Here I need to enter a caveat. What might adaptation of these models in practice mean? I just described adaptation in the VSM as open ended, but Beer imagined and was prepared to implement something less than this in his models. He understood them as sets of mathematical equations linking long lists of variables such as demand, revenue, technological and economic change, dividends, share prices, and the money market. And the basic form of these sets of equations was not, in itself, revisable, at least as part of Beer's description of the regular functioning of a viable system. What could be revised in practice were the parameters figuring in these equations which specified the intensity of the couplings between variables. Beer's models were thus adaptive, but only to a degree, within a fixed overall form.24
One further point. The symbolic models of the VSM were envisaged as conventional simulations programmed on digital computers. In this respect, there was no distinctively cybernetic aspect to the VSM. But it is still instructive to review Beer's thoughts on the computerization of industry. It is important to note that Beer was himself an enthusiast for computers. As early as 1956 at United Steel he had had installed one of the first computers in the world to be dedicated to management science, a Ferranti Pegasus (Harnden and Leonard 1994, 4). He was nevertheless a consistent critic of the way in which computers were being introduced more generally into industry and business. His argument was that "the first and great mistake" was that "people set out to automate the procedures and therefore the organisations they already knew. These themselves were frozen out of history and fixed by professionalism." Computers were, in other words, being used to automate existing clerical tasks while leaving the overall structure of the traditional organization untouched: "Companies have exchanged new lamps for old, and set them in the window as marks of progress. . . . We are using a powerful control instrument competent to reorganise the firm, its departments and functions, and encapsulating it in a received system geared to the quill pen." Instead, Beer argued, we should ask, "What should my enterprise be like, now that computers exist?" (Beer 1967, 214–17).
Beer was especially critical of the use of computers in business to automate and augment record keeping, and this gets us back to the ontological question. If the world is beyond our capacity to know it, and if, even worse, it continually changes, knowing the past is of limited utility. Our information processing should therefore be forward looking, as in the system 4 model of the VSM. "It is worth making a tremendous effort to burst through the barrier marked 'now,' and to make managers concern themselves with what can be managed—namely the future, however near—rather than peruse a record of what can be managed no longer—namely the past, however recent. We may learn from that past record of course, but we cannot influence it in retrospect. . . . Look straight ahead down the motorway when you are driving flat out. Most enterprises are directed with the driver's eyes fixed on the rear-view mirror" (1981, 127, 199). Beer's idea in the VSM was thus that most of the information that one can collect on an organization is useless and can be discarded. This was what the filtering operations at the various levels did, keeping only anomalous signals for transmission to higher levels.
Seen from this angle, the object of the VSM was to reorganize the firm around the computer—to effect a transformation that was social as well as technological, to rearrange the human components as part of an adaptive technosocial system of information flows and transformations. Here, too, then, social relations and ontology hung together. And this contrast between the VSM and the traditional structure of the organization is another nice example of how ontology can make a difference in practice. Further aspects of this are apparent below.
The VSM in Practice
The VSM was a normative vision of the organization. Organizations had to look like the VSM if they were to survive and grow in time. The obvious implication of that would seem to be that they needed to be remade from the ground up to exemplify the VSM. Beer had one serious chance at that, which is reviewed in the next section. But Beer could hardly claim that all existing organizations were nonviable—some of them had been around for a long time, the Catholic Church, for example. He therefore made a more nuanced argument. Just like organisms, organizations could be more or less viable— some struggling to survive, others actually dying, others springing happily into the future: "The amoeba succeeded, the dinosaur failed, the coelacanth muddles along" (Beer 1981, 239). And the problem was that organizations had no way to discuss this temporal viability; they lacked any language or conceptual apparatus for it.
What organizations had instead was organization charts of hierarchical power relationships running downward from the board of directors through vertical chains of command devoted to production, accounting, marketing, and so on. Beer's claim was that such charts did not, and could not, represent how firms actually worked. They functioned, at most, as devices for apportioning blame when things went wrong.25 Already, then, whether anyone recognized it or not, the VSM was a better description of how the firm really worked, and Beer's pitch was that the formal VSM could therefore function as a diagnostic tool (1981, 155). One could examine the firm, or any other organization, and see just which bits of it corresponded to the five levels of the VSM, and one could examine the ways in which they were connected together. Certain aspects of the firm might thus be identified as especially deficient as compared to the VSM diagram and made the targets of therapeutic intervention. Beer claimed that an experienced VSM practitioner could often walk into a factory and identify the major problems within a day or two and that, once pointed out, management would recognize the veracity of the judgment—such problems having already been subconsciously recognized and papered over (Beer 1989a, 27). Of course, addressing the problems thus identified might take much longer—conceivably a period of years. "My guess would be that organizations cannot face up to more than a quarter of the reshaping that their long-term viability demands. This is of course the reason why so many enterprises are in a state of continuous. . . reorganisation" (Beer 1981, 239).
One simple example of what might be at stake here, of continuing practical and scholarly interest, concerns automation and the interlinkages between the systems 1 of figure 6.13. Beer noted that such linkages between different subsidiaries of a single company found no formal representation on the typical organization chart. But (Beer 1981, 107) "I have collected scores of examples of this. Sometimes, very often perhaps, the foremen in the related departments make it their business to keep in intimate touch. Maybe they walk across the road and drink tea together; maybe they telephone: 'You'd better know, Charlie, that . . .' In a few extreme case, it was not possible to discover how the messages were transmitted—but transmitted they certainly were." Beer was quite happy with such informal channels of communication; his only concern was that Ashby's law should be respected—that there should be enough variety at each end to cope with that at the other, and that there be enough bandwidth between them to mobilize and organize those varieties appropriately. Instead, Beer argued, the introduction of computers as information-processing devices often acted to sever such channels completely. Because the channels did not appear on the organization chart, they did not become automated; at the same time, their human conduits—the foremen, in this example—might be forbidden to step outside their own domains, or their positions eliminated entirely. "In the limiting case where the departmental outstation is fully automated, there is no possible way in which the social link can be maintained. Computers do not just happen to develop the trick of shouting to each other across the void, as human beings always do" (108). A technological transformation which appeared progressive on the surface might thus be regressive as seen from the perspective of the VSM.26
Beer also claimed that many organizations were entirely lacking a system 2 (1981, 175), and in the absence of the "sympathetic" damping generated by the 1-2-3 system would thus always be prone to pathological competition and "oscillations" between their subsidiaries. More generally, Beer worried about the higher
levels of the brain of the firm. Pieces of the organization which he felt should lie directly on the "command" axis were often found to be misplaced. This was true especially of parts of the organization that had grown up since World War II, including management accounting, production control (Beer's first job in the steel industry), and operations research (his first love in management). These had no place on prewar organization charts and thus found themselves a position almost at random (Beer 1981, 82–83). OR groups, for example, might be found buried in subsidiaries and thus serving the overall organization asymmetrically—to the benefit of some subsidiary rather than the whole firm. The moral of the VSM was that there should be an OR group on the command axis itself, at level 3. Beer also argued that "in most firms System 4 is a fiasco" (153–54). Elements of system 4—the monitoring and planning organ at the base of the conscious brain—were usually to be found in any large organization, but they tended to be dispersed across the organization instead of grouped coherently together on the command axis. Certainly very few clubby operations rooms were to be found in industry in this period.
We need to remember that from 1970 onward Beer made his living primarily as an independent management consultant, and his writings on the VSM were integral to that. In 1989, he produced a list of consultancies he had been engaged in (Beer 1989a, 35):
Small industrial businesses in both production and retailing, such as an engineering concern and a bakery, come to mind; large industrial organizations such as the steel industry, textile manufacturers, ship-builders, the makers of consumer durables, paper manufacturers are also represented. Then there are the businesses that deal in information: publishing in general, insurance, banking. Transportation has figured: railways, ports and harbours, shipping lines. Education, and health (in several countries), the operations of cities, belong to studies of services. Finally comes government at all levels—from the city, to the province, to the state and the nation itself—and the international agencies: the VSM has been applied to several.
Obviously . . . these were not all major undertakings, nor is "success" claimed for massive change. On the other hand, none of these applications was an academic exercise. In every case we are talking about remunerated consultancy, and that is not a light matter. The activities did not necessarily last for very long either, since speedy diagnosis is a major contribution of the whole approach. On the other hand, some of them have lasted for years. Undoubtedly the major use of this work to date was in Chile from 1971–1973.
Chile is next. Here I can just emphasize what is obvious from this list: Beer operated not only at the level of commercial companies; many other kinds of social organizations were likewise open to his interventions. We should also remember what was noted earlier—that by the 1980s the VSM had gained a significant following among management consultants and their academic counterparts, leading to the publication of at least one multiauthor book on the VSM (Espejo and Harnden 1989). The interested reader can look there for case studies written up by Beer and his followers, including Beer's sixtypage account of his association over nine years with a mutual life assurance company (Beer 1989a), as well as for various methodological and substantive reflections on and extensions of the VSM. The VSM was never one of those great fads that seem to have periodically overtaken the world of management since the Second World War. Given its subtlety and complexity, to which I have done scant justice here, this does not seem surprising. But it has been at the heart of a significant movement.
Chile: Project Cybersyn
In Chile in the autumn of 1970 Salvador Allende became the world's first democratically elected socialist president. The new government started nationalizing the banks and major companies operating within Chile, operating through an existing organization known as CORFO (Corporacíon de Fomento de la Produccíon). On 13 July 1971, the technical general manager of CORFO, one Fernando Flores, wrote to Beer (Beer 1981, 247): "This letter spoke of 'the complete reorganization of the public sector of the economy,' for which it appeared its author [Flores] would be primarily responsible. He had read my books, and had even worked with a SIGMA team ten years before. He went on to say that he was now 'in a position from which it is possible to implement, on a national scale—at which cybernetic thinking becomes a necessity—scientific views on management and organization.' He hoped that I would be interested. I was." Beer's commitment to the project became "total" (245), and he subsequently published a long account of the project's evolution and termination, in five chapters added to the second edition of Brain of the Firm (Beer 1981, 241–399). Beer's chapters are, as usual, very dense, and I can only attempt an overview of his account as a way of sketching in the main features of what was undoubtedly the world's most striking cybernetic project.27
Taking up Flores's invitation, Beer flew into the capital of Chile, Santiago, on 4 November 1971, remaining for eight days and returning to London on 13 November. In Santiago he met Flores and his collaborators, and together they made plans to implement the VSM at the level of the national economy. Beer had just completed the manuscript of Brain of the Firm; the Chileans studied it while he was there, and it became the basis for their vision of Chile's future. On 12 November Beer met President Allende himself and explained the VSM to him. When Beer drew the box for system 5 of the VSM diagram, he was thinking of it as representing the president, but Allende "threw himself back in his chair: 'at last,' he said, 'el pueblo' " (Beer 1981, 258)—the people. Beer was so impressed by this that he told the story often. Allende was apparently similarly impressed with Beer and the VSM: " 'The President says: Go ahead—fast' " (257).
What did the plan sketched out on Beer's first visit look like—Project Cybersyn, for "cybernetic synergy," as it became known? Beer felt that speed was of the essence—"within a year . . . the foreign reserves would run out" (251)—so he aimed to begin by installing a cut-down version of the VSM by, astonishingly, 1 March 1972. This was less than four months after his first visit, and he promised to return on 13 March 1972. The initial plan aimed to achieve real-time (meaning daily) communications between system 1 productive activities at the level of individual factories, and a system 4 control room to be constructed in Santiago.
OR teams were charged "to construct a quantitative flow chart of activities within each factory that would highlight all important activities" (253). OR models would then be used in consultation with management—typically workers' committees, foreign managers having fled the country—to construct indices of performance analogous to those Beer had devised in the steel industry and reported upon in the 1953 OR paper discussed above (163).28 "In practice, it turned out that some ten or a dozen indices were adequate to monitor the performance of every plant" (253). Among these was to be an index to measure morale as a ratio depending inversely on absenteeism (253).
The question of what to do with all the data thus generated, how to handle it, then arose. Ideally, every plant should have its own computer to "process whatever information turned out to be vital for that factory's management" (252)—this, thinking of each plant as a viable system in its own right. "But such computers did not exist in Chile, nor could the country afford to buy them. . . . Therefore it was necessary to use the computer power available in Santiago: it consisted of an IBM 360/50 machine and a Burroughs 3500 machine" (252). The remaining technical problem was to connect plants all over the country up to Santiago. This was to be accomplished by requisitioning telex machines, augmented by microwave and radio links whenever necessary. "The plan allowed just four months for this to be accomplished (and it was)" (252). This national information system was known as Cybernet; the data it brought to Santiago were processed there "and examined for any kind of important signal. . . . If there were any sort of warning implied by the data, then an alerting signal would be sent back to the managers of the plant concerned" (253). Beer himself took two tasks back to England with him (256): "I had to originate a computer program capable of studying tens of thousands of indices a day, and of evaluating the
m for the importance of any crucial information which their movements implied. . . . I had done this kind of system building many times before. . . . Secondly, I should need to investigate prospects for a simulation system in the operations room that could accept the input of realtime data. This would be a completely novel development in operational research technique." The basic blueprint and timetable for Cybersyn were thus set. Beer's own account covers subsequent developments in some detail; we can review some of the main features.
As indicated above, the Cybernet national information system was indeed established by the deadline of March 1972. The first computer program mentioned in the above quotation took longer than hoped to construct, partly because of the incorporation of very new OR techniques in forecasting. A temporary version was indeed implemented in March 1972, but the permanent version only became operational in November that year. By that time "something like seventy percent of the socio-industrial economy was operating within this system, involving about four hundred enterprises" (Beer 1981, 262, 264).29
These "Cyberstride" programs sat at the system 3 level, contributing to the homeostasis of the 1-2-3 economic assemblage while at the same time filtering data upward into the 3-4-5 system. A key element of the latter was a computer model of the Chilean economy and its national and global environment. This was to be the centerpiece of system 4 planning, intended to enable future projections according to different inputs and assumptions. This program was also Beer's responsibility. Lacking time to design such a model afresh, Beer announced in a January 1972 report that he had decided "to make use of the immediately available DYNAMO compiler extensively developed by J. W. Forrester of MIT. I have directed three projects in the past using this compiler, and have found it a powerful and flexible tool" (266). Forrester's work had grown by devious routes out of his World War II work at the Servomechanisms Laboratory at MIT and was just about to become famous, or notorious, with the publication of the Club of Rome's Limits to Growth report, which, on the basis of DYNAMO simulations, predicted an imminent collapse of the global economy and ecosystems.30 Work in London and Chile under Chilean direction had developed a tentative version of the Checo (for Chilean economy) program by June 1972, and by September a better model was running. "I wanted to inject information in real time into the Checo program via Cyberstride. Thus any model of the economy, whether macro or micro, would find its base, and make its basic predictions, in terms of aggregations of low-level data—as has often been done. But Checo would be updated every day by the output from Systems 1-2-3, and would promptly rerun a ten-year simulation; and this has never been done. This was one of my fundamental solutions to the creation of an effective Three-Four homeostat; it remains so, but it remains a dream unfulfilled" (268). This continual updating was the way in which Checo simulations were foreseen as evolving in time, responsively to real-time input, thus exemplifying the performative epistemology of the VSM discussed in general terms in the previous section.
The Cybernetic Brain Page 32