The Cybernetic Brain

by Andrew Pickering

  Back in England, Beer married Cynthia, and they had six children together, though the first was stillborn. Following a divorce, Beer married Sallie Steadman, a widow and mother of a daughter, Kate, and they had two more children, for a total of eight, but this marriage, too, ended in divorce, in 1996. From 1974 onward Beer lived alone in Wales for much of the year (see below). In 1981 he met and fell in love with another cybernetician, Allenna Leonard (then a mature graduate student and later president of the American Society for Cybernetics), and she was Beer's partner for the remainder of his life.

  Leaving the army, Beer hoped to do a PhD in psychology at University College, but when told that he would have to recommence his studies as a first-year undergraduate he turned his back on the academic life, and in 1949 he began work for Samuel Fox in Sheffield, a subsidiary company of United Steel, where he created and ran its Operational Research Group (probably the first such group to exist in Britain outside the armed forces). From 1956 until 1961 he was head of the Operational Research and Cybernetics Group of United Steel, with more than seventy scientific staff based in the appropriately named (by Beer) Cybor House in Sheffield. In 1961 he founded Britain's first operational research consulting firm, SIGMA (Science in General Management). In 1966 he moved on to become development director of the International Publishing Corporation (IPC), then the largest publishing company in the world, where his work largely concerned future initiatives around computing and information systems. In 1970, Beer left IPC "following a boardroom disagreement about development policy." From 1970 until his death in Toronto on 23 August 2002 he operated as an independent consultant in a variety of arenas, some of which are discussed below.

  Besides his career in management and consultancy, Beer was a prolific writer of scholarly and popular works, including more than two hundred publications and ten books on cybernetics, which he referred to as "ten pints of Beer" (Beer 2000). After 1970, he occupied many institutional roles and gained many honors. At different times he was president of the Operational Research Society, the Society for General Systems Research, and the World Organization of Systems and Cybernetics. He had several footholds in the academic world, though none of them full-time. His most enduring academic base was at the Business School of Manchester University, where he was visiting professor of cybernetics from 1969 to 1993. He was research professor of managerial cybernetics at University College Swansea from 1990 to 1997, visiting professor of management science at the University of Durham from 1990 to 1995, visiting professor of cybernetics at the University of Sunderland and life professor of organizational transformation at Liverpool John Moores University, both from 1997 until his death. And so on, including visiting professorships at many other universities in Britain, Canada, Sweden (Stockholm), and the United States dating from 1970 onward. He was awarded major prizes for his work in operations research and cybernetics by the Operations Research Society of America, the American Society for Cybernetics, the Austrian Society for Cybernetics, and the World Organization of Systems and Cybernetics. A festschrift in Beer's honor was published in 2004 (Espejo 2004), and two volumes of his key papers have also appeared (Beer 1994a; Whittaker 2009).

  Figure 6.1.Beer as businessman. Source: Beer 1994a, facing p. 1. (This and other Beer images in this chapter, where otherwise unattributed, are courtesy of Cwarel Isaf Institute and Malik Management Zentrum St. Gallen [www.management.kybernetik .com,www.malik-mzsg.ch].)

  Figure 6.1 is a photograph of Beer in the early 1960s when he was director of SIGMA—the smartly trimmed hair and beard, the three-piece suit, the cigar: the very model of a successful English businessman. In the early 1970s, however, Beer changed both his lifestyle and appearance. Partly, no doubt, this was in disgust at events in Chile with which he had been deeply involved, culminating in the Pinochet coup in 1973 (as discussed below). But also, as he told me, approaching his fiftieth birthday, he was moved to take stock of his life—"I had had two wives, I had eight children, a big house and a Rolls- Royce"—and the upshot of this stock taking was that in 1974 Beer renounced material possessions and went to live in a small stone cottage in a remote part of Wales.4 He retained the cottage for the rest of his life, but after the mideighties he divided his time between there and a small house he shared with Allenna Leonard in Toronto. This break in Beer's life was registered by a change in his appearance (fig. 6.2) and also in his writing style. Until this change, Beer's writing took a fairly conventional form. His first book in its wake was Platform for Change: A Message from Stafford Beer, printed on paper of four different colors, signaling different modes of argument and presentation. The introduction, printed on yellow paper, begins thus (Beer 1975, 1):

  HELLO

  I would like to talk to you

  if you have the time

  in a new sort of way

  about a new sort of world.

  It ends (6):

  I am fed up with hiding myself

  an actual human being

  behind the conventional anonymity

  of scholarly authorship.

  Figure 6.2.Beer after the move to Wales. Source: Beer 1994a, 315. Photo: Hans-Ludwig Blohm. © Hans-Ludwig Blohm, Canada.)

  From Operations Research to Cybernetics

  Beer's route into cybernetics began with his work in operations research (OR) which in turn grew out of his work in the British Army in India. We do not need to delve deeply into the history of OR, but some brief remarks are relevant. As its name suggests, OR developed in World War II as a scientific approach to military operations. "Scientific" is to be contrasted here with traditional approaches to tactical and strategic planning based on the accumulated expertise of military commanders, and wartime OR can be broadly characterized in terms of a quantifying spirit aimed at modelling military activities with an eye to optimizing performance. One could try to calculate, for example, the optimal U-boat search pattern to be flown by a specified number of aircraft of given speed and range. OR was first developed in Britain in conjunction with new radar technologies but was also taken to a high art in the United States.5

  Beer was not himself involved in the wartime development of OR. On his own account, he rather wandered into it while he was in the army, first by attempting to use symbolic logic, which he had studied at University College, to organize large numbers of men into functioning systems.6 He first heard of OR as a field on his return to England and plunged himself into it as he moved into civilian life. Two early papers, published in 1953 and 1954, for example, outline novel statistical indices for measuring the productivity of manufacturing processes which he developed and implemented at the Samuel Fox steel company. These papers have a very practical bent, including ideas on how the sampling of productivity should be done and how the information could be systematically and routinely collected, assembled, and presented. The aim of the indices in question was the ability to forecast how long it would take to perform any given operation, a topic of interest both to the managers and customers of the mill (Beer 1953, 1954).

  Figure 6.3.Beer's classification of systems. Source: S. Beer, Cybernetics and Management (London: English Universities Press, 1959), 18.

  Beer's career in OR was very successful, as is evident from the biographical sketch above, and OR continued to play an important part throughout his subsequent work, both as an employee and as a consultant. But at an early stage he began to look beyond it. The second of the OR papers just mentioned is largely devoted to the development and use of performance measures for individual production operations in the factory, but it concludes with a section entitled "The Future Outlook" (also the title of Grey Walter's novel in its English publication two years later) looking forward to the development of "models . . . which would embrace the whole complex manufacturing structure of, say, an integrated steelworks." Beer notes that such models would themselves be very complex to construct and use and mentions some relevant mathematical techniques already deployed by OR practitioners, including game theory and linear programming, before conti
nuing, "Advances in the increasingly discussed subject of cybernetics, allied with the complex models mentioned, might result in a fully mechanized form of control based on the technique described here" (1954, 57).

  What did cybernetics mean, in assertions like that, for Beer, and how did it differ from OR? This takes us straight back to questions of ontology and a concept that I have been drawing on all along, that of an exceedingly complex system. Here we need only return briefly to its origin. In his first book, Cybernetics and Management(1959), Beer distinguished between three classes of systems (while insisting that they in fact shaded into one another): "simple," "complex," and "exceedingly complex" (fig. 6.3). He gave six examples of the first two types (subdividing them further into "deterministic" and "probabilistic" systems). Under "simple" came the window catch, billiards, machine shop layout, penny tossing, jellyfish movements, and statistical quality control; under "complex" we find electronic digital computers, planetary systems, automation, stockholding, conditioned reflexes and industrial profitability. What those examples have in common, according to Beer, is that they are in principle knowable and predictable, and thus susceptible to the methods of the traditional sciences. Physics tells us about billiard balls; statistics about penny tossing; OR about stockholding and industrial profitability—this last, of course, being especially relevant to Beer. OR was, then, a classical science of production, a science appropriate to those aspects of the world that are knowable and predictable, in the same space as modern physics. However, under "exceedingly complex" systems (which, according to Beer, can have only probabilistic forms) we find just three examples: the economy, the brain, and the company. And Beer's claim was that these are "very different" (Beer 1959, 17):

  The country's economy, for example, is so complex and so probabilistic that it does not seem reasonable to imagine that it will ever be fully described. The second, living, example—the human brain—is also described in this way. Moreover, it is notoriously inaccessible to examination. . . . Inferential investigations about its mode of working, from studies such as psychiatry and electroencephalography, are slowly progressing.

  Probably the best example of an industrial system of this kind is the Company itself. This always seems to me very much like a cross between the first two examples. The Company is certainly not alive, but it has to behave very much like a living organism. It is essential to the Company that it develops techniques for survival in a changing environment: it must adapt itself to its economic, commercial, social and political surroundings, and it must learn from experience.

  Beer's exceedingly complex systems, were, then, as discussed already, in a different ontological space from the referents of OR (or physics). They were not fully knowable or adequately predictable, and they were "the province of cybernetics" (18). Beer's enduring goal was precisely to think about management cybernetically—to inquire into how one would run a company, or by extension any social organization, in the recognition that it had to function in and adapt to an endlessly surprising, fluctuating and changing environment.7

  Toward the Cybernetic Factory

  MY GOD, I'M A CYBERNETICIAN!

  STAFFORD BEER,ON FIRST READING WIENER'S CYBERNETICS (BEER 1994C)

  Beer first read Norbert Wiener's Cyberneticsin 1950 and plunged into the field, establishing an individual presence in it and close personal connections as he went. By 1960, "I had known McCulloch for some years, and he would stay at my house on his Sheffield visits. . . . The British pioneers in cybernetics were all good friends—notably Ross Ashby, Frank George, Gordon Pask, Donald MacKay and Grey Walter" (Beer 1994 [1960], 229). "Norbert Wiener, as founder of cybernetics, was of course my great hero," but Beer did not meet him until his first trip to the United States when, on 25 May 1960, Wiener "almost vaulted over his desk to embrace me," greeting Beer with the words "I have become increasingly conscious that the growing reputation of my work [Wiener's] in Europe derives in large measure from your lectures and writings, and from the fact that you have built Cybor House. For this I should like to thank you" (Beer 1994 [1960], 281, 283).

  In what follows, we will be largely concerned with connections between Beer's cybernetics and Ashby and Pask's. Beer and Pask actively collaborated in the work on biological and chemical computers discussed below and in the next chapter, and one can trace many parallels in the development of their work. But the defining features of Beer's cybernetics were Ashby's homeostat as key model for thinking about adaptive systems and Ashby's law of requisite variety, as a tool for thinking realistically about possibilities for adaptive control. Much of what follows can be understood as a very creative extension of Ashby's cybernetics into and beyond the world of organizations and management. During the 1950s, Beer experimented with a whole range of cybernetic approaches to management (e.g., Beer 1956), but two ideas quicky came to dominate his thinking. First, one should think of the factory (or any complex organization) in analogy with a biological organism. Second, and more specifically, to be adaptive within an unknowable environment, the factory as organism should be equipped with an adaptive brain.

  Beer laid out an early and striking version of this vision in a paper he presented to a symposium on self-organization held at the University of Illinois's Allerton Park on the 8 and 9 June 1960 (Beer 1962a). He opened the discussion with the notion of the "automatic factory," then attracting great interest, especially in the United States. This was a vision of industrial automation taken, one might think, to the limit. In the automatic factory, not only would individual machines and productive operations be controlled by other machines without human interference, but materials would be automatically routed from one operation to the next. In the "lights out" factory, as it was sometimes called, the entire production process would thus be conducted by machines, and human labor made redundant—literally as well as metaphorically.8

  Beer was not at this stage a critic of the automatic factory, except that he did not feel it was automatic enough. He compared it to a "spinal dog"—that is, a dog whose nervous system had been surgically disconnected from the higher levels of its brain. The automatic factory (1962a, 164) "has a certain internal cohesion, and reflex faculties at the least. [But] When automation has finished its work, the analogy may be pursued in the pathology of the organism. For machines with over-sensitive feedback begin to 'hunt'—or develop ataxia; and the whole organism may be so specialized towards a particular environment that it ceases to be adaptive: a radical change in the market will lead to its extinction." Beer's argument was that to make it adaptive and to avoid extinction in market fluctuations, the automatic factory would need a brain.

  At present, such an automatic factory must rely on the few men left at the top to supply the functions of a cerebrum. And . . . the whole organism is a strange one—for its brain is connected to the rest of its central nervous system at discrete intervals of time by the most tenuous of connections. The survival-value of such a creature does not appear to be high. . . .

  This will not do. The spinal dog is short of a built-in cerebrum; and the automatic factory is short of a built-in brain. The research discussed in this paper is directed towards the creation of a brain artefact capable of running the company under the evolutionary criterion of survival. If this could be achieved, management would be freed for tasks of eugenics; for hastening or retarding the natural processes of growth and change, and for determining the deliberate creation or extinction of whole species. (Beer 1962a, 165)

  The reference to eugenics is provocative to say the least, but the idea is an interesting one. The cybernetic factory, as Beer imagined it, would be viable—a key term for Beer: it would react to changing circumstances; it would grow and evolve like an organism or species, all without any human intervention at all. The role of humans in production would thus become that of metamanagement—managers would survey the field of viable production units and decide on which to promote or retard according to metacriteria residing at a level higher than production itself. Figure 6.4 is Beer'
s schematic vision of what the cybernetic factory should look like. and much of his essay is devoted to a formal, set-theoretic definition of its contents and their relations. This is not the place to go into the details of the formalism; for present purposes, the important components of the diagram are arranged around the circumference: the T- and V-machines at the left and right, bridged by the U-machine and "states of the world" at the bottom. The symbols within the circumference represent processes internal to the U-machine.