necessary in line with the decided-upon goals of the organization. For Occupy this is crucial as at this level the people engaging in the discussions are not separate or distinct from those involved in the working groups. The same individuals step out of their functional role as working group members and into that of reflecting on their practice within working groups. This is an essential distinction in the anarchist form of viable system developed here: there is a hierarchy in terms of function but not in terms of structure and it is not a hierarchy that issues commands from one group to another group. Rather, decisions are made democratically by all members of the organization. Limits are imposed on their autonomy but these are limits that are agreed upon together.
System 4 involves the same individuals again, and also the General Assemblies, reflecting on the activities of the working groups and the organization as a whole as well as its overall strategy in relation to events in the outside world. System 4 has a view to the tactical activities of lower levels in the model, strategic decisions made at the higher fifth level and the external environment in which the organization exists. This allows for adjustments to both tactics and strategy in light of changes in the environment that individual working groups might not be aware of. Again, for Occupy the distinction is functional and the activities of System 4 would involve all members of the organization discussing together at General Assemblies.
The final level of the model of Occupy I’m presenting here is that which is concerned with the strategy and overall goals of the organization as a whole. This is again a level of discussion and decision-making that is, or should be, open to all in the organization. It is where decisions are made about the objectives and priorities of the organization and is ultimately what limits the autonomy of the working groups; but again, this is not a limitation coming from a distinct group of leaders but is something that is agreed upon democratically by all members of the organization. In addition to these five levels within this ideal model of the Occupy camp, there are of course the flows of information that are represented by the lines with arrows. These highlight how information flows through the organization, between the different functional levels and to and from the external environment.
What this representation of Occupy as a viable organizational cybernetic system allows is not only an understanding of how things went well for the 2011 uprisings and similar experiences that, in one way or another, adopted a broadly anarchist form of organization (Not An Alternative’s critique of Occupy in fact reinforces the importance of autonomy within its structure), but also crucially where and why things went wrong. Mark Bray, for example, in a recent article on the failures of Occupy, identifies the divergent strategies between liberals and radicals in terms of the role of the individual within the collective. According to the model presented here, this comes down to a difference in strategy at System 5, which in turn provides differing messages to working groups in terms of the limits to their autonomy and their role in the organization. Without a clearly agreed upon strategy, individuals and groups operate along different lines and in often conflicting and mutually-detrimental ways. Agreement at System 5 allows for a clear paradigm in which to work and creates a clear objective for the organization and gives working groups the strategic perspective they need in order to effectively work autonomously within their own niches.
While using Stafford Beer’s organizational cybernetics to explain the successes and failures of Occupy undoubtedly misses out a lot of the analysis, its focus on levels of action and decision-making as well as on the lines of communication and flows of information can, I believe, help in understanding some of what was going on at an organizational level. The modifications to the model to account for the radical democratic processes put to work in Occupy and other uprisings are important in providing for a greater understanding of non-hierarchical organization and the role of autonomy and functional hierarchy within them. The working group structure of Occupy, combined with the space for strategic decision-making in the General Assemblies, made possible the kind of anarchist organization Colin Ward discussed in his 1966 essay. Working groups engaged functionally in different niches and, as a cybernetic analysis makes clear, were able to do so with a level of autonomy that was limited not by a hierarchical command structure but by decisions made by the very same individuals at the General Assemblies. The relationship between working groups, as represented in the anarchist version of the Viable Systems Model sketched here, mirrors in many ways that of the anarchist federation. Indeed, viewing the Occupy movement through the lens of the work of Stafford Beer in organizational cybernetics helps bring out the anarchist characteristics of this as a form of effective and at the same time radically democratic organization.