---
title: "Ch16 objections that land content"
description: "In 1981 Werner Ulrich publishes twenty-seven pages in the Journal of Applied Systems Analysis, putting to Project Cybersyn the questions C. West Churchman had…"
url: "https://wiki.cybernetic-futures.org/punch-and-judy/part-02-outside-the-model/ch16-objections-that-land/ch16-objections-that-land-content"
modified: "2026-11-12T15:26:25.914Z"
language: "en"
site: "Cybernetic Futures"
---

# The Objections That Land, and What They Cost

In 1981 Werner Ulrich publishes twenty-seven pages in the *Journal of Applied Systems Analysis*, putting to Project Cybersyn the questions C. West Churchman had formulated for any systems study. His conclusion is one sentence: "Cybersyn's alleged goal of devolving power to workers is not confirmed by its design." Beer sent his comments on tape, and the tape, if it survives, has not been found.

Apart from five pages Beer printed two years later, which we have not yet read, that is the whole of this field's reply. In the same year the second edition of *Brain of the Firm* addressed the charge of technocracy in general terms, conceding that the participatory design "was not always being done in the intended spirit"; it does not mention Ulrich and does not answer his argument about the client.[^ch16-beer-techno] The objections in the pages that follow deserved a better reply.[^ch16-opener]

## What lands, and by what standard

The chapters before this one kept us company: neighboring literatures that arrived at our components without us, mostly without knowing we had them. What follows is opposition, and a reviewer will turn here first.

Three objections land: Werner Ulrich's, F. A. Hayek's and Robert Cooper's. We take them first, in that order, before the positions that do not land, and they turn out to be one wound seen from three directions. It is a wound about boundaries.

The standard is narrow, and it has to be stated, because a chapter of this kind can reach any verdict by adjusting what counts as a hit. An objection lands here when it identifies something the Viable System Model↗[^ch16-glossary] needs in order to do the job it claims, cannot supply from its own resources, and has not been given by any amendment since. An objection that prices an amendment (an added index, a demoted claim, a stated scope condition) does not land in that sense, however uncomfortable it is to receive. We want to tell a refutation from a repair bill, and to pay the repair bill rather than argue with it.

In one respect all the positions collected here are of the more serious kind. None of them thinks the model merely imprecise or under-tested. They think the job it proposes is not well posed: there is no bounded steerable object to diagnose, or no vantage point from which to diagnose it, or no innocent way to draw its edges. If any of them is right in that strong form, better measurement makes things worse, because it refines an instrument aimed at nothing.

One structural fact belongs before the detail. It is the strongest defensive resource this book has, and it is a fact, not a rhetorical move: these positions are mutually inconsistent. Cooper and Burrell classify Luhmann among the modernists.[^ch16-cooperburrell] DeLanda is a realist about emergent wholes at several scales, which actor-network theory denies; actor-network theory abolishes the levels DeLanda keeps. Ulrich and Jackson both presuppose that a systems study is a legitimate thing to do properly, which the process theorists deny. The model does not have to survive all of them, only whichever of them survive each other.

:::: {.note-on-language}
**A note on language**

This chapter borrows four vocabularies, and each of them uses a familiar word in an unfamiliar way.

Luhmann's *communication* is not people talking: it is the unit a social system is made of, and it excludes the people. His *operational closure* does not mean isolation; it means that no operation begins in one system and completes in another. DeLanda's *relations of exteriority* is the claim that a component is not constituted by the role it plays in a whole — which is why he can detach it and the model, as we teach it, cannot. Ulrich's *boundary judgment* is not a boundary but the decision, usually unnoticed and always somebody's, about where one goes. And Jackson's *coercive* classifies a context in which interests cannot be reconciled; it is not an accusation against anyone standing in it.

Where these authors and this book use the same word for different things, we have kept theirs and marked ours.
::::

After the three that land come four positions that do not land in the form they are usually given: Luhmann's operational closure, the assemblage and network arguments about levels, Michael Jackson's placement of the model within a system of methodologies, and the wider process argument of which Cooper's boundary point is one part. For each we say what it does establish. Each converts into something we owe, and the closing section prices the only one of them that can be paid immediately.

## Ulrich: whose boundary, and who witnesses

Werner Ulrich is the sharpest of these critics because he does not deny that systems modeling works. He denies that it can be innocent.

Nor is he a distant critic. His 1981 case study, built on Beer's own reports and checked by an insider,[^ch16-ulrich] finds that the design does not do what its designer said it did. The client Beer declared was the workers; the client the design serves is the government, which alone can act on the system's reports without a homeostat↗ above it, and an algedonic↗ signal is a cry of pain and also a transfer of control upward. Hence the sentence in the paper that is hardest to answer: "what distinguishes Cybersyn from a traditional state bureaucracy is not a less hierarchical structure but the greater speed at which vertical communication (and control) takes place." That is the objection Part Four is built to answer, made in 1981 and aimed at the founder.

Beer's own pages answer part of this and concede part. The design was meant to be the reverse of a transfer of control: "the schema represents maximum decentralization, since any given level of recursion receives directly only its own Cyberstride reports", and the algedonic signal was reserved for a level that failed to deal with its own trouble, so that "if all concerned do their agreed jobs properly, algedonic signals will rarely be fired."[^ch16-beer-algedonic] He had also conceded, as the opening of this chapter records, that the participatory design was not always kept, and he reported that when the Minister drew back from Cybersyn in December 1972 the gratification "was experienced by the technocratically minded, who wanted effective management regardless of the political framework."[^ch16-beer-technocrats] Ulrich's point survives as a claim about who sets the filters and the thresholds, a question these passages do not settle.

Chapter 8, *The Loop That Stayed an Intention*, narrates the episode, including the taped comments Beer sent. Beer also published a short reply in the same journal in 1983, which we have not yet read. Here the story matters less than the standing of the argument, and that is undiminished. The paper was revised in the light of a participant's comments and checked by another. It makes no claim about the political purpose the project served, and says so twice. It was aimed at a designer who was alive, working and perfectly capable of answering it. Beyond Beer's own five pages, the reply was ours to write, and we did not write it.

The critique generalized into a method. Critical systems heuristics asks twelve boundary questions, each in an *is* and an *ought* mode: who is and ought to be the client; what is and ought to be the purpose; what is and ought to be the measure of improvement; who decides and what they control; who counts as an expert and what counts as relevant knowledge; what is assumed to be the guarantor of success; and, at the end, who is witness to the interests of those affected but not involved, and what secures their emancipation from the premises of those who are.[^ch16-csh] Chapter 4, *Channels Without Owners*, sets those questions beside the channels the model does specify. Here we ask them of the diagnostic act itself, where they were always pointed.

Applied to us, they are uncomfortable in a precise way. The model leaves the naming of identity to the eye of the beholder and assigns the naming only loosely. Purpose is imputed by "a GOOD OBSERVER" and stated for the system by the higher recursion; the System-in-focus is chosen by the diagnostician; Cybersyn's measures were to be agreed with the people in each plant.[^ch16-beer-naming] The model does not say who is *entitled* to do any of this, how it is recorded, or how a disagreement is settled. It does not say who fixes the recursion levels↗, or on whose authority. It does not say whose measure of improvement the System One↗ indices encode, though they always encode somebody's, and Beer's own Chile account concedes that the participatory design was not always kept. Its assumed guarantor of success is either viability itself or the model's own theorems. Both are internal to the designer's frame, so neither can justify anything to someone standing outside it. Beer's own stated guarantor is predictive utility, which could lie outside the frame; the tradition has not specified the predictions that would put it there.

The tenth question is this chapter's line. The algedonic channel has no seat for the person who is affected by an operation and was never a member of it. Pain is signaled by the units that feel it from inside the system. The neighbor downstream of a plant is not a unit, has no channel, and appears in the design only as environment.

Now what the objection establishes, and what it does not. It does not establish incompatibility. Nothing in critical systems heuristics contradicts the Viable System Model; the two operate at different logical levels, and Ulrich interrogates the boundary that Beer designs inside. Nor does it establish that the model conceals boundary judgments deliberately, and Ulrich does not say so. It establishes that a diagnosis produced with our tools does not justify itself, and that the missing justification takes the form of answerable questions with names attached. System Five↗ and the algedonic channel are, in intent, a witness function. The model has the slot. Beer states who ought to be there, "All stakeholders are represented in System Five," with a consumers' council among his examples,[^ch16-beer-s5] but he gives no procedure by which those affected and not involved obtain or use the seat, and we have not supplied one. That defect can be fixed, and fixing it is the cheapest high-value reform available to this tradition. The receipt at the end of the chapter buys exactly this and nothing else.

## Hayek: the knowledge that cannot be sent

Hayek put the second objection in one sentence: "The sort of knowledge with which I have been concerned is knowledge of the kind which by its nature cannot enter into statistics and therefore cannot be conveyed to any central authority in statistical form."[^ch16-hayek] It has never been answered, in this tradition or, so far as we can find, anywhere adjacent to it.

The objection comes in layers, and treating it as one is how we keep failing to meet it. Three layers are Hayek's own. The last two belong to his modern defenders, and Cosma Shalizi has put them most sharply.[^ch16-shalizi-planning] First, the knowledge is tacit, constituted in the act of using it, not merely dispersed. Second, it is *discovered* rather than transmitted: the data a regulator would need does not exist until rivalrous action generates it.[^ch16-discovery] Third, there is a measure problem. Indices must be chosen in advance. Beer's actuality-capability-potentiality triple is fixed beforehand, by the analysis teams and, by design if not always in practice, the plant's workers, and Beer bounds potentiality by "what is already known to be feasible,"[^ch16-beer-potential] while Hayek's circumstances of time and place are by definition the ones the modeler did not anticipate. Fourth, there is an objective-function problem. Even with perfect information, weighting goods against each other requires preferences that differ between people and that no mechanism reconciles satisfactorily. In the VSM that weighting has to be smuggled into System Five, which Beer equipped with deliberative devices, Team Syntegrity and the algedonic meter, and with no procedure that reconciles divergent preferences and could claim legitimacy for the result.[^ch16-beer-deliberate] Last comes the computational layer, the weakest of the five and the one most often mistaken for the whole. When our literature answers the objection at all, it answers this layer, because it is the only one that hardware improvements can address.

Note where the third layer lands. Hayek's measure problem and Ulrich's third boundary question are the same question, asked by an economist and a philosopher who would agree on nothing else: whose measure of improvement is this, and what secures it against the improvements it cannot register? When two hostile literatures converge on a point, the point is load-bearing.

Beer shared Hayek's premise. He wrote that the prerogative to intervene indiscriminately "does not have Requisite Variety," and he answered it with autonomy rather than with central knowledge.[^ch16-beer-hayek] The best defense available to us is that Cybersyn was not built to compute a plan. It attenuated↗, and in that it was a Hayekian architecture, leaving the decision to the person on the spot and escalating only the residue. Chapter 8 sets out this reading of the design, and we still think it is the right reading of the intent.

The historical record neither confirms nor refutes the defense. By Beer's own count the system ran across some four hundred enterprises, but never under conditions that could test the claims at issue: the operations room was only "in experimental working order" in January 1973.[^ch16-beer-scale] Cosma Shalizi's empirical verdict, that to call Cybersyn a decentralized planning system "is nonsense" and that at best it would have been a dashboard, is set out in Chapter 8.[^ch16-cybersyn] Beer's first-person account reads October 1972 as filtration, with some signals algedonic and others attenuated, and he staged and rejected the objection that the computers were "merely frills."[^ch16-beer-frills] We have no evidential answer to either reading, because the trial ended before it could produce evidence either way.

So the measure-problem layer stands unanswered. Preselected indices cannot certify that they captured the surprises nobody selected for. No amount of real-time reporting closes that gap, because the gap lies in the choice of what to report, and the choice is made before the surprise arrives. We do not have a reply to this. We have a research program that would begin to constitute one, and Part Five prices its first two steps.

One counter-argument has never been properly staged, and staging it falls to us. Hayek's price system is itself a variety attenuator, carrying a single scalar per commodity. The Ashbyan question is whether one scalar has requisite variety↗ for every regulatory task, and whether its settling time is short enough for every disturbance. Beer came close. He argued that a money-supply regulator lacks requisite variety for an economy that learns, and called money "the metric of *constraint*."[^ch16-beer-money] Nobody, Beer included, has applied the argument carefully to Hayek's price mechanism itself. It is an opening, not yet a citation, and it is the sort of work this book exists to commission.

## Cooper: the act that manufactures the object

The third objection that lands is the least often stated and the hardest to shake off. It belongs to a family of process arguments, most of which do not land, so the one that does has to be separated from the rest.

"Change must not be thought of as a property of organization," write Tsoukas and Chia. "Rather, organization must be understood as an emergent property of change."[^ch16-whitehead] The family position is that an entity is an artifact of a description, and that the description moves more slowly than what it describes. Applied here: by the time the System Ones have been identified, the boundaries drawn and the indices chosen, the flux has reorganized. A diagnosis produces a system-effect rather than a snapshot of a system. Then it feeds back, because people rearrange themselves to fit the model, and the model's apparent accuracy becomes a performative achievement rather than a finding.

Robert Cooper adds the edge that does the damage. In our compression of his argument, the diagnostician's first act, drawing the boundary, is the act that manufactures the object, and it is exactly the act for which the method supplies no procedure of examination.[^ch16-cooper] His complaint against systems theory is that it most neglects the concept which originates system: boundary, division.

Beer named the act. "A system is not something given in nature, but something defined by intelligence"; two observers "will not necessarily agree on … the boundaries"; the analyst should "not gaily accept the system delineated by the conventions of management structure." His workbook makes the boundary its first task.[^ch16-beer-cut] He even revised one of his own first cuts in print, the Chilean model that had mixed two levels of recursion, and what triggered the revision was events, not a method.[^ch16-beer-chile]

Against the general form of the argument Beer built a defense in advance, and it is a real one. The Three-Four homeostat is the response to a description that lags its object. System Four↗'s business is continuous re-modeling, and the insistence on real-time data and short settling times is an attack on lag as such. Process philosophy names a problem the model was partly designed against. It also offers no criterion for telling a better abstraction from a worse one, and some abstractions kill people when they fail. Whitehead's fallacy, in any case, was mistaking the abstract for the concrete, not abstracting.[^ch16-misplaced]

Cooper's version survives all of that, because it concerns the first cut rather than lag. Nothing in the Three-Four homeostat examines the act that produced the object the homeostat then tracks, and no amount of faster re-modeling reaches back to it. Part One has already made this concession in its own terms, and the convergence is now visible. Ulrich says the first cut is somebody's and should be published with a name on it. Hayek says the first cut fixes what can afterward be measured. Cooper says the first cut makes the object. Three literatures with nothing else in common have found the same act at the start of our method: named by Beer as the observer's, revised once in public, and never given a procedure.

## Luhmann: nothing crosses the boundary

The remaining positions are serious and do not land in the form they are usually given. Luhmann's is the most metaphysically ambitious.

Niklas Luhmann's social systems consist of communications, not of people or machines or actions. Psychic systems are separately closed and belong to the environment. Organizations reproduce themselves out of *decisions*, communications that announce their own contingency, that they could have been otherwise. Their structures are decision premises: programs, channels, personnel, and the undecidable premises we call culture.[^ch16-luhmann] Operational closure carries the weight. No operation begins in one system and completes in another. Information is not transported; the system produces it internally, by its own distinctions.

Follow that through the VSM and the damage is specific. On this account a System One operational unit (people, plant, process, making the thing the enterprise exists to make) is not a unit of the social system at all. System Three↗ presupposes a position from which the operations it regulates can be observed. But management is itself decision-making, one more run of communications inside the same autopoiesis, not a stratum above it. "System Three regulates System One" is a self-description the organization produces about itself, and the model mistakes the self-description for the structure.

System Four takes the sharper damage. Environmental scanning does not deliver the environment; it produces internal communications about a distinction the system itself drew. So the celebrated homeostat between outside-and-then and inside-and-now couples no internal reality to an external one. It is an internal oscillation between two self-produced descriptions, with no channel by which the environment could correct either. Beer's requisite-variety justification for System Four is stated in *The Heart of Enterprise* as if it assumed exactly the access closure denies, although by 1980 Beer had accepted that "we observers invent the codes" and had defined closure as "a *self-referential process*, and not the isolation of the system within an adiabatic shell."[^ch16-beer-codes] And System Five, on this reading, does not close the system. It is where the system de-paradoxifies its own founding paradox, that decisions are only needed where the matter is undecidable: the fiction by which it conceals its contingency.

The reply has four parts. The first is that "cannot be steered" is a strong reading, and Luhmannians dispute it among themselves. Helmut Willke built an entire steering theory inside the tradition on *Kontextsteuerung*: you cannot steer a closed system directly, but you can alter the contexts and couplings within which it steers itself. That is a description of what System Three actually does. System Three commands little. Besides the command channel, on which it "exerts authority",[^ch16-beer-command] it sets the resource bargain, the accountability and the standards, the context within which System One regulates itself. Beer's principle of autonomy and Willke's context steering are the same idea, arrived at from opposite ends.[^ch16-willke]

The second is that closure permits a great deal. Structural coupling means systems can be durably linked so that each is a standing source of irritation for the other; contracts, budgets, membership and technology are structural couplings. Second-order observation, observing observers and one's own self-descriptions, can be institutionalized as routine, which is functionally close to what we call Systems Four and Five. Organizations change themselves by deciding about the premises of future decisions, which is not far from what we call design. And for Luhmann organizations are the only social systems that can act as units and be addressed as actors. That last point helps us.

The third is that the integration has already been done, and we mostly have not read it. Achterbergh and Vriens devote parts of one book to Luhmann and to Beer and argue that the two can be brought into a single framework. Johnson and Leydesdorff staged the confrontation directly, reporting Luhmann's position that businesses "do not 'live' as viable systems but operate because they relieve human beings from environmental complexity". Schwaninger and Groesser used operational closure and self-reference as *cybernetic* tools to explain organizational change.[^ch16-integration] Luhmann's own debts are cybernetic in any case: Ashby, von Foerster, Maturana and Varela, Spencer-Brown.[^ch16-luhmann-debts] Reduction of complexity is a variety argument. The dispute is over the unit of analysis, not over whether regulation occurs.

The fourth is that the argument proves too much. If it holds, medicine, engineering and Luhmann's own sociology are equally self-description. A book about practice may fairly reply that a theory under which all intervention is impossible has not explained why some interventions reliably work.

What the objection does establish is a matter of grammar, and it is not trivial. It lands cleanly against an ontology, the claim that an organization *is* a viable system with five subsystems. It glances off a design language that offers a variety-based test of whether a set of governance arrangements can absorb the disturbances it faces. The cost to us is to stop making the first kind of claim, and to read the integration work before anyone in this tradition asserts incompatibility again.

## Assemblage and network: the levels are not there

The next two positions attack the same structural feature from opposite directions, and both convert into amendments we should make.

Manuel DeLanda's assemblage theory insists on relations of *exteriority*: "a component part of an assemblage may be detached from it and plugged into a different assemblage in which its interactions are different", and "the properties of the component parts can never explain the relations which constitute a whole." A component's role in a larger whole does not define it.[^ch16-delanda] System One, as we teach it, denies exactly this. A unit is an operational element because it produces what the organization exists to produce, and its function in the whole is taken to confer its identity. Those are relations of interiority, the organicist model assemblage theory was built to destroy. Beer's own text is less organicist than our teaching. He calls a subsidiary "a viable entity in itself" that belongs to its parent only "in a specially defined way", and he reports a divisional chairman "secretly moving to sell his division."[^ch16-beer-detach] The model assigns a unit its role by function; its identity, on Beer's page, is its own, and DeLanda's point lands on how the role is assigned. The emblematic bad object of that tradition is the arborescent diagram, a root and branches, and the VSM diagram is usually drawn as one, although Beer wrote that its vertical dimension "has logical but not organizational implications."[^ch16-arborescent][^ch16-beer-tree]

But DeLanda does not abolish levels. He builds an explicitly multi-scale social ontology in which persons, networks, organizations, cities and states are each real and irreducible. What he denies is that levels are internally *constitutive*. That is close to Beer's principle of autonomy, on which each System One is a viable system in its own right, with its own identity and purpose, whose cohesion with the whole is bought rather than assumed.[^ch16-autonomy] DeLanda's territorialization axis measures precisely how detachable a component is: cheap for a contractor, expensive for a licensed plant. The model does not assume uniform embeddedness in Beer's text: systems of concentrated purpose are "closely-knit," those of general purpose "more loosely coherent," and freedom is "in principle a computable function" of purpose.[^ch16-beer-cohesion] Nobody has computed it, and our diagnoses treat embeddedness as uniform. The repair is still an index, now as the measure of a quantity Beer named and left uncomputed, and the model stays. A diagnosis should record how detachable each unit is, and stop treating a franchise, a subsidiary and a regulated utility as the same kind of part.

Actor-network theory goes further and flattens everything. "Macro no longer describes a wider or a larger site in which the micro would be embedded like some Russian Matryoshka doll, but another equally local, equally micro place, which is connected to many others through some medium transporting specific types of traces"; "size and zoom should not be confused with connectedness."[^ch16-latour-scale] Without a scale ordering, recursion↗ would have nothing to recurse over, if recursion were a ladder of sizes. Beer's is ordered by the containment of viable systems, not by size or zoom, and he warned against reading it as organizational level.[^ch16-beer-recursion] Latour's scale point therefore misses Beer's definition. What it hits is the claim that every level is a viable system in its own right.

Except that the oligopticon concedes Beer's mechanism. Latour's centers of calculation accumulate inscriptions and so act at a distance while seeing very little, very well. Cyberstride and the operations room are centers of calculation in precisely his sense.[^ch16-oligopticon] Actor-network theory does not deny that they work; it denies that they work by being *above*. Beer's own account says a metasystem↗ works by attenuating, by throwing almost everything away. That is Latour's oligopticon in Ashbyan clothing, and the objection converts into a friendly amendment that takes Beer at his word. He called the recursive description "a vast attenuator of the variety proliferated by any large organization."[^ch16-beer-attenuator] Keep recursion as that bookkeeping of who attenuates what for whom, and drop the ontological ladder our teaching has built on it.

## Jackson: a placement, not a verdict

Michael Jackson's system of systems methodologies, first published with Paul Keys in 1984 and developed since, sorts approaches by the complexity of the system and the relations among participants: unitary, pluralist, and in the later versions coercive. It places organizational cybernetics in the complex-unitary cell. The characterization that gets quoted is that the VSM is a "unitary, functionalist" approach, "unable to deal with individual, social and political issues": useful where interests genuinely converge, untrustworthy where they do not, and in coercive contexts liable to make efficiency into the efficiency of domination.[^ch16-sosm]

Two things about this are usually got wrong, and both errors are ours. The first is the tone. Jackson's 1988 paper on Beer is titled an *appreciation*, his 2019 treatment devotes a section to the value of organizational cybernetics to managers, and his stated project has always been complementarism rather than elimination. A chapter that reads him as an executioner will be corrected in review, and will deserve it. The second is that the exchange is live and recent. Angela Espinosa, writing in a Festschrift for Jackson, argues that he has become more supportive over the decades and that Beer's successors have addressed several of the original criticisms, and she proposes a posture she calls critical and empathetic. Other recent work has taken the power criticism seriously enough to graft an explicit power framework onto the model. Beer himself published his own account of the model's provenance and pathology in the same 1984 volume that carried Jackson and Keys.[^ch16-jackson-recent]

In one place the classification is weaker than it looks: "unitary" may be a category error. The principle of autonomy and the recursive system theorem make each System One a viable system with its own identity and purpose. The model *assumes* divergent sub-unit purposes; that is why the anti-oscillation function and the resource bargain exist at all. Whether a resource bargain amounts to pluralism is a real question and not yet a settled verdict. And a grid that sorts methodologies into cells is itself an exercise in boundary judgment, with premises of its own that Ulrich's twelve questions would find. What we owe here is modest and overdue: state the placement, and stop presenting the model as a method to be executed in every context.

## Which of these land

Three, and they land on the same place.

Ulrich lands, and the concession is the boundary argument Part One has already made: the first cut is somebody's, and we do not publish whose. Hayek's measure problem lands, and it is Ulrich's third question in an economist's register: the indices are chosen before the surprise, by the person who did not anticipate it. Cooper's point about the boundary lands, and it is the same concession a third time, stated as an act instead of a question or a measure. One defect, three descriptions: an act Beer named and never gave a procedure, and no reply from us to any of them.

The rest price amendments. Luhmann's objection lands against an ontology and glances off a design language; the cost is to stop making ontological claims and to read the integration work that already exists. Assemblage theory buys a territorialization index, the measure of an embeddedness Beer said varies and our diagnoses treat as uniform. Actor-network theory buys the demotion of recursion from a ladder of being to a record of who attenuates what for whom. Jackson's classification places the model and does not refute it, and its author has been saying so for decades. Process philosophy, minus Cooper's boundary point, describes a problem the Three-Four homeostat was built for.

## The priced receipt

**Receipt — publish the boundary judgments ·** Require every published diagnosis to answer Ulrich's twelve questions in both *is* and *ought* modes, including question 10, and to publish the boundary judgments beside the system map. Retrofitting one completed diagnosis is a five-day desk exercise; adding the protocol prospectively is roughly two interview-days per engagement. If competent analysts cannot answer the questions from the engagement record, the diagnosis is not reproducible enough to support its boundary claims. If the answers expose affected parties without a witness role, the design must record that as a defect rather than hide it in the system boundary.

That receipt addresses the external critics, and only the first of the three wounds. It buys nothing against the measure problem, which needs evidence rather than a protocol. Nor does it buy anything against Cooper, who is answered only by examining the first cut, not by documenting it after the fact. The hardest political charge is internal, what this tradition itself says about freedom and control, and Part Four holds it open rather than pretending this chapter has paid it.

[^ch16-cooperburrell]: Robert Cooper and Gibson Burrell, "Modernism, Postmodernism and Organizational Analysis: An Introduction," *Organization Studies* 9, no. 1 (1988): 91–112. Luhmann appears there alongside Bell and Habermas as an exemplar of modernist discourse, which is the classification the text relies on.

[^ch16-luhmann]: Niklas Luhmann, *Social Systems*, trans. John Bednarz Jr. with Dirk Baecker (Stanford University Press, 1995); Niklas Luhmann, *Organization and Decision*, ed. Dirk Baecker, trans. Rhodes Barrett (Cambridge University Press, 2018). The account of organizations as reproducing themselves out of decisions, and of decision premises as programs, communication channels and personnel, is from the latter.

[^ch16-willke]: Helmut Willke, *Systemtheorie entwickelter Gesellschaften: Dynamik und Riskanz moderner gesellschaftlicher Selbstorganisation* (Juventa, 1989); Helmut Willke, *Systemtheorie III: Steuerungstheorie; Grundzüge einer Theorie der Steuerung komplexer Sozialsysteme* (Gustav Fischer, 1995). *Kontextsteuerung* is Willke's term. The identification of it with Beer's principle of autonomy is ours: we have not found a place where Willke discusses Beer, and he should not be read as endorsing the equation.

[^ch16-integration]: Jan Achterbergh and Dirk Vriens, *Organizations: Social Systems Conducting Experiments* (Springer, 2009); Mark William Johnson and Loet Leydesdorff, "Beer's Viable System Model and Luhmann's Communication Theory: 'Organizations' from the Perspective of Meta-Games," *Systems Research and Behavioral Science* 32, no. 3 (2015): 266–82; Markus Schwaninger and Stefan N. Groesser, "Operational Closure and Self-Reference: On the Logic of Organizational Change," *Systems Research and Behavioral Science* 29, no. 4 (2012): 342–67. The quoted words about businesses that "do not 'live' as viable systems but operate because they relieve human beings from environmental complexity" are Johnson and Leydesdorff's own summary of Luhmann's position and not a quotation from Luhmann.

[^ch16-luhmann-debts]: W. Ross Ashby, *An Introduction to Cybernetics* (Chapman & Hall, 1956); Heinz von Foerster, ed., *Cybernetics of Cybernetics; or, The Control of Control and the Communication of Communication* (Biological Computer Laboratory, University of Illinois, 1974); Humberto R. Maturana and Francisco J. Varela, *Autopoiesis and Cognition: The Realization of the Living* (D. Reidel, 1980); George Spencer-Brown, *Laws of Form* (George Allen and Unwin, 1969).

[^ch16-delanda]: Manuel DeLanda, *A New Philosophy of Society: Assemblage Theory and Social Complexity* (Continuum, 2006), 10–11. Both quoted sentences come from his statement of relations of exteriority in the opening chapter.

[^ch16-arborescent]: Gilles Deleuze and Félix Guattari, *A Thousand Plateaus: Capitalism and Schizophrenia*, trans. Brian Massumi (University of Minnesota Press, 1987), 3–25, originally published as *Mille plateaux* (Éditions de Minuit, 1980). The arborescent-rhizome opposition is theirs, not DeLanda's; he inherits it. We name them here because the text uses the term without naming them.

[^ch16-autonomy]: Stafford Beer, *The Heart of Enterprise* (John Wiley & Sons, 1979), 158, for what Beer calls "the cybernetic argument for autonomy (or freedom)", which is what the text means by the principle of autonomy — Beer does not count it among his Principles of Organization — and 118 for the recursive system theorem; DeLanda, *A New Philosophy of Society*, for territorialization and deterritorialization as an axis of variation.

[^ch16-latour-scale]: Bruno Latour, *Reassembling the Social: An Introduction to Actor-Network-Theory* (Oxford University Press, 2005), 176 and 187. Both quoted sentences belong to the "First Move: Localizing the Global" section of Part II (173–90).

[^ch16-oligopticon]: Bruno Latour, *Science in Action: How to Follow Scientists and Engineers through Society* (Harvard University Press, 1987), chap. 6, "Centres of Calculation," 215–57, for centers of calculation; Latour, *Reassembling the Social*, 181–83, for the oligopticon, which sees "much too little" but sees it well (181). The application of either to Cyberstride and the operations room is ours, not Latour's; he does not discuss Cybersyn.

[^ch16-ulrich]: Werner Ulrich, "A Critique of Pure Cybernetic Reason: The Chilean Experience with Cybernetics," *Journal of Applied Systems Analysis* 8 (1981): 33–59. A Spanish translation appeared as *Una crítica de la razón cibernética pura (el caso del proyecto CYBERSYN)*, Memos de Investigación 36 (Universidad de los Andes, 1989). Earlier drafts of this book recorded that we had not been able to obtain the text and were relying on Ulrich's own one-line description of it; we have read it since, and this note replaces that admission. The paper has two halves. The first argues that systems theory has been absorbed by a "scientistic" school which mistakes theoretical for practical reason, and proposes a critically-normative paradigm against what Ulrich names the *syntactic fallacy* — taking Ashby's law as a criterion of systems rationality when the level of organization is socio-cultural rather than organic. The second applies five of Churchman's categories — client, purpose, decision maker, expert, guarantor — to Cybersyn, and concludes that "Cybersyn's alleged goal of devolving power to workers is not confirmed by its design. Instead, its strongly hierarchical organisation and its concept of 'autonomy' one-sidedly serve the top decision maker, the government." Three things should be said about its standing. Ulrich states at the outset that the political purpose Cybersyn served is *not* the subject of his critique — "It is not my intent to question the social and democratic ideals that both Salvador Allende and Stafford Beer associated with the project" — and closes by regretting that his argument falls on "one of the very rare efforts undertaken by systems scientists in the service of democratic values." The account was revised after comments from Fernando Flores, Allende's Secretary of Economics, and its accuracy was confirmed by "an anonymous reviewer who was involved in the project"; where Ulrich contradicts the contemporary *New Scientist* coverage, he does so on Flores's authority. And on the question of a reply: the acknowledgements record that "Stafford Beer too was kind enough to transmit his oral comments on tape to me, though he could not agree with my critique." Beer did answer in print: Stafford Beer, "A Reply to Ulrich's 'Critique of Pure Cybernetic Reason: The Chilean Experience with Cybernetics,'" *Journal of Applied Systems Analysis* 10 (1983): 115–19, listed and quoted in M. C. Jackson, "An Appreciation of Stafford Beer's 'Viable System' Viewpoint on Managerial Practice," *Journal of Management Studies* 25, no. 6 (1988): 557–73, at 572 and 564; we have not read it, and earlier drafts of this book said no reply had been published. Beer's letter in *New Scientist* of 22 March 1973 answers Herbert Grosch's letter of 15 March, not Ulrich (Eden Medina, *Cybernetic Revolutionaries: Technology and Politics in Allende's Chile* (MIT Press, 2011), 186–87 and 289 n. 55), and the tape, if it survives, would be worth finding; the search is priced once, in the receipt of Chapter 8. One limit of the paper is worth stating in its favor as well as against it: it is built on *Decision and Control*, *Brain of the Firm*, *Platform for Change*, the Fanfare lecture (the Richard Goodman Memorial Lecture, Brighton Polytechnic, 14 February 1973, reprinted in *Platform for Change*, 421–52) and the 1974 Zaheer lecture, and engages neither *The Heart of Enterprise* nor *Designing Freedom*, so it does not meet the Beer of Chapter 26, *Freedom as a Quantity*.

[^ch16-csh]: Werner Ulrich, *Critical Heuristics of Social Planning: A New Approach to Practical Philosophy* (Paul Haupt, 1983); Werner Ulrich and Martin Reynolds, "Critical Systems Heuristics," in *Systems Approaches to Managing Change: A Practical Guide*, ed. Martin Reynolds and Sue Holwell (Springer, 2010), 243–92. The ordering used here is Ulrich's own: the twelve questions fall in four groups of three — sources of motivation, of power, of knowledge, of legitimation — and each is asked in an *is* and an *ought* mode. The measure of improvement is question 3, the guarantor question 9, and the witness to those affected but not involved question 10.

[^ch16-sosm]: Michael C. Jackson and Paul Keys, "Towards a System of Systems Methodologies," *Journal of the Operational Research Society* 35, no. 6 (1984): 473–86. The 1984 grid has two axes only: systems, simple or complex, and decision-makers, unitary or pluralist. Coercive contexts are named in that paper and deliberately left outside the grid (483, the page Jackson himself cites for it in 2019), on the ground that no systems methodology then addressed them, so the coercive column belongs to the later restatements — Michael C. Jackson, "Beyond a System of Systems Methodologies," *Journal of the Operational Research Society* 41, no. 8 (1990): 657–68; *Systems Thinking: Creative Holism for Managers* (John Wiley & Sons, 2003); and *Critical Systems Thinking and the Management of Complexity* (John Wiley & Sons, 2019), which further adds a "complicated" row and renames the horizontal axis from participants to stakeholders (163–64). We have not been able to fix from a primary source which publication carried the coercive column first: Jackson's 2019 account says only that "later renditions of the grid explicitly included 'coercive' problem contexts," and dates the grid's stable form to his *Systems Methodology for the Management Sciences* (Plenum, 1991) and *Systems Approaches to Management* (Kluwer/Plenum, 2000) (157–59). We have therefore left the text's "in the later versions" unqualified. The quoted characterization — a "unitary, functionalist" approach "unable to deal with individual, social and political issues" — is Angela Espinosa's rendering of Jackson's criticism, and is the form in which it circulates; see the next note.

[^ch16-jackson-recent]: M. C. Jackson, "An Appreciation of Stafford Beer's 'Viable System' Viewpoint on Managerial Practice," *Journal of Management Studies* 25, no. 6 (1988): 557–73; Jackson, *Critical Systems Thinking and the Management of Complexity*, §13.6, "The Value of Organizational Cybernetics to Managers," 339–40; Angela Espinosa, "Revisiting the Viable System Model as an Emancipatory Systems Approach," *Systems Research and Behavioral Science* 42, no. 1 (2025): 171–88, in the festschrift introduced by Ashish Dwivedi and Gerald Midgley, "Festschrift for Mike Jackson," *Systems Research and Behavioral Science* 42, no. 1 (2025): 3–10; Nahed T. Zeini, "The Viable System Model Through a Power-Relations Lens: Towards a Power-Aware Conceptual Framework," *Systems Research and Behavioral Science* 43, no. 3 (2026): 1133–43; Stafford Beer, "The Viable System Model: Its Provenance, Development, Methodology and Pathology," *Journal of the Operational Research Society* 35, no. 1 (1984): 7–25. Espinosa's own phrase is a "critical empathetic approach." Beer's paper is in issue 1 and Jackson and Keys's in issue 6 of the same volume 35, which is what "the same 1984 volume" means; they are not in the same issue.

[^ch16-hayek]: F. A. Hayek, "The Use of Knowledge in Society," *American Economic Review* 35, no. 4 (1945): 519–30. The sentence quoted is on 524, in the last paragraph of section IV. The phrase "the particular circumstances of time and place" first appears on 521 and recurs on 522 and 524.

[^ch16-shalizi-planning]: Cosma Shalizi, "In Soviet Union, Optimization Problem Solves You," *Crooked Timber* (blog), May 30, 2012. The objective-function layer is his: planners cannot take the relative weighting of goods as given, because "people have different preferences" and "there is no universally satisfactory mechanism for making such choices." The computational layer is the arithmetic of the same essay.

[^ch16-discovery]: F. A. Hayek, "Competition as a Discovery Procedure," trans. Marcellus S. Snow, *Quarterly Journal of Austrian Economics* 5, no. 3 (2002): 9–23, a translation of a lecture first published in German in 1968. This is the source for the layer that treats the relevant data as generated by rivalrous action rather than merely dispersed; the 1945 paper does not carry that argument on its own.

[^ch16-cybersyn]: Cosma Shalizi, "Project Cybersyn," Notebooks, http://bactra.org/notebooks/cybersyn.html, consulted 29 September 2026; the page is revised continuously and then showed "Last update: 01 Aug 2026" (first version 30 January 2022). The quoted verdict is that "to describe it as a decentralized planning system is nonsense"; the supporting points are that there was "the one (!) mainframe computer available to the project," that the lag between measurement, processing and delivery "was so long that it seems to have been a complete flop," that, in words Shalizi quotes from Eden Medina, "dismantling one of the primary safeguards of [enterprise] autonomy might have been as easy as having someone from the telex room walk down the hall," and that what the project would have produced at best was a "pioneering example of what we'd now call a 'dashboard.'" The telex-room sentence is Medina's, not Shalizi's: Eden Medina, *Cybernetic Revolutionaries: Technology and Politics in Allende's Chile* (MIT Press, 2011), 184, where it reads "of that autonomy" (Shalizi cites the passage as 183–84). For the October 1972 strike and the scale at which the system actually ran, Medina, *Cybernetic Revolutionaries*, chap. 5, "The October Strike," 141–70.

[^ch16-whitehead]: Haridimos Tsoukas and Robert Chia, "On Organizational Becoming: Rethinking Organizational Change," *Organization Science* 13, no. 5 (2002): 567–82, at 570. Tsoukas and Chia are frequently enrolled as Whiteheadians and are not: their process sources are James, Bergson, Mead and Rescher. The Whiteheadian strand in organization studies runs through Robert Cooper and later writers. Chapter 10, *Six Components*, names Whitehead for the position, which is right for the position and wrong for these authors.

[^ch16-cooper]: Robert Cooper, "Organization/Disorganization," *Social Science Information* 25, no. 2 (1986): 299–335. Cooper's complaint against systems theory is that "the very concept which originates system — namely, boundary or division — is the one which is most neglected." We have that sentence from the article's own notes; we did not have access to the full body of the text, and the sharper formulation attributed to him here — that drawing the boundary is the act that manufactures the object — is our compression of the argument rather than a quotation. See also Cooper and Burrell, "Modernism, Postmodernism and Organizational Analysis."

[^ch16-misplaced]: Alfred North Whitehead, *Science and the Modern World* (Macmillan, 1925), where the fallacy of misplaced concreteness is named, and *Process and Reality: An Essay in Cosmology* (Macmillan, 1929), where it is restated. The fallacy is mistaking an abstraction for the concrete fact it abstracts from, which is not an objection to abstracting.

[^ch16-opener]: Werner Ulrich, "A Critique of Pure Cybernetic Reason: The Chilean Experience with Cybernetics," *Journal of Applied Systems Analysis* 8 (1981): 33–59 — twenty-seven pages. The quotation and the taped reply are from the paper and its acknowledgements; the note below on Ulrich sets out the argument in full and what its standing rests on. The five categories are C. West Churchman's, applied by Ulrich.

[^ch16-glossary]: Terms marked ↗ are defined in the Glossary at the back of the book. The mark is set on the first appearance of each term in each chapter, and not afterwards.

[^ch16-beer-techno]: Stafford Beer, *Brain of the Firm*, 2nd ed. (John Wiley & Sons, 1981), 271, where the lapse "seemed likely" to have "aided detractors who later wished to call Cybersyn 'technocratic'"; see also 322 and 328.

[^ch16-beer-naming]: Stafford Beer, *Diagnosing the System for Organizations* (John Wiley & Sons, 1985), 99 and 16; Beer, *Brain of the Firm*, 271. Purpose is "a subjective phenomenon, rather than a property of the system independent of its instigators, participants, and observers": Stafford Beer, *The Heart of Enterprise* (John Wiley & Sons, 1979), 158. On predictive utility, Stafford Beer, *Decision and Control* (John Wiley & Sons, 1966), 113 and 243.

[^ch16-beer-s5]: Stafford Beer, *Beyond Dispute: The Invention of Team Syntegrity* (John Wiley & Sons, 1994), 149; Beer, *Heart of Enterprise*, 263–64.

[^ch16-beer-potential]: Beer, *Brain of the Firm*, 163 and 271.

[^ch16-beer-deliberate]: Beer, *Beyond Dispute*, 149–59; Beer, *Brain of the Firm*, 283.

[^ch16-beer-scale]: Beer, *Brain of the Firm*, 264 and 270; at 263, "nothing was known about the performance" of the forecasting filter in advance.

[^ch16-beer-frills]: Beer, *Brain of the Firm*, 312–13 and 394. The account is declared first-person and partisan at 245–46.

[^ch16-beer-money]: Beer, *Brain of the Firm*, 308; Stafford Beer, *Platform for Change* (John Wiley & Sons, 1975), 159. See also *Decision and Control*, 139.

[^ch16-beer-cut]: Beer, *Decision and Control*, 242, 243 and 246; Beer, *Diagnosing the System*, 2 and 16.

[^ch16-beer-chile]: Beer, *Brain of the Firm*, 322–26.

[^ch16-beer-codes]: Stafford Beer, *Think Before You Think*, ed. David Whittaker (Wavestone Press, 2009), 69; Beer, *Heart of Enterprise*, 260 and, for the variety-matching statement, 235.

[^ch16-beer-cohesion]: Beer, *Heart of Enterprise*, 158.

[^ch16-beer-algedonic]: Beer, *Brain of the Firm*, 275 and 276.

[^ch16-beer-technocrats]: Beer, *Brain of the Firm*, 322.

[^ch16-beer-hayek]: Stafford Beer, *Diagnosing the System for Organizations* (John Wiley & Sons, 1985), 38: "the so-called prerogative to intervene indiscriminately does not have Requisite Variety. It cannot be competently done."

[^ch16-beer-command]: Beer, *Diagnosing the System*, 86.

[^ch16-beer-detach]: Beer, *Diagnosing the System*, 1; Beer, *Heart of Enterprise*, 351.

[^ch16-beer-tree]: Beer, *Heart of Enterprise*, 200. Against this, *Brain of the Firm*, 228, calls the same construction "the five-tier hierarchic model."

[^ch16-beer-recursion]: Beer, *Diagnosing the System*, 2; Beer, *Heart of Enterprise*, 308. Beer's criterion is set out in Chapter 10.

[^ch16-beer-attenuator]: Beer, *Heart of Enterprise*, 308.