The Cybernetic Brain

by Andrew Pickering

  A more detailed version of this same critique acknowledged that there must be some vertical communication within organizations but questioned the automaticity of "cries for help." In the VSM, this was simply a matter of statistical filtration of data. If production indices remained anomalous after an agreed period of time, the algedonic signal automatically passed on to the next level. Werner Ulrich (1981, 51–52) pointed out that in a less automated system there would be a place for management learning—managers come to recognize patterns in the signals arriving at their level and thus to discriminate between which needed to be passed on and which did not—thus protecting the lower levels to some extent from vindictiveness above. I do not know whether Beer ever addressed this point, but, again, the VSM was not exemplary of the cybernetic ontology in action to just the degree to which this automaticity was a fixed part of the VSM.

  3. Following the lines set down by Hanlon, the VSM's critics asserted that the VSM prescribed a "top-down" mode of organizational control: management or government gave orders that the workers were then expected simply to implement. Cybersyn "has some kind of built-in executive power. . . . Its strongly hierarchical organisation and its concept of 'autonomy' one-sidedly serve the top decision maker, the government" (Ulrich 1981, 52, 54). As before, there is something to this critique, but it is worth taking it slowly. Though the critics seem to have read Cybersyn as implementing a classic "command and control" form of organization, with a unilinear flow of orders descending from on high, in this they were wrong. Beer did not think of viable systems in that way. This was precisely the significance of the adaptive couplings that pervaded the VSM, especially the couplings between the various levels. As discussed earlier, these were modelled on the reciprocal vetoing in Ashby's multihomeostat setups and implied that the parties at different levels had to cast around for mutually agreeable initiatives and plans, precisely not the traditional command-and-control mode. These adaptive couplings were the most definitively cybernetic component of the VSM, and it is significant that the critics failed to get to grips with them or even to recognize their distinctive character. Beer often complained that outsiders erred in a similar way concerning all sorts of cybernetic machines and contrivances, utterly failing to grasp their adaptive aspects, and this seems to have been the case here. If ontology makes a difference, then that difference eluded the VSM's critics. But more needs to be said.

  Cybersyn was, on one occasion, operated in both a surveillance and a command-and-control mode. This was the time of the gremiostrike in October 1972, a "CIA-instigated trucker's strike" in Chile (Ulrich 1981, 54n; Beer 2004 [2001], 860) which threatened to halt flows of goods around the country.36 The Cybernet information system was then switched temporarily to monitoring shortages around the country and figuring out how to use the transportation available to overcome them. Beer was very pleased that this approach worked and that the strike was defeated (Beer 1981, 312–15), but there was no homeostatic give-and-take involved in this episode in negotiating plans between different levels, and it serves to show just how readily the organic quality of the VSM could be conjured away, and, indeed, this possibility seems to have appealed to Allende's enemies.37 "At the end of July [1973] . . . several strange messages reached me. . . . They were coming from the political opposition. It seemed that this [Cybersyn] was the best project undertaken under Allende's aegis, and that his (self-assumed) successor would continue it in his own way. This would not, of course, involve any 'nonsense' about worker participation. . . . I found these overtures obnoxious; but our strategies were well prepared" (Beer 1981, 345). The strategies, I believe, were intended to render Cybersyn useless in the event of a coup, but three comments are called for. First, in its genuinely cybernetic aspect—the adaptive couplings between levels—the VSM did serve to undo hierarchies of command and control. Second, these adaptive couplings could easily be "switched off" and replaced by asymmetric ones. It is fair to say, then, that the VSM was hardly a potent bulwark against the institutional arrangements that Beer wanted to obviate. This, too, was much on his critics' minds. But third, as Beer might have advised, we should be concerned here with the future more than the past. Even if key components of the VSM were readily erasable, the VSM remains interesting as a model for a democratic subpolitics.

  4. We can return to the question of goals. In chapters 3 and 4 we looked largely at systems with fixed goals. Ashby's homeostats adapted open-endedly, but so as to keep their essential variables within given limits. According to Beer, the quasi-organic viable system likewise had goals that patterned its adaptation. But, unlike Ashby, Beer was not attempting to construct models of the adaptive brain, and he therefore did not have to take a sharp position on what the goals of a viable system are. I said earlier that one could think of the profitability of an enterprise as the sort of thing at issue, but actually Beer had something different and more interesting in mind, which we can get at via the critique of the VSM. At the heart of Werner Ulrich's (1981, 35) long critique, for example, is a contrast between "purposive" and "purposeful" systems, which relates to a more familar distinction between means and ends: a "purposive" system is a means to some extrinsically specified end, while a "purposeful" one can deliberate on its own ends. Ulrich criticized the VSM as purposive, and at one level this is correct. Beer was keen not to try to build any substantive goals beyond adaptability into the VSM; this is an aspect of what was entailed in my earlier description of the VSM as a form of subpolitics.

  Ulrich, however, went on from this observation to claim that because the VSM had no substantive goals, then whatever goals a system came to manifest would have to be supplied in a top-down fashion, from systems 4 and 5 of the model—we are back to technocracy from a different angle. But here there are some complications worth discussing. One reply would be that Beer was working for a democratically elected government responsive to "the will of the people," but that is an observation about the specific context of Cybersyn rather than an intrinsic feature of the VSM in general. Another reply would go along the lines indicated above: that the adaptive couplings between the VSM's levels are reciprocally adaptive, not one-way. But here, still, some asymmetry remained in the VSM. Beer does not seem to have envisaged the formulation of new plans and goals from below; the higher levels of management and government do seem to have held the advantage here in his thinking (though this assertion will be qualified below when we come to his work on "syntegration," which indeed focused on inclusive processes of goal formation). Nevertheless, Project Cybersyn, as it evolved, did at least try to close the loop between government initiatives and their popular reception in various ways, and I want to examine just one of these.

  On Goals

  In March 1972 . . . we addressed the basic issue of the organization of the state that is not economic but societary. . . . I wrote a second paper about a project to examine:

  "the systems dynamics

  of the interaction

  between government and people

  in the light of newly available technology

  such as TV

  and discoveries in the realm

  of psycho-cybernetics"

  (Beer 1981, 278)

  There were, of course, many channels by which the Chilean government could communicate with the Chilean population at large and vice versa. But the reference to TV immediately suggests an asymmetry. Governments could transmit information over the television in great detail and length—a highvariety channel, in the language of information theory. The people, in contrast, could not reply via the TV at all—an exceedingly low-variety channel. Of course, the people could communicate via other channels, such as forming political parties and voting in elections, but Beer felt that it was necessary to do something to increase the information flow from people to government if a homeostatic equilibrium was to be achieved. He also, as usual, felt that the channel from people to government should be a real-time one, so that the latter could react to how the former felt today rather than last week or last m
onth or last year.38 The solution Beer proposed, novel and endearing, is shown in figure 6.15. The aim here was to supplement the economic algedonic feedback of the VSM with social feedback. TV viewers, for example, would be provided with very simple "algedonic meters" of the form shown in the lower left of figure 6.15. These would be simple semicircular devices in which a partition could be rotated clockwise (toward "happy") or counterclockwise ("unhappy") in response to whatever was happening before them—a televised political speech, say. Some simple wiring arrangements would aggregate these algedonic signals (the precise arrangements being left open in the initial proposal) and transmit them for display in real time on the TV screen. In this way, the politicians would get instantaneous feedback on their proposals or arguments. And—this is the clever bit—the viewers could also see how the politicians would react to the feedback, and so on in a cascade of feedbacks between the TV studio and its audience (Beer 1981, 285). In effect, some channel, however crude, would thus be opened for mass debate—or, better, a dance of agency—with the government. Again, policy making could thus emerge in real-time interaction.

  Figure 6.15.Feedback from the people. Source: S. Beer, Brain of the Firm, 2nd ed. (New York: Wiley, 1981), 281, fig. 45.

  Like many of the cybernetic devices we have been exploring, these algedonic meters of Beer's were at once serious and amusing, and even startling in spanning the gap between the two. Their origins, I would guess, lay in the clapometers and swingometers of the BBC's popular music TV shows and election reporting.39 An interesting feature is that they were truly algedonic in being able to register pleasure as well as pain, unlike the algedonic signals in the basic VSM, which were regarded as warnings that something was wrong. Though Beer initially conceived their use in mass communication, they could obviously be deployed in much more limited contexts—in small meetings, say, where some planning group reported to its constituents, or at factory gates as feedback from the workers to management.

  Beer's son Simon, an electrical engineer, built a prototype system "of ten algedonic meters, linked by a single wire in a loop through a large summation meter" (Beer 1981, 284), and took it out to Chile, where experiments were done on its use with a group of fifteen friends. These friends, however, "rapidly learned how to rig the system. They joined in plots to 'throw' the lecturer by alternating positive and negative responses, for instance" (286). The algedonic meter was, in this instance, too much fun. And one can easily imagine less amusing forms of rigging—the political party instructing its supporters to slam the indicator to the left whatever an opponent said—or even arguments about whether "unhappy" should be at the left or the right. This form of feedback was thus never introduced in Chile, leaving Beer to reflect that its design was a tricky problem and that more cybernetic research was needed. Nevertheless, it is interesting to stay with them just a little longer.

  Beer contrasted his algedometers favorably with another and more familiar form of quasi-real-time feedback from the people to government: questionnaires and opinion polls (Beer 1974a, 334–38). From Beer's perspective, the great thing about the algedonic meters was that they were inarticulate, wordless. They measured "happiness," but the nature of happiness and its causes were left undefined. They simply indicated a positive or negative response on some undefined scale. Beer's enthusiasm for this mode of communication had to do with his intense interest in performance and his associated suspicion of representational knowledge. The trouble with opinion polls, Beer argued, is that the domain of inquiry is circumscribed by the questions asked (themselves framed by politicians, journalists, academics, and so on) and lacks variety. Articulated questions might therefore be able to determine how people feel about specific government policies, but they can never find out whether people's real concerns lie entirely elsewhere. Polls can never contribute, then, to the emergence of real novelty in real-time politics, only to a fine-tuning of the status quo. In contrast, the algedonic meters constituted an open invitation to genuine experiment. If a politician or journalist were to float some wild idea and the integrated meter reading went from lethargically neutral to wildly positive, there would be reason to think that some genuine but hitherto unthought-of social desire had been tapped.

  And here we can return to Ulrich's critique of the VSM as purposive rather than purposeful. Though Beer did not try to build into the VSM any substantive goals, he did try to think through the ways in which the system could articulate its own goals, in practice, in a nonhierarchical fashion. We can think of the algedonic meters as expanding the VSM as a subpolitical diagram of social relations and information flows in such a way as to enable any organization to become purposeful, rather than purposive, on its own terms. Ulrich is wrong here about the VSM, at least in principle, though, as above, practical concerns are not hard to find: it would have been less difficult for General Pinochet and his friends to eliminate algedonic meters than, say, rifles in the hands of the workers.

  One last thought about the algedonic meters. What did they measure? At the individual level, an unanalyzed variable called "happiness." But for the aggregated, social, level Beer coined a new term—eudemony, social well-being (Beer 1974a, 336). Again he had no positive characterization of eudemony, but it is important that he emphasized that it is not any of the usual macrovariables considered by politicians and economists. Eudemony is not, or not necessarily, to be equated with GNP per capita, say, or life expectancy (Beer 1974a, 333). Eudemony is something to be explored in the adaptive performance of a viable social system, and, obviously, Beer's algedonic meters were an integral part of that. This thought is perhaps the most radical aspect of Beer's subpolitics: the idea that social systems might continually find out what their collective ends are, rather than, indeed, having those ends prescribed from above (the wonders of the free market, for example). And this remark gets us back to the general question of cybernetics and goals. Beer's cybernetics, unlike that of Walter and Ashby, did not enshrine any idea of fixed goals around which adaptation was structured. Goals, instead, could becomein Beer's (and Pask's) cybernetics. As ontological theater, then, the VSM staged a vision of open-ended becoming that went an important step beyond that of the first-generation cyberneticians. Beer had not, of course, solved the problem of building a machine that could mimic the human facility of formulating goals; his systems could be adaptive at the level of goal formation precisely because they contained human beings within themselves.

  _ _ _ _ _

  Where does this leave us? After reviewing the critiques of the VSM and Project Cybersyn, I continue to think that we cansee the VSM as enshrining a very interesting approach to what I have called subpolitics. The VSM offers a considered topology of social locations and relations, information flows and transformations that, to a considerable degree, promises a dispersal of autonomy throughout social organizations. The key elements of the VSM, from this perspective, are the adaptive, homeostat-like couplings between the various levels of the VSM, and the algedonic signals that travel back up the system. Like Beer's earlier experimentation with biological computing, his work on the VSM seems original and singular to me. It is hard to think of any equivalents in more conventional approaches to political theory and practice. And for this reason I am inclined to point to the VSM as another item on my list of striking examples of the cybernetic ontology in action, in politics and management. Here again we can see that the cybernetic ontology of unknowability made a difference.

  Turning to the critics, it is significant that they seemed unable ever quite to get the VSM into focus. Beer's overall cybernetic aim, to bolster the adaptability of organizations, was never, as far as I can make out, mentioned by the critics; neither was the key cybernetic idea of adaptive coupling between levels. Instead, the critics focused on a cybernetically denatured version of the VSM, a version from which the distinctively cybernetic elements had been removed, turning it into a nightmare of command and control. The critics mapped the VSM onto a distinctively modern space in which it did not belong, and th
ey found it wanting there. This inability to contemplate the thing in itself I take to be further evidence that ontology makes a difference.40

  Having said that, I have also recognized that the critics' concerns about the VSM were not empty. It does seem clear that systems like that envisaged in Project Cybersyn could be readily stripped down in practice and turned into rather effective systems of command, control, and surveillance, the very opposite of what both Beer and the critics aimed at. But as I have said before, the object of this book is not to resurrect any specific cybernetic project, including Cybersyn. It is to exhibit and examine a whole range of such projects—as a demonstration of their possibility and their difference from more conventional projects in cognate areas, and as models for the future. A future cybernetic politics that followed Beer's lead into subpolitics might well want to bear in mind the democratic fragility of the VSM—while contemplating algedonic meters as, shall we say, a desperate but entertaining attempt to open up a politically deadening status quo.

  Pinochet's coup in Chile was not the end of Beer's involvement in politics at the governmental level, especially in Central and South America. He went on to consult for the governments of Mexico, Venezuala, and Uruguay, as well as, in other directions from the United States, Canada, India, and Israel (Beer 1990a, 318–21), and "bits and pieces of the holistic approach have been adopted in various other countries, but by definition they lack cohesion" (Beer 2004 [2001], 861).41 I will not pursue that line of development further here; instead, I want to explore Beer's cybernetic politics from another angle.