From Cybernetics to Chiasm to Category: The Evolution of Organisational Ontology

Abstract#

This paper argues that organisations exceed mathematical capture in a structural sense, and that the project of reducing organisational reality to KPI-level formalisations destroys the intelligence the formalisations claim to measure. The argument does not deny that firms can count. Units shipped, revenue booked, headcount, defects per thousand: these are real magnitudes and the paper denies none of them. The argument separates two incapacities the conventional defence of metrics conflates. The first concerns people, whose relationships and accumulated judgement are constitutive of organisational capacity and are not magnitudes admitting measurement error because they are not magnitudes at all. The second concerns trajectory: even granting every real number a firm books, the measurement frame cannot register what the firm is building or dismantling in the act of booking it, because the would-be calculator is a constituent moment of the process it proposes to total, in the second-order cybernetic sense that Beer’s own architecture could not absorb without dissolving. The metric reports the real component of the firm’s condition and presents the projection as the whole; the discarded component is not a residual magnitude awaiting a finer instrument but a relational and constitutive process that no scalar frame ranges over. A firm can therefore post record efficiency in the same quarter it consumes an engineering substrate twenty years in the building, with the efficiency figure and the consumption standing in no tension within the frame because the frame has no term for the consumption.

The argument proceeds in two registers. In the proof register, two results establish the lower bound. The horizon-mismatch theorem, developed with full force in §24 from Ashby’s law of requisite variety (1956) and elementary information-theoretic constraints on finite measurement frames, demonstrates that any regulator operating at temporal horizon shorter than its substrate cannot capture the substrate’s variety within its measurement window. A corollary holding under the condition that the firm persists, that it goes on meeting its metric rather than coming apart, then locates the unregistered variety: read as the contrapositive of Ashby’s law over the whole coupled system, persistence forces the difference onto the people operating the substrate, onto the substrate itself as accumulating degradation, or onto externalities discharged to coupled systems. The corollary converts the epistemic result into a material one and accounts for the cannibalisation-read-as-efficiency case, where a firm posts rising efficiency in the same quarter it consumes the reserve that produced its capacity. The Novikov-Boone result on the unsolvable word problem in finitely presented groups (Novikov, 1955; Boone, 1958), applied to sufficiently rich organisational categories, demonstrates that algorithmic equivalence-checking over organisational state sequences is undecidable. In the structural-analogue register, the apparatus of sheaf cohomology (Grothendieck, 1957), derived categories (Verdier, 1996), and homotopy type theory (Univalent Foundations Program, 2013) articulates what any adequate formalisation would need to capture: local-global obstruction, transformation remainder, and path-constitutive identity. The articulation is rigorous in the formal sense without claiming constructive embedding of organisations into specific mathematical objects.

The argument is grounded historically through a genealogy from Stafford Beer’s Viable System Model (1972, 1979, 1985) through second-order cybernetics (Bateson, 1972; Maturana and Varela, 1980; von Foerster, 1981; Spencer-Brown, 1969) to the chiasmic ontology developed in Merleau-Ponty’s late work (1968) and extended through Barad’s agential realism (2007) and the enactive cognition of Varela, Thompson, and Rosch (1991). The argument is grounded empirically through three publicly documented cases that instantiate the apparatus across three different institutional contexts and three different evidentiary registers: the Hertz Corporation v. Accenture LLP litigation of 2019, which demonstrates the three-mechanism reform foreclosure operating in commercial digital transformation; the Mid Staffordshire NHS Foundation Trust scandal documented in the Francis Report of 2013, which demonstrates the same mechanisms operating under statutory regulation in public healthcare; and the MacCormack, Baldwin, and Rusnak (2012) test of the mirroring hypothesis in Research Policy, which establishes the chiasmic co-constitution of organisation and product architecture as a quantitative empirical regularity in the software industry.

The consequence is that the widespread reliance on Key Performance Indicators, Objectives and Key Results, and similar metric-based management frameworks is not a suboptimal strategy awaiting refinement. The reliance proceeds from a misapprehension of what organisations are. The misapprehension destroys the intelligence the metrics claim to measure, generates the configuration this paper diagnoses as the N-th Order Epistemic Acknowledgement State, and forces the human absorption of impossibility into bodies, relationships, and biographies. The argument has theoretical, phenomenological, and mathematical grounds, and it has consequences the paper articulates in §34 as a philosophy of cultivation rather than control.

Guide to the Reader#

The argument runs from genealogy through ontology through proof through practice. Part I sets out the Viable System Model and the cybernetic foundations Beer drew on; Part II develops the second-order cybernetic critique that undermines Beer’s observer-stance assumption; Part III articulates the chiasmic ontology that follows from the critique, drawing on Merleau-Ponty, Barad, and the enactive cognition tradition; Part IV demonstrates the chiasmic apparatus in the ordinary moment of a Tuesday-morning bug report at a software company; Part V deploys the mathematical apparatus, with the horizon-mismatch theorem at the foundation and the structural-analogue articulations extending upward; Part VI sets out the strategic adjustments and empirical grounding, including the three documented cases that instantiate the argument across institutional contexts. Section 8, sitting between Parts II and III, develops the Terminal Attractor of Epistemic Denial: the N-th Order State that organisations enter when they maintain frameworks they collectively know to be inadequate, and the three institutional mechanisms by which reform within such configurations is foreclosed.

The paper is long because the argument is cumulative, and its parts require the genealogy, the ontology, the formalism, and the empirical anchoring together. The core thesis can be stated concisely: an organisation cannot calculate the field it is part of co-creating through the act of calculation, and the attempt to do so generates a stable configuration in which the inadequacy is collectively known and institutionally maintained.

Keywords#

organisational ontology, cybernetics, Viable System Model, chiasmic organisation, autopoiesis, second-order cybernetics, horizon-mismatch theorem, three-mechanism reform foreclosure, N-th Order Epistemic Acknowledgement State, phenomenology, Merleau-Ponty, Beer, Bateson, von Foerster, management metrics, KPIs, recursive systems

Part I: The Beautiful Machine That Almost Worked#

§1. Beer’s Viable System Model#

§1.1 The Context#

When Stafford Beer published Brain of the Firm in 1972, organisational theory was dominated by mechanistic models derived from Taylor’s scientific management (1911) and Weber’s bureaucratic theory (1922/1947). These approaches treated organisations as machines to be optimised through division of labour, standardisation, and hierarchical control. Beer’s Viable System Model departed from this tradition by drawing on Wiener’s cybernetics (1948) and Ashby’s law of requisite variety (1956) to reconceptualise organisations as adaptive systems maintaining identity through environmental perturbation. The reconception was architectural rather than terminological: viability, in Beer’s argument, requires specific functional capacities regardless of the system’s particular domain. As Beer put it, “the cybernetic science of effective organization is the science of organizational invariance” (1972, p. 7).

§1.2 The Strategic Purpose of This Analysis#

This paper undertakes a genealogical investigation of organisational ontology, tracing the intellectual evolution from Beer’s cybernetic insights through their critique to a phenomenological alternative. The analysis serves three connected purposes.

The first purpose is historical, since Beer’s VSM, innovative as it was for its time, embedded an assumption about system-environment separation that subsequent theorists would render untenable. The recognition is not a criticism of Beer’s achievement; it is a recognition that theoretical frameworks carry implicit ontological commitments that only become visible through their successors.

The second purpose is constructive, developing the concept of chiasmic organisation through phenomenological investigation that draws on Merleau-Ponty’s late ontology (1945/2012, 1968), Barad’s agential realism (2007), and empirical observation from software development and organisational practice. The chiasmic ontology treats organisations as patterns within fields they continuously co-create, rather than as entities observing antecedent environments.

The third purpose is strategic, deploying mathematical apparatus to establish rigorous boundaries on what formalisation can and cannot do with organisational phenomena. The horizon-mismatch theorem of §24 supplies the foundation; the formal-analogue articulations of §§27-29 extend the argument through the apparatus of sheaf cohomology, derived categories, and homotopy type theory. The approach has precedent in the tradition of impossibility theorems that have proven as valuable as constructive results, exemplified by Gödel’s incompleteness theorems (1931), Arrow’s impossibility theorem (1951), and Turing’s halting result (1936).

The argument carries an urgent practical implication, since the widespread reliance on Key Performance Indicators, Objectives and Key Results, and similar metric-based management frameworks rests on a misapprehension of organisational reality. The misapprehension does not produce neutral inadequacy; it actively destroys the intelligence it seeks to measure, and it generates conditions in which the human cost is absorbed into burnout, organisational fragility, and the systematic erosion of the capacities the metrics purport to cultivate. The paper provides theoretical, phenomenological, and mathematical grounds for the claim, and it sets out in §22 the consequences for organisational practice.

§1.3 The Architecture of the Viable System Model#

The VSM consists of five subsystems, each necessary for viability (Beer, 1979, 1985). System One comprises the primary activities that justify the system’s existence, what Beer called “the collection of primary activities, the muscle and bone of the organization” (1985, p. 14). System Two provides anti-oscillatory mechanisms preventing destructive interference between operational units, the “sympathetic nervous system” providing stability through damping (Beer, 1979, p. 176). System Three handles internal regulation and resource allocation across operations, the “automatic pilot” maintaining homeostasis (Beer, 1972, p. 178). System Four conducts environmental scanning and adaptation: as Beer emphasised, “System Four is there to scan the total environment and to project the total system into it” (1979, p. 226). System Five maintains identity and defines purpose, what Beer called “the closure of the whole system” providing essential unity (1985, p. 128).

Beer’s recursive insight was that each System One unit is itself a complete viable system containing all five functions. The recursive embedding distributes intelligence across levels rather than concentrating it at an organisational apex, and the model permits autonomous viability at every level of the recursion. The architecture broke new ground at the time of its publication, and it remains analytically powerful where contemporary organisational theory continues to treat hierarchical control as the default.

§1.4 The Cybernetic Foundations#

Beer grounded the VSM in cybernetic principles, particularly in Ashby’s law of requisite variety (1956, p. 207): “only variety can destroy variety.” Beer’s organisational interpretation read the law as a regulatory constraint: “the variety of the control system must be equal to or greater than the variety of the system being controlled” (1985, p. 89). Variety engineering, in Beer’s apparatus, proceeds through two complementary mechanisms. Attenuation filters and aggregates incoming environmental variety, reducing what the regulator must process to what its measurement frame can resolve. Amplification multiplies regulatory variety through parallel processing, hierarchical decomposition, and adaptive subroutines, raising what the regulator can absorb to match the environmental complexity it must handle. Beer (1979, p. 88) argued that organisations fail when “the variety of the environment exceeds the variety of the system designed to regulate it,” and the VSM architecture distributes variety-engineering operations across the five subsystems so each handles the variety appropriate to its register.

Beer’s use of Ashby’s law anticipates the formal result this paper develops in §24. There the law is sharpened into the horizon-mismatch theorem, which states the sense in which a regulator operating at temporal horizon shorter than its substrate cannot, even in principle, satisfy the variety condition Ashby’s law imposes. The cascade argument of Part V then builds on the theorem to extend the variety constraint through more elaborate mathematical apparatus.

§1.5 The Departures from Traditional Organisational Theory#

The VSM departs from traditional organisational theory along six axes. Each departure carries weight on its own terms; together they constitute a coherent reframing of what organisational structure is and how it operates.

The first departure is from hierarchical trees to recursive embedding. Traditional organisational theory, rooted in Weber (1922/1947) and Taylor (1911), conceived organisational structure as a hierarchical tree with clear superior-subordinate relationships. Beer’s recursive embedding gave each operational unit all five subsystems internally, with intelligence distributed across levels rather than concentrated at an apex, and every recursive unit possessing the capacity for autonomous viability within its scope.

The second departure runs from concentrated to distributed intelligence: traditional models concentrated decision-making and environmental scanning at senior levels, following Fayol’s (1916/1949) principle of centralisation. Beer’s System Four function appears at every recursive level, permitting local environmental adaptation without central mediation. The recursive presence of System Four enacts the proposition Beer formulated as “every viable system contains and is contained in a viable system” (1979, p. 234).

The third departure is from fixed divisions to dynamic boundaries. Where traditional approaches imposed fixed departmental divisions with clear demarcations, the VSM treats boundaries as permeable membranes managing variety flows, so that boundaries become zones of variety management and the architecture admits dynamic reconfiguration in response to environmental demand.

The fourth departure is from top-down commands to autonomous coordination. The VSM replaces command hierarchies with autonomous units that achieve coordination through System Two’s anti-oscillatory function. Beer described this as “maximum autonomy consistent with cohesion” (1985, p. 67), and the formulation encodes the entire coordination philosophy: autonomy and cohesion, which conventional management treats as a trade-off, are instead jointly maintained through System Two’s damping function.

The fifth departure is from data flows to variety management. Traditional approaches understood information as data flowing up hierarchies for decision and commands flowing down for execution. Beer reconceptualised information processing as the cybernetic operations of attenuation and amplification distributed throughout the system, so that information itself becomes variety to be managed.

The sixth departure is from central planning to local adaptation. Where traditional management assumed centralised planning and execution, the VSM admits local adaptation: each recursive level responds directly to its environment, with overall coherence maintained through Systems Three and Five.

§1.6 Early Influence and Practical Adoption#

The VSM’s influence extended across multiple domains, the most ambitious application occurring in Chile, where Beer attempted to coordinate the national economy through Project Cybersyn between 1971 and 1973. The project, documented by Medina (2011), created a real-time economic management system using telex networks and operations rooms. It ended when the September 1973 military coup overthrew Allende and destroyed the project’s institutional context.

In healthcare, the UK National Health Service attempted VSM implementation during the 1980s reforms, seeking to balance local hospital autonomy with system-wide coordination (Beckford, 1995). The NHS did not achieve full implementation, though VSM principles influenced subsequent healthcare management approaches. In industrial settings, manufacturing sectors, particularly steel and chemical industries, adopted VSM principles for production planning and control. Espejo and Harnden (1989) document implementations at British Steel Corporation and ICI, where the model coordinated complex production processes.

Management consultancies later incorporated VSM principles into their methodologies, often without acknowledgement of the cybernetic origins or the model’s design limitations (Jackson, 2003, p. 89). The adoption is worth marking explicitly given this paper’s later analysis: the same consultancies whose methodologies are diagnosed in §8.6 and §33.1 as displacement-as-defence drew, often invisibly, on Beer’s apparatus when constructing their proprietary frameworks. The VSM’s spread into management consulting demonstrated both the model’s appeal and the conditions under which advanced cybernetic insight could be flattened into the very practices it had been designed to displace.

Academic development followed a steadier path, with the VSM generating extensive scholarly research and dedicated conferences and journals exploring its theoretical extensions and practical applications, and Espejo and Harnden (1989, p. 3) characterised Beer’s work as “the most comprehensive attempt to apply cybernetic thinking to organizational design.” The influence persists in contemporary approaches to enterprise architecture and adaptive management, though as Jackson notes the cybernetic origins are often unacknowledged in the contemporary literature.

Adoption demonstrated both the model’s appeal and its practical difficulties. The implementation problems anticipated the second-order cybernetic critique that follows in Part II.

§2. The Limitations Embedded in the VSM#

§2.1 The Observer Problem#

The VSM presupposes that organisations can observe their environments objectively, since System Four’s function rests on exactly that presupposition. Von Foerster (1973/1981, p. 258) would later render the presupposition untenable: “the environment as we perceive it is our invention.” Beer’s own formulation makes the dependence explicit: “System Four observes the total environment and models it” (1985, p. 56). The formulation requires three commitments: that the environment exists independently of observation, that observation does not alter the observed, and that models can faithfully represent external reality. Each commitment will be challenged by a different strand of second-order cybernetics: Maturana and Varela challenge the first through autopoiesis, von Foerster challenges the second through the constructivist account of perception, and Bateson challenges the third through the map-territory analysis.

§2.2 The Boundary Paradox#

The VSM requires clear boundaries between system and environment. Beer himself poses the question without resolving it: “where does the organization stop and the environment begin?” (1979, p. 405). The model requires boundaries to fix what is inside (controllable) versus outside (observable), to determine where System Four stops modelling itself and begins modelling the environment, and to specify how System One units interface with external entities. As Luhmann would later argue, “the boundary is not a natural given but a construction of the system itself” (1995, p. 17). Beer’s framework cannot accommodate the constructive character of organisational boundaries because doing so would undermine the observer-stance the VSM presupposes.

§2.3 The Sequential Assumption#

The VSM implies a temporal asymmetry at each recursive level, in which observation precedes modelling, modelling precedes decision, and decision precedes action. Beer’s summary “information flows up, control flows down” (1972, p. 160) captures the asymmetry. The asymmetry is internal to each level; the recursion across levels does not eliminate it. Each level requires the temporal sequence observation → model → decision → action. Weick’s empirical work on organisational sensemaking shows that the assumption is reversed at the very moment organisational actors are doing the work the framework presupposes. “People often don’t know what they think until they see what they say” (Weick, 1979, p. 165). Sensemaking is retrospective. The temporal asymmetry the VSM assumes does not survive close examination of organisational practice.

§2.4 The Centralisation Tendency#

Despite the VSM’s commitment to distributed intelligence, System Four in practice tends towards centralisation. Beer’s requirement that System Four “model the total system and total environment” (1985, p. 226) imposes a variety condition that no single regulator can absorb (per Ashby’s law and the horizon-mismatch result of §24). Organisations consequently centralise System Four by default, because no single locus possesses the variety the model demands. Empirical studies of VSM implementation (Beckford, 1995; Jackson, 2003) document the same pattern: implementing teams strengthen System Four at higher recursive levels, weaken it at operational levels, and create de facto centralised intelligence functions that contradict Beer’s distributed-intelligence commitment.

The next sections trace how second-order cybernetics rendered each of the VSM’s observer assumptions untenable. Bateson, Maturana and Varela, von Foerster, and Spencer-Brown each attack a different facet of the observer stance, and the convergence of the four critiques produces the conditions under which the chiasmic ontology of Part III becomes available.

Part II: The Critique Through Second-Order Cybernetics#

§3. Bateson’s Recursive Epistemology#

§3.1 Information as Difference That Makes a Difference#

Bateson’s epistemology of information and observation arrives in the essay “Form, Substance, and Difference,” collected in Steps to an Ecology of Mind (1972, p. 453): “information, the elementary unit of information, is a difference which makes a difference.” Information, in Bateson’s articulation, is relational, selective, and observer-dependent. It requires a difference, must make a difference, and exists only relative to whom it makes a difference to. The contrast with classical information theory is direct: information is not data, bits in channels, or signals to be transmitted; it is the difference-event itself. Bateson elaborates the position later in the same essay: “the territory never gets in at all. What gets in is news of difference, and news of difference is transformed (i.e., coded version) of events which preceded it” (1972, p. 459).

§3.2 The Map-Territory Relation#

Bateson extends Korzybski’s (1933) formulation “the map is not the territory” through recursive elaboration: “the map is not the territory, and the name is not the thing named, but the map is all we have access to” (1972, p. 455). Environmental models are maps with attendant distortions, simplifications, and abstractions, and organisations respond to these maps rather than to the territory itself, so that different observers construct different maps and consequently encounter different organisational realities. Bateson extends the analysis to push the regress further: “the name is not the thing named, and the thing named is not the ‘thing’ itself” (1979, p. 32). The regress is unterminating, since every representation is itself a representation of a representation, leaving System Four no ground from which to start.

The map-territory relation has a geometric form that renders the distortion exact rather than cautionary, and that exactness matters for the argument Part V will make formally. A globe cannot be represented on a flat sheet without distortion. This is not a limitation of cartographic technique that a better projection would remove; it is a theorem. Gauss’s Theorema Egregium (Gauss, 1827) establishes that the curvature of a surface is intrinsic, preserved under any bending that does not stretch, so a surface with curvature cannot be laid flat onto a plane, which has none, without tearing or stretching somewhere. Every flat map of the round earth is therefore a projection, and every projection makes a choice about which property to preserve and which to sacrifice. The Mercator projection preserves angles, which is why it served navigation, and pays for the preservation by inflating area without bound toward the poles. An equal-area projection preserves area and pays by shearing every shape. No projection preserves angle, area, distance, and direction together, because the plane lacks a property the globe has, and the missing property has to surface somewhere as error. The distortion is not removed by a better map; it is only relocated by a different one.

This geometry gives the organisational dashboard its exact diagnosis, and three distinct breakdowns follow from the single fact that the flat map sacrifices what it cannot carry. Each breakdown disables a different organisational function, and each is invisible to the reader of the map, because a projection is internally consistent and gives no sign from within that anything has been lost.

The first breakdown disables coordination, for on the globe the shortest path between two points is an arc of a great circle. Drawn on the flat Mercator map, that shortest path appears as a curve bending far away from the straight line a ruler would draw between the same two points, which is why the shortest flight from northern Europe to the far east arcs over the pole and looks, on the flat map, like a detour. The person reading the flat map sees the ruler-straight line, takes it for the shorter one, and asks why anyone would take the bending route. The person reading the globe knows the bending route is the shortest one and the ruler-straight line is longer. Both are looking at the same journey; each calls the other’s path irrational; neither can see that they are reading off different projections, because the projection is the medium of seeing and not an object within it. This is the structure of the recurring disagreement between organisational functions. One team sees the path through the actual system, which on the flat map of the metric looks like a wilful detour, and another team sees the flat-map straight line and asks why the first will not simply take the direct route. Each accuses the other of taking the long way, and the dashboard is the flat map on which the real shortest path through the territory is rendered as obstruction.

The second breakdown disables estimation, since the Mercator inflation of area toward the poles, far from being a small effect at the margins, grows enormous and counterintuitive at the scale of whole continents. Greenland is rendered at roughly the visual size of Africa and is in reality about fourteen times smaller. Russia is rendered at roughly twice its true area relative to equatorial land. A country near the equator that reads as small on the map can exceed in true area a high-latitude country that reads as large, so that two regions appearing equal on the flat map can differ several-fold in the ground they actually cover. The organisational consequence is that the apparent size of a task on the dashboard is a function of where the task sits in the projection rather than of the work the task contains. The visible, measured, high-latitude work is inflated and draws resources proportionate to its apparent size; the quiet infrastructure maintenance and the relational labour sit near the equator of the metric, read as small, and are resourced as though they were small, when in true area they are the larger continent. Two initiatives that read as equal commitments on the dashboard can differ by a factor of several in the labour they actually demand. Management sizes the task by its projected area and is wrong in a direction the map itself conceals, over-committing to the inflated and under-committing to the compressed, with no warning from the instrument, because the instrument is internally consistent and reports apparent area as though it were real area.

The third breakdown is the one the first two are instances of, and it is the one Bateson named. The map is not a lie: a navigator crosses oceans with the Mercator chart and arrives, and a planner allocates with the dashboard and ships product. The flat map and the metric both do real work on the axis they preserve, and the error lies not in using the projection but in forgetting that it is one, in reading the inflated continent as vast in itself and the bending geodesic as perverse in itself, in mistaking the chosen preservation for the whole of the territory. The error and its correction are both geometric, so that one does not respond to the distortion of the Mercator map by abolishing maps; one responds by knowing which projection is in hand and what it was built to sacrifice, and by reaching for a different projection when the task changes from navigation to the measurement of land. The dashboard admits the same discipline and forbids the same forgetting.

The geometry also states, in advance and in intuitive form, the conservation result Part V will reach formally. The distortion that the flat map cannot avoid is not destroyed by any choice of projection; it is only moved. Mercator moves it onto area to keep angle; the equal-area projection moves it onto shape to keep area. The cartographer does not decide whether to distort but only what to distort, which is to say who bears the distortion. The variety the map cannot hold does not vanish for want of a projection to hold it; it relocates to whatever the projection sacrifices. In the organisation, as the later argument develops, the variety the metric cannot hold relocates in the same way, and the term it relocates onto is most often the people who absorb the difference between the flat map and the round world. The map analogy and the cohomological obstruction of Part V are the same structure stated at two registers: the impossibility of flattening a curved surface without remainder, and the impossibility of patching locally consistent sections into a globally consistent one without obstruction. The first is the intuitive face of the second.

§3.3 Double Binds in Organisational Communication#

Bateson’s theory of double binds was developed in the clinical context of schizophrenia in families (Bateson, Jackson, Haley, and Weakland, 1956; Bateson, 1972, pp. 206-227). The structure transposes to organisational communication when the same conditions hold: a primary injunction, a secondary injunction contradicting the first, an inability to escape the field, and a prohibition on commenting on the contradiction. Organisations frequently produce exactly this structure: a primary injunction directs the agent to action A, while a secondary injunction, often emerging through a different channel or organisational subsystem, directs the agent to incompatible action B; the agent cannot leave the field, since resignation is materially or socially foreclosed; and explicit articulation of the contradiction is institutionally prohibited.

Bateson offers concrete organisational forms (1972, p. 208): “Be spontaneous” combines a primary injunction (act spontaneously) with the secondary feature that following the injunction is by definition not spontaneous. “Ignore these instructions” is the same form in compressed shape. “Be independent” is the form mapped to organisational autonomy: the directive to behave independently, issued by a hierarchical authority, undermines the independence it commands.

System Two (coordination) operates within a single logical level, whereas double binds straddle two: the primary injunction and the meta-injunction prohibiting comment on the contradiction. System Two cannot reach the meta-level, by its own functional definition. As Bateson observed, “the victim cannot make the metacommunicative statement, ‘this is a paradox’” (1972, p. 212).

§3.4 Levels of Learning and Logical Types#

Bateson identified hierarchical learning levels grounded in Russell and Whitehead’s theory of logical types (Principia Mathematica, 1910-1913; Bateson, 1972, pp. 279-308). Learning Zero is mechanical response without learning; Learning One acquires specific responses to specific contexts; Learning Two reframes contexts and adapts the patterns of Learning One; Learning Three reframes the patterns of Learning Two themselves; and Learning Four operates at the level of evolutionary change in entire ways of categorising experience.

The VSM was designed for Learning Two, where variety engineering modifies the regulatory parameters that Systems One through Four operate within, and System Five modifies the framework that defines System Four’s environment. What the VSM cannot reach is Learning Three: questioning the framework itself. Bateson observes that “Learning III is likely to be difficult and rare even in human beings” (1972, p. 303), and the difficulty is built into the VSM at the level of its architecture. Learning Three would require questioning the observer-observed separation that the VSM presupposes. The VSM cannot accommodate the questioning without abandoning its own architecture.

§3.5 The Recursive Nature of Epistemology#

Bateson developed the recursive epistemology in the late work: “the observer must always be included in the observed” (1979, p. 201). He named the resulting meta-pattern “the pattern which connects” (1979, p. 8). Two consequences follow. First, when System Four observes its environment, this is environment observing itself through the recursive structure that contains the observer. Second, boundaries are not given prior to observation; observation creates them. The chiasmic-organisation argument of Part III turns on exactly these two consequences. Bateson’s closing formulation captures the position: “the mental world, the mind, the world of information processing, is not limited by the skin” (1979, p. 242).

§4. Maturana and Varela: Autopoiesis and Organisational Closure#

§4.1 The Santiago Theory of Cognition#

Maturana and Varela, working in Santiago during the period of Beer’s Cybersyn project, developed autopoiesis theory. The theory challenges the VSM’s presupposition that observation imports information from environment into system. Maturana and Varela’s formal definition appears in Autopoiesis and Cognition (1980, p. 78): “an autopoietic machine is a machine organized (defined as a unity) as a network of processes of production (transformation and destruction) of components that produces the components which: (i) through their interactions and transformations continuously regenerate and realize the network of processes (relations) that produced them; and (ii) constitute it (the machine) as a concrete unity in the space in which they (the components) exist by specifying the topological domain of its realization as such a network.”

§4.2 Organisational Closure and Structural Determinism#

Maturana and Varela’s theory was developed for biological systems, with the nervous system as the paradigm case. As they state, “the nervous system is organizationally closed, and therefore, no inputs or outputs occur in its operation” (1987, p. 169). The transposition to organisations requires justification, which Luhmann (1995) supplied by extending autopoiesis to social systems. Luhmann argues that organisations exhibit operational closure in the analogous sense: they generate their own categories of legitimate action through internal operations. The formal-analogue argument (per §25.3) is that organisations exhibit the autopoietic features Maturana and Varela identify in the nervous system, with Luhmann’s framework providing the bridge from biological to social application.

Three consequences follow. Organisations do not receive inputs in the cybernetic sense; they are perturbed. Perturbations trigger internal changes, but the changes are determined by internal structure, not by the perturbations themselves. Maturana captures the relation: “the structure of the environment cannot specify the changes of state of the system; it can only trigger them” (1978, p. 42).

For Beer’s System Four, the consequence is direct: System Four cannot scan the environment as if it were an external object, cannot import information across an inside-outside boundary, and can only undergo internal changes triggered by perturbations, where the triggers do not determine the changes but the internal structure does.

§4.3 Structural Coupling#

Structural coupling differs from adaptation: where adaptation implies an organism conforming to a fixed environment, structural coupling describes mutual specification. Organisation and environment specify each other through the history of their interactions, and the changes that result are congruent with environmental conditions without being caused by them. Maturana and Varela (1987, p. 75) formulate the position: “the continued interactions of a structurally plastic system with its environment will result in selections of the system’s structure such that it will ‘fit’ into its environment.” The fit is mutual specification, not conformity to a pre-existing shape.

Varela (1979, p. 48) sharpens the point: “in describing autopoiesis, one is not describing the properties of components, but relations which must be regenerated constantly.” The shift from properties to relations is the move that distinguishes autopoiesis from earlier systems-theoretic approaches.

§4.4 Cognition as Effective Action#

Maturana and Varela characterise cognition: “living is a process of cognition” (1987, p. 29). Cognition is effective action in the domain of existence rather than information processing in an internal representation. Organisations cognise by maintaining their organisation, not by computing models of an external world. As Varela formulated the position, “the concrete realization of the living system determines what can be a perturbation and what cognitive domain it can have” (1979, p. 268). Beer’s System Four assumes cognition as representation, whereas the autopoietic account treats cognition as effective action, the bringing forth of a world through embodied engagement (the language of “bringing forth” developed further in Varela, Thompson, and Rosch, 1991).

§4.5 The Observer as a Living System#

Maturana formulates the position: “everything said is said by an observer” (1978, p. 31). The observer is itself a living system with its own autopoiesis. Observation occurs from within an autopoietic organisation, not from a position external to it, and the consequences follow without further premise: observations are structural determinations of the observer rather than reports of the observed, so that different observers, constituted differently, encounter different realities, and no privileged observation point is available, because every observation is the operation of some autopoietic system. Maturana and Varela’s closing formulation captures the position: “we do not see what we do not see” (1987, p. 242).

§5. Von Foerster: Second-Order Cybernetics#

§5.1 The Cybernetics of Observing Systems#

Heinz von Foerster, who edited the proceedings of the later Macy Conferences (conferences 6-10, 1949-1953), later coined the term “second-order cybernetics” in lectures during the late 1960s and formalised the position in his Biological Computer Laboratory work at Illinois. The new register identifies a constraint internal to the cybernetic programme: cybernetics studying observed systems must itself be studied as a cybernetics of observing systems. Von Foerster’s distinction (1981, p. 258) is direct: “first-order cybernetics is the cybernetics of observed systems. Second-order cybernetics is the cybernetics of observing systems.” The shift relocates the observer inside the system being studied, and observation becomes participation rather than spectatorship. Von Foerster’s formulation: “objectivity is the delusion that observations could be made without an observer” (2003, p. 285).

§5.2 The Construction of Reality#

Von Foerster’s central claim (1981, p. 275) is that “the environment as we perceive it is our invention.” The position avoids solipsism. The claim is not that nothing exists outside perception; the claim is that perception constructs rather than receives, that reality is enacted in observation, and that organisations create the environments they then perceive. Von Foerster’s 1973 paper “On Constructing a Reality” (reprinted in Observing Systems) demonstrates the neurophysiological case: the nervous system is operationally closed, no information enters from outside, and all perception is internal construction.

For organisations, the consequence is that System Four constructs environmental opportunities through its distinctions rather than discovering them as antecedent objects. The categories of risk, opportunity, threat, and adaptation are not given by the environment; they are imposed on the environment by the organisation’s observing apparatus.

§5.3 Eigenforms and Organisational Stability#

Von Foerster’s 1976 paper “Objects: Tokens for (Eigen-)Behaviors” (reprinted in 2003, p. 261) introduces eigenforms, stable forms arising from recursive operations: “objects are tokens for eigenbehaviors.” Stable patterns emerge from recursive interactions, the patterns appear as objects or features of an environment, and yet they are eigenbehaviours: the stable products of recursive operation rather than external givens. Von Foerster’s mathematical formulation (2003, p. 264): “the eigenvalues are the stabilities that occur when an operation is recursively applied to itself.”

Beer’s System Five (identity) operates as an eigenform: a stable pattern emerging from and maintained by recursive organisational operations, rather than a fixed property of the organisation. The recognition matters for the impossibility argument of §7: an organisation’s identity is something the organisation does, recursively, and reform requires modifying the recursion itself rather than substituting one fixed identity for another.

§5.4 Ethics and Responsibility#

Von Foerster (1992, p. 14) formulates two imperatives. The ethical imperative: “act always so as to increase the number of choices.” The aesthetic imperative: “if you desire to see, learn how to act.” The imperatives invert VSM variety management. Beer’s variety engineering reduces variety to manageable levels; von Foerster’s ethical imperative increases the number of choices available. Beer’s architecture proceeds from observation to action; von Foerster’s aesthetic imperative makes action prior to observation, with seeing made possible by acting. Von Foerster’s closing formulation captures the recursive structure: “I am the observed relation between myself and observing myself” (2003, p. 289).

§5.5 The End of the Observer Illusion#

Von Foerster drives the position to its conclusion: “there are no observers; there are only actors who construct their universe through their actions” (2003, p. 287). The consequences for the VSM are direct, since the observer stance the architecture requires does not exist as a feature of any actual organisation. What appears as observation is action, what appears as scanning is construction, and what appears as objective modelling is the recursive operation of an observing system constructing the world it then claims to describe.

§6. Spencer-Brown: The Laws of Form#

§6.1 The Calculus of Indications#

Spencer-Brown’s Laws of Form (1969) develops a calculus of distinctions. The calculus treats organisational boundaries as acts of distinction made by an observer rather than as divisions among pre-existing entities. The opening axiom (1969, p. 1) sets out the structure: “we take as given the idea of distinction and the idea of indication, and that we cannot make an indication without drawing a distinction.” The calculus rests on two laws:

• Law of Calling: the value of a call made again is the value of the call.
• Law of Crossing: the value of a crossing made again is not the value of the crossing.

The laws encode the formal properties of distinctions: distinctions can be referenced repeatedly without altering the distinction itself, but crossings (acts of moving across the distinction) cancel themselves when repeated.

§6.2 Boundaries as Distinctions#

Spencer-Brown’s foundational formulation (1969, p. 1) states that “distinction is perfect continence.” The act of distinction creates what it distinguishes; boundaries do not separate pre-existing entities but constitute the entities they distinguish, with every boundary an observer’s distinction. Organisational boundaries follow the same pattern: department boundaries do not separate antecedent departments, since the boundary-making constitutes the departments. Customer-versus-employee distinctions, product-versus-service categories, internal-versus-external divisions: each is the calling-into-being of what the distinction marks. Spencer-Brown’s later remark (1969, p. 69) gives the cosmological extension: “the universe is the inside of the severance of the first distinction.”

§6.3 Re-Entry and Organisational Recursion#

Spencer-Brown’s re-entry construction (1969, Chapter 11) resolves a problem Beer’s VSM cannot explain: organisational self-reference without infinite regress. When a distinction re-enters its own indicational space, the system can refer to itself without generating an infinite tower of meta-levels. The construction introduces imaginary values to resolve the resulting oscillation. Organisations exhibit re-entry when they observe themselves observing, when System Four models itself modelling, when boundaries include their own drawing. The phenomenon is recognisable in any organisational moment where the activity of describing the organisation becomes itself part of what is described, and the system Beer attempted to formalise as observation from outside dissolves into recursive self-reference from within.

§7. The Accumulated Critique#

Five impossibilities follow from second-order cybernetics. Objective observation is impossible, per von Foerster’s recognition that the observer is part of the system. Information transfer is impossible in the cybernetic sense, per Bateson: only difference-events occur, and they are observer-coded. Environmental inputs are impossible, per Maturana and Varela: organisations are organisationally closed. Pre-existing boundaries are impossible, per Spencer-Brown: boundaries are distinctions made by observers. External control is impossible, as the consequence of the first four. The observer stance the VSM requires cannot exist. Organisations are patterns within the field they themselves participate in producing, the chiasmic structure Part III develops.

Before turning to that structure, the paper takes up a question the second-order critique sharpens but does not answer. Organisations continue to operate as if the observer-stance were available, continue to deploy frameworks the operators collectively know to be inadequate, and continue to invest substantial resources in maintaining the gap between official framework and actual practice. The configuration is stable, well-documented across institutional contexts, and resistant to reform from within. The next section develops the formal structure of this state and the institutional mechanisms by which reform within it is foreclosed.

§8. The Terminal Attractor of Epistemic Denial#

The configuration this section develops sits between the second-order critique of Part II and the chiasmic ontology of Part III. The section is positioned here because this state is what becomes visible when an organisation maintains a first-order framework while operating in conditions the second-order critique has rendered untenable. The state is stable, and internal reform within it is foreclosed. The mechanisms of foreclosure are institutional rather than psychological, and they operate invariantly across widely different organisational contexts.

§8.1 The N-th Order Epistemic Acknowledgement State#

The N-th Order Epistemic Acknowledgement State is the configuration in which an organisation $\mathcal{O}$ maintains an operative framework $F$ while its constituent agents hold tacit knowledge $K$ that the framework cannot capture. The state is characterised by four features.

Feature 1 (Representational Inadequacy). The tacit knowledge $K$ is not contained in the operative framework $F$:

The exclusion is built into how the framework operates, since the framework works in one register (formal procedures, metrics, audit-compatible language) while the tacit knowledge belongs to another (lived experience, pattern recognition, embodied skill). The two registers are not different by degree; they are incommensurable, and the framework cannot accommodate the tacit register without ceasing to function as a framework.

Feature 2 (Universal Recognition). Every participant recognises that $K \not\subset F$:

The recognition is distributed across the organisational field rather than localised to a dissident minority or held privately by senior leadership; it is common knowledge in the precise sense the next feature formalises.

Feature 3 (Recursive Acknowledgement to Depth $N$). Each participant knows that every other participant recognises the inadequacy, and knows that they know, to arbitrary recursive depth:

The recursion can extend to arbitrary $N$. The structure is the game-theoretic condition of common knowledge (Lewis, 1969; Aumann, 1976) applied to framework inadequacy: each participant knows, knows that the others know, knows that the others know that they know, and so on.

Feature 4 (Mandatory Framework Persistence). Despite Features 1 through 3, the framework $F$ persists as the sole legitimate language of organisational discourse. Alternative articulations using $K$-language lack institutional standing. The persistence is not maintained by any agent’s preference. It is maintained by the architecture: by audit requirements, regulatory compliance, evaluation procedures, and the credentialing systems that determine which utterances count as organisationally legitimate.

An organisation satisfying Features 1 through 4 simultaneously occupies the N-th Order Epistemic Acknowledgement State. This state is no exception or pathological case but a structurally stable attractor in the institutional state-space, and the historical record shows that organisations enter it across a wide range of contexts and remain in it for extended periods.

§8.2 Homeostatic Mechanisms#

The N-th Order State maintains itself through three institutional mechanisms. Each mechanism is observable in organisational practice, documented in the empirical literature, and reducible to a specific function within this state.

§8.2.1 Credentialism as Variety Attenuation#

Information processing in the N-th Order State filters incoming variety through credential conditions before content. Let $C: \text{Sources} \to [0,1]$ map information sources to credential scores. Let $V$ represent information value. In healthy systems, information processing follows $\text{Uptake} = f(V)$: uptake scales with value. In the N-th Order State, processing becomes $\text{Uptake} = C(\text{source}) \cdot V$, with $C$ operating as a gating function. When $C(\text{source}) \approx 0$ for sources articulating $K$ in non-$F$-compatible terms, even high-value information receives effectively zero organisational uptake. The mechanism is documented in Collins and Evans (2007) on expertise and in Alvesson and Robertson (2006) on elite identities in consulting firms.

The credentialism serves a dual function: explicitly, it maintains quality standards for organisational knowledge; implicitly, it filters information through compatibility with $F$, ensuring that only $F$-compatible knowledge gains organisational legitimacy. Information contradicting $F$, however valid in substantive terms, arrives from sources lacking the requisite credentials and consequently attenuates to insignificance in organisational decision-making (Abbott, 1988). A developer with fifteen years of system knowledge is overridden by a management consultant from a major firm who lacks the domain knowledge but possesses the institutional standing the framework recognises.

§8.2.2 Performance Amplification as Divergence Response#

As the gap between official framework and lived reality widens, the organisation intensifies its commitment to the framework rather than questioning it. Metrics proliferate. Ceremonies elaborate. The worse $F$ performs, the more organisational labour is devoted to demonstrating compliance with $F$. The dynamic is one of positive feedback: divergence between $K$ and $F$ generates additional investment in $F$-maintenance, which generates additional divergence as more of the actual work is forced into $F$-incompatible shadow practices.

Let $D = \int_{\mathcal{O}} |K(a) - F(a)|, da$ represent aggregate divergence between tacit and formal across the organisational field. Let $E_F$ represent organisational labour invested in $F$-compliant performance. In healthy systems, recognition of inadequacy reduces investment: $\partial E_F / \partial D < 0$. In the N-th Order State, the relation inverts and accelerates: $\partial E_F / \partial D > 0$ and $\partial^2 E_F / \partial D^2 > 0$. The mechanism is documented across empirical literatures: Power (1997) shows that audit inadequacies generate more audit rather than less; Strathern (2000) shows that metric inadequacies generate more metrics; Alvesson and Spicer (2016) describe the broader pattern of functional stupidity in which framework intensification substitutes for framework questioning.

Feynman’s (1974) characterisation of cargo cult science captures the dynamic in compressed form: if the planes are not landing, build longer runways and wave the signal flags more vigorously. The intensification operates with full ritual seriousness within the framework while the framework’s relation to the substrate it claims to regulate continues to deteriorate.

§8.2.3 Source Isolation as System Defence#

Individuals who persistently articulate the gap between $K$ and $F$ are marginalised through subtle exclusion rather than explicit punishment. The marginalisation operates through three characteristic formulations: “they are not team players,” “they are too negative,” “they do not understand the big picture.” The problem, organisationally, is not the analytical content of what they articulate but the institutional pressure their articulation places on the stability of the state. Morrison and Milliken (2000) document the pattern as organisational silence; Ryan and Oestreich (1991) describe the broader phenomenon of fear-driven communication suppression. The isolation operates as an immune response: the state treats articulation of $K$ as a threat to homeostasis and deploys institutional mechanisms to neutralise the threat without acknowledging the analytical content of what is being articulated.

§8.3 The Three-Mechanism Reform Foreclosure#

The N-th Order State resists reform from within by its very architecture, the resistance following from the three mechanisms operating in coordinated foreclosure: any reform proposal expressed in organisational discourse is processed by one of three modalities, each of which preserves $F$ and the architectural feature targeted by the reform.

Theorem (Three-Mechanism Reform Foreclosure). Let $\mathcal{O}$ be an organisation in the N-th Order State with operative framework $F$, tacit knowledge $K$, and $K \not\subset F$. Let $R$ be any reform proposal articulated within organisational discourse. Then $R$ is processed by one of three mechanisms.

Mechanism 1 (Absorption). If $R$ is articulated in $F$-language, then $R$ is compatible with the framework’s existing categories. The reform’s content can be expressed without challenging the framework’s foundations. The framework absorbs $R$: it implements the reform’s surface content while leaving the architectural condition $K \not\subset F$ unchanged. Example: a reform demand for better velocity tracking is absorbed into the existing velocity-as-meaningful-metric paradigm. The reform succeeds locally and the architectural condition is preserved.

Mechanism 2 (Attenuation). If $R$ is articulated in $K$-language, in terms the framework does not recognise as legitimate, then $R$ lacks the institutional standing required for uptake. The reform is filtered through credentialism (Mechanism 2 of §8.2.1): only $F$-compatible justifications gain regulatory or evaluative standing. The reform is attenuated regardless of substantive content. Example: a reform demand that velocity metrics create perverse incentives is dismissed as “not understanding how we work” because the dismissal-grounds operate within $F$ and the demand operates outside it.

Mechanism 3 (Decomposition). If $R$ attempts a hybrid articulation, partly in $F$-language and partly in $K$-language, then the mechanisms decompose $R$ into its constituent registers. The $F$-compatible elements are processed by Mechanism 1 (absorbed); the $K$-elements are processed by Mechanism 2 (attenuated). The hybrid does not survive the decomposition: what passes through is only the $F$-compatible content that does not challenge the framework. Example: a reform proposal that “we should improve velocity tracking ($F$-compatible) while addressing the metric-gaming behaviour it incentivises ($K$-element)” produces an organisational response that implements “improved velocity tracking” and ignores the gaming critique.

Corollary. For any reform $R$ expressed in organisational discourse, the architectural feature targeted by $R$, the condition $K \not\subset F$, is preserved across the institutional processing of $R$. The N-th Order State is therefore stable against internal reform. Escape requires either external shock that bypasses the institutional processing (regulatory intervention, financial crisis, public disclosure) or ontological reframing that dissolves the $K/F$ distinction itself.

The theorem is proved in the proof register by exhaustion of the three discourse possibilities ($F$-language, $K$-language, hybrid) combined with the empirical observation that each possibility activates one of the three mechanisms. The empirical grounding for the mechanisms (credentialism as variety-attenuation, performance amplification as divergence response, source isolation as system defence) was established in §8.2 and is historically validated across the cases of §8.4.

The Sartrean reading of this state as institutional bad faith is available but requires care. Sartre’s analysis (1943/1956, pp. 86-116) treats bad faith as individual: the café waiter performing the waiter role with exaggerated precision while knowing himself to be more than this role exemplifies simultaneous knowing and not-knowing held in place by refusing to examine the contradiction. The N-th Order State manifests something structurally analogous at institutional scale. The distinction with Sartre is important and worth marking: the institutional bad faith is not the aggregation of individuals’ Sartrean bad faith. It persists even when no individual is in Sartrean bad faith. It is a property of the architecture’s discourse-rules, not of any agent’s self-deception. Fleming and Spicer (2003) analyse the cynical-distance variant in which agents maintain reflexive distance from their organisational performances while continuing to perform them; the analysis is consistent with the present account, but the mechanism rests in the architecture rather than in the agents’ relation to it.

§8.4 Historical Validation#

The N-th Order State and the three-mechanism foreclosure operate invariantly across widely different organisational contexts. The invariance is the argumentative ground for treating the state as a feature of institutional life rather than a contingent property of any particular case, and three historical cases instantiate the apparatus.

§8.4.1 Soviet Production Statistics (1920s-1980s)#

Soviet enterprises operated under a command economy that required fulfillment of centrally planned production quotas. Berliner (1957) and Kornai (1992) document how enterprises systematically reported meeting quotas everyone in the chain knew were fictional. The recursive acknowledgement structure was four-deep. Factory managers knew their statistics were inflated. Workers knew the managers knew. Ministry officials knew the factories were inflating. Central planners knew the ministry was aggregating inflated returns. And every participant knew that every other participant knew.

Yet the reporting continued because the Soviet system required the performance of plan fulfillment. The gap between actual production ($K$) and official statistics ($F$) was openly acknowledged in private while $F$ was maintained in public. The performance itself became the product (Nove, 1977). The three mechanisms operated in coordinated foreclosure across the period: credentialist defence (only Gosplan-issued framework counted as legitimate; reform proposals articulated outside the Gosplan register lacked institutional standing); performance amplification (planning bureaucracies grew, reporting frequencies increased, framework intensification substituted for framework revision through every reform episode from the New Economic Policy through the Liberman reforms); and source isolation (the dissident analytical economists were marginalised institutionally, with substantive critiques absorbed only as ammunition for inter-bureaucracy conflict rather than as a route to framework revision).

The system persisted in this stable N-th Order state for decades. External shock (glasnost policies opening the public-acknowledgement channel beyond the architecture’s processing capacity) triggered rapid collapse (Goldman, 1991). Internal reform had been foreclosed throughout the seventy-year period by the three mechanisms operating in coordination.

§8.4.2 NASA Challenger Disaster (1986)#

Vaughan’s (1996) investigation of the Challenger launch decision reveals the same state in an engineering organisation operating under conditions Beer’s VSM was designed to handle. Engineers had documented O-ring concerns for years; management had received the documentation; Thiokol engineers argued against the cold-weather launch the night before the disaster. Every participant in the chain recognised the safety concerns the documentation articulated. The recursive acknowledgement was as deep at NASA as it had been in the Soviet planning ministries.

The launches continued because organisational commitment to schedule and budget targets exceeded the institutional capacity to act on explicit safety knowledge. Vaughan termed the resulting pattern “normalisation of deviance,” the slow drift through which known violations of safety standards become routine. Reading the case through the three-mechanism apparatus: “do not launch” (articulated in $K$-language about engineering judgement and felt risk) was attenuated through the absence of $F$-compatible justification within the flight readiness review process; “is this within flight readiness review process?” (articulated in $F$-language about institutional procedure) was absorbed without disturbance to the procedure itself; the hybrid articulations Thiokol engineers attempted in the launch-readiness meeting were decomposed into the $F$-compatible elements (which entered the launch decision) and the $K$-elements (which did not).

The state was stable until external shock (the physical disaster, the subsequent Rogers Commission with subpoena power beyond NASA’s institutional processing capacity) generated reform from outside the system that could not be generated from within.

§8.4.3 The 2008 Financial Crisis Risk Models (1990s-2008)#

Financial institutions deployed Value-at-Risk models for risk assessment despite widespread recognition of their inadequacy. MacKenzie (2011) and Taleb (2007) document the epistemic structure with depth comparable to Vaughan’s analysis of NASA. Traders knew the models did not capture tail risk. Risk managers knew the traders knew. Regulators knew the risk managers knew. Academic literature documented the limitations publicly throughout the 1990s and 2000s.

Yet the models remained operative because they provided the legitimate language for risk assessment, “what one uses” for regulatory compliance and internal reporting (MacKenzie and Spears, 2014), and the three mechanisms operated as before. Credentialism: only Basel-compliant model risk frameworks qualified as legitimate, and substantive critiques articulated outside the Basel register attenuated to insignificance in regulatory practice. Performance amplification: the response to LTCM’s 1998 collapse was elaboration of Value-at-Risk methodology rather than questioning of the assumption that tail risk admits parametric modelling. Source isolation: the analysts who articulated tail-risk concerns in non-VaR language (Taleb, Roubini, Rajan) were institutionally marginalised in the pre-2008 period as fringe figures, with their analytical content unable to gain regulatory standing until after the crisis demonstrated its substantive correctness.

When the crisis arrived, every participant claimed surprise despite the widespread prior knowledge of model limitations. The surprise was not about the models breaking down; it was about their inadequacy becoming publicly undeniable in a manner that exceeded the institutional processing capacity (Engelen et al., 2011). External shock from the financial collapse itself generated regulatory reform that decades of internal critique had not.

The three cases converge in their structure across three widely different institutional contexts: a planned economy across seventy years, a government engineering agency across two decades, and a deregulated financial sector across two decades. The cases share no domain features beyond the formal ones the N-th Order State and the three mechanisms specify. This invariance across structure is the argument: the state is not a feature of Soviet planning, NASA engineering culture, or financial regulation but a feature of institutional architecture, and it appears in any organisational form that combines formal frameworks with substantive activity those frameworks cannot capture.

§8.5 Phase Transition Dynamics#

Organisations enter the N-th Order State through a four-stage transition. The stages are sequential and cumulative: each establishes the conditions for the next.

Stage 1: Local Adaptation. Small gaps emerge between official framework and actual practice. A team discovers the ticketing system does not capture their workflow and maintains a shadow spreadsheet. Another team finds the testing framework does not fit their architecture and develops informal practices. Individual managers find prescribed metrics do not capture meaningful progress and supplement formal reports with informal updates. The adaptations seem reasonable, even necessary, local optimisations improving effectiveness that no participant views as subversive but as pragmatic responses to framework limitations. The organisation explicitly tolerates the gap: “use the official system for reporting, but do what actually works” (Orlikowski, 2000).

Stage 2: Proliferation. More teams develop similar workarounds. The shadow systems grow elaborate, sometimes more so than the official frameworks. Knowledge about how things actually work circulates informally through hallway conversations, tribal knowledge, and onboarding that happens outside official channels (Brown and Duguid, 1991). Leadership acknowledges the duality implicitly: “we know the process is bureaucratic; just get the real work done and we will handle the paperwork.” The gap between $K$ and $F$ is recognised, and not problematised (Zuboff, 1988).

Stage 3: Performance Becomes Primary. At a critical threshold, maintaining both systems becomes the primary organisational work. Substantial time is devoted to translating actual work into official framework language for reporting. Teams spend more time on optics, making work visible in $F$-terms, than on the work itself. Meetings focus on what they will report rather than what they will do (Alvesson and Spicer, 2016). The organisation has crossed the phase boundary: it is no longer using $F$ to accomplish goals, it is maintaining performance of using $F$ while goals get accomplished despite $F$. As one practitioner put it, “we spend 60% of our time making the other 40% look good in the reports” (Graeber, 2018, p. 97).

Stage 4: Terminal State. The shadow system and official framework achieve complete separation and reciprocal stabilisation. Every participant operates in a shadow system, every participant reports through the official framework, every participant knows that every other participant does this. The performance is complete, stable, and recognised as performance. Escaping this state requires external shock because all internal language is $F$-language, and $F$-language cannot express its own inadequacy (Brunsson, 1989).

The transition can be characterised through phase-space concepts. The N-th Order State corresponds to a meta-stable attractor in the institutional state-space. Let $x$ represent institutional state and $x^*$ the N-th Order equilibrium. Perturbations $\delta$ below a critical threshold $\delta_c$ are absorbed by the three-mechanism apparatus: the system returns to $x^*$. Perturbations exceeding $\delta_c$ bypass the apparatus and the system exits the basin of attraction. The argument is qualitative, characterising the shape of the phase space rather than modelling specific trajectories. The historical cases of §8.4 show the empirical pattern: long stability punctuated by sudden catastrophic transition rather than gradual degradation. The transition discontinuity is inherent rather than accidental: the three mechanisms maintain $x^*$ against any perturbation they can process, and the system exits this state only when perturbations exceed the apparatus’s processing capacity.

§8.6 The Demand for Better Metrics as Displacement#

Organisations in terminal epistemic states exhibit a characteristic response to a framework’s inadequacy: demand for more elaborate versions of the inadequate framework. “These metrics are not working; we need better metrics.” “This methodology fails; let us try a more rigorous methodology.” “The model does not capture complexity; we need a more complex model.”

The response sounds reasonable, since improving inadequate tools makes sense as a default heuristic, but it displaces an ontological problem into the methodological domain (Power, 1997). The issue is not that current metrics are poorly designed; it is that organisational phenomena exceed what metrics can capture. The response misidentifies the category of problem, treating ontological limitation as methodological insufficiency (Flyvbjerg, 2001).

The pattern is recognisable in OKR implementations, where an organisation adopts Objectives and Key Results and, after two years, the workforce recognises that OKRs create perverse incentives and do not capture meaningful progress. Rather than questioning whether objective-setting frameworks can capture emergent organisational work, the consultancies propose “improved OKR implementation,” “transitioning to Evidence-Based Management,” or “integrating with North Star metrics” (Pfeffer and Sutton, 2006). Each new framework arrives with elaborate theoretical backing, promises to repair the previous framework’s limitations, and enters with organisational fanfare. The consultancy industry depends on the cycle: frameworks must prove inadequate enough to justify replacement, while remaining appealing enough that the inadequacy does not generalise into questioning frameworks-as-such (Kieser, 2002). Each new framework eventually reveals the same inadequacy, by which point the organisational memory has faded, the consultancies have moved on, and new consultancies arrive proposing the next framework. The cycle perpetuates through the framework’s inadequacy rather than despite it, since each round of inadequacy justifies the next attempt (March, 1991).

The displacement is not the product of stupidity; it is anxiety management (Heidegger, 1927/1962). Acknowledging that organisational phenomena exceed formalisation entirely creates existential difficulty for management structures premised on measurement and control. The institutional response prefers the belief that the right framework has not been found yet to the recognition that no framework can exist (Alvesson and Willmott, 2002).

The Part V mathematical argument bears on the displacement directly. The cascade demonstrates that even the most expressive available formalisms, sheaf cohomology, derived categories, homotopy type theory, articulate rather than capture the chiasmic features of organisational phenomena. If those formalisms cannot capture the phenomena (and Part V’s argument is that they cannot), then elementary formalisms (arithmetic for counting tickets, basic statistics for velocity calculations) fall short a fortiori. The demand for “better metrics” amounts to the demand for a more elaborate formalism within a register the horizon-mismatch theorem of §24 has already established as inadequate to the substrate. The demand is not unreasonable in its register; it is incoherent in its category.

Part III: The Phenomenological Turn to Chiasmic Organisation#

§9. Merleau-Ponty’s Chiasmic Ontology#

§9.1 The Chiasm#

Merleau-Ponty’s late philosophy, in The Visible and the Invisible (1968), supplies the ontological framework for chiasmic organisation. The chiasm, from the Greek chi (Χ), names the crossing or intertwining structure underlying perception and action. Merleau-Ponty’s central formulation (1968, p. 139): “the chiasm, reversibility, is the idea that every perception is doubled with a counter-perception. One no longer knows who speaks and who listens.” The reversibility is a simultaneous ambiguity rather than an alternation between distinct states, both aspects present at once. This is not a property some organisations have; it is what organisations are once the observer-observed separation is recognised as untenable.

§9.2 The Touching-Touched Illustration#

The classical illustration is tactile self-perception: when the right hand touches the left, Merleau-Ponty observes that “my hand while it is felt from within is also accessible from without. If it takes its place among the things it touches, [it] is in a sense one of them” (1968, p. 133). The touching hand is also touched. Subject and object are aspects of a single phenomenon. The reversibility occurs without coincidence: the touching hand and the touched hand are not identical at any moment, yet neither precedes the other as a pre-existing entity. The écart (the gap) between touching and touched is what makes both possible. As Merleau-Ponty puts it elsewhere, “there is a circle of the touched and the touching, the touched takes hold of the touching” (1968, p. 148).

§9.3 Flesh as Element#

Merleau-Ponty introduces flesh (chair) as the element of reversibility itself. His own definition (1968, p. 139): “the flesh is not matter, is not mind, is not substance. To designate it, we should need the old term ‘element,’ in the sense it was used to speak of water, air, earth, and fire, that is, in the sense of a general thing, midway between the spatio-temporal individual and the idea.” Flesh is what subject and object share as their common tissue, the formative medium of both. Merleau-Ponty’s later formulation (1968, p. 147) describes flesh as “the formative medium of the object and the subject.” The element is not added to the objects it connects; it is the medium through which they emerge as objects at all.

§9.4 Organisational Implications#

The chiasmic ontology has direct organisational implications: at every organisational boundary, inside becomes outside and outside becomes inside. The customer service representative embodies both company and customer in a single role. The salesperson creates needs while discovering them. As Merleau-Ponty notes, “the boundary is not a line but a zone of reciprocal insertion” (1968, p. 123). The recognition recovers what Beer’s VSM could not formalise: organisational boundaries that constitute the entities they distinguish, rather than separating antecedent units.

Organisations do not observe environments; they are visible within the same field they observe. As Merleau-Ponty puts it, “he who sees cannot possess the visible unless he is possessed by it, unless he is of it” (1968, p. 134). The reversibility of seeing and being-seen makes the observer-stance the VSM presupposes unavailable in principle.

What allows organisational visibility is itself invisible, a position Merleau-Ponty’s formulation captures: “the invisible is not another visible. It is the invisible of this world, that which inhabits this world, sustains it, and renders it visible” (1968, p. 152). The constitutive invisibility is the chiasmic-organisation parallel of what the present paper develops elsewhere as the constitutive outside: the conditions that must remain non-thematic for the visible aspects of organisational life to operate at all.

§10. Barad’s Agential Realism#

§10.1 Intra-Action and Interaction#

Karen Barad (2007) extends the chiasmic insight through agential realism. Where interaction presupposes pre-existing agents that subsequently encounter each other, Barad’s neologism intra-action describes the mutual constitution of agencies through their entanglement. Barad’s foundational formulation (2007, p. 33): “the neologism ‘intra-action’ signifies the mutual constitution of entangled agencies. That is, in contrast to the usual ‘interaction,’ which assumes that there are separate individual agencies that precede their interaction, the notion of intra-action recognizes that distinct agencies do not precede, but rather emerge through, their intra-action.”

For organisations, the implications are direct: workers and managers do not pre-exist their organisational relations. The cut that separates them, the agential cut in Barad’s terminology, creates the entities it separates rather than recognising antecedent ones. Customer and provider, employee and employer, internal and external: each pair emerges through the apparatus that distinguishes them, with the apparatus itself part of what the distinction produces.

§10.2 Cutting Together-Apart#

Barad’s formulation captures the dual operation: “cutting together-apart (one move) is how differentiating-entangling is enacted” (2007, p. 176). The move differentiates (creates the cut between agencies) and entangles (binds the agencies in mutual specification) in a single operation. The organisational examples are immediate: organisational charts cut roles together-apart, the chart simultaneously separating the roles (employee A is not in role X) and binds them in a single specification (the entire role system is defined together). Strategic plans cut markets together-apart: defining the target market separates it from non-target markets, and the operation that defines it is also the operation that produces the entire market typology in which the target sits.

§10.3 Apparatus and Phenomenon#

Barad’s central claim about measurement (2007, p. 142): “apparatuses are not passive observing instruments. On the contrary, they are productive of (and part of) phenomena.” Organisational measurement apparatus produces what it measures, becomes part of the organisational phenomenon, and shapes the future operation of the organisation that deploys it. KPIs do not measure performance; they construct what counts as performance. Dashboards do not display organisational state; they constitute the organisational state they display. The recognition extends Maturana’s “everything said is said by an observer” into the domain of measurement: every measurement is made by an apparatus that is part of what it measures. As Barad puts it, “apparatuses are boundary-making practices” (2007, p. 206).

§11. Conway’s Law and Its Reversal: Empirical Validation#

§11.1 The Chiasmic Nature of System Architecture#

Melvin Conway’s (1968) original observation, that “organizations which design systems are constrained to produce designs which are copies of the communication structures of these organizations,” reveals a chiasmic dynamic in software practice. The empirical literature (MacCormack et al., 2012; Colfer and Baldwin, 2016) demonstrates that the reverse direction is equally operative: systems shape the organisations that maintain them. The two-way relation, what may be called Conway’s Chiasm, runs in both directions simultaneously: organisational structure constitutes system architecture, and system architecture constitutes organisational structure, with neither term temporally prior to the other.

§11.2 Evidence from Software Development#

Three empirical studies anchor the case. Mockus, Fielding, and Herbsleb (2002) studied Linux kernel development and the Apache project, finding that code architecture shapes developer communities (architectural modularity drives community formation around components) and that developer communities reshape code architecture (community structure drives architectural evolution). The two directions co-evolve, with neither preceding the other in any temporal sequence the study could identify.

Howison and Crowston (2014) studied open source projects more broadly, finding the same pattern of continuous mutual specification: project architectural decisions and contributor relationships shape each other through ongoing recursive operation. Cataldo, Wagstrom, Herbsleb, and Carley (2006) studied commercial software development at Bell Labs and confirmed that technical dependencies create organisational dependencies and that organisational dependencies create technical dependencies, with the intertwining inseparable in the empirical record.

The strongest single empirical confirmation comes from MacCormack, Baldwin, and Rusnak (2012), discussed at length in §33.3 of this paper. Their test of the mirroring hypothesis uses matched pairs of software products with identical functional requirements developed by organisationally tightly-coupled commercial firms versus organisationally loosely-coupled open-source communities. In every matched pair examined, the product developed by the loosely-coupled organisation exhibited significantly higher architectural modularity than the product from the tightly-coupled organisation. The effect size reached a factor of six in propagation cost between component pairs. The mirroring hypothesis was supported without exception across the matched pairs.

§11.3 Why Planned Boundaries Do Not Hold#

The empirical literature on organisational boundary work converges on the same chiasmic conclusion. Mergers collapse at rates of 50-70% according to Cartwright and Schoenberg’s review (2006), with the collapses typically attributed to “cultural incompatibility.” The chiasmic reading interprets cultural incompatibility as incompatibility of intertwined chiasmic structures: what cannot merge is what is internally intertwined differently in each pre-merger organisation, and the post-merger organisation cannot impose a coherent intertwining on the combined units through any planning operation external to the units’ own ongoing self-constitution.

Studies of digital transformation (Vial, 2019) document the same pattern across a different domain. Boundaries between technology and organisation prove resistant to the planned reconfiguration that transformation programmes assume. The boundaries are zones of transformation rather than lines of separation, and the transformation programmes themselves become part of what they were intended to operate on. Studies of agile transformations (Dikert, Paasivaara, and Lassenius, 2016) document the same pattern within a more bounded scope: top-down agile transformations collapse at rates above 80% in large enterprises, with the collapses attributable to the assumption that an agile organisation can be designed and then implemented, rather than the chiasmic recognition that the agile organisation emerges through the practices that constitute it.

§12. Gibson, Varela, and Enactive Cognition#

§12.1 Affordances as Relational#

James Gibson’s (1979, p. 127) concept of affordances anchors the chiasmic reading of organisational capabilities: “the affordances of the environment are what it offers the animal, what it provides or furnishes, either for good or ill.” Affordances exist only in the organism-environment relation. They are real but properties of neither party alone, emerging through interaction history rather than residing in either organism or environment in isolation. The organisational analogue is direct: markets afford opportunities only relative to organisational capabilities, capabilities afford action only relative to environmental conditions, and neither markets nor capabilities exist independently of the chiasmic relation through which both come into focus.

§12.2 Enaction and Sense-Making#

Varela, Thompson, and Rosch (1991, p. 173) develop the enactive account: “cognition is not the representation of a pregiven world by a pregiven mind but is rather the enactment of a world and a mind on the basis of a history of the variety of actions that a being in the world performs.” Organisations enact opportunities through their actions, co-create environments through their participation, and perform capabilities rather than possessing them as static properties. As Varela and colleagues argue, “the organism and environment are mutually specified” (1991, p. 205).

Karl Weick’s organisational sensemaking research aligns with the enactive account. His central formulation about retrospective sensemaking has already appeared in §2.3: “people often don’t know what they think until they see what they say” (Weick, 1995, p. 30). Organisations create environments through their categorical operations, enact constraints through their procedural decisions, and produce opportunities through their attentional patterns. Weick’s closing formulation captures the position: “enacted environments are outputs of organizing, not inputs to it” (1995, p. 34).

§12.3 The Chiasmic Organisation#

The four concepts: reversibility (Merleau-Ponty), intra-action (Barad), enaction (Varela, Thompson, and Rosch), and flesh (Merleau-Ponty), together specify the chiasmic-organisation ontology. The argument is ontological rather than metaphorical: organisations are patterns within fields they continuously co-create, exhibiting reversibility of observer and observed, intra-action through which agencies emerge, enactment through which environments come into being, and the flesh-medium through which the recursive relations are sustained.

§13. The Mathematical Requirement#

Set theory cannot represent chiasmic phenomena. The formalism requires that elements be either in a set or not, that boundaries be sharp (characteristic functions $\chi: X \to {0,1}$), and that sets be distinct or identical without interpenetration. Organisational reality systematically presents elements simultaneously inside and outside their containing categories, boundaries operating as zones of transformation rather than separation, interpenetration without coincidence between organisational subsystems, and reversibility without identity between observer and observed. The mismatch is direct, and no refinement of set-theoretic formalism resolves it: set theory’s foundational commitments are the commitments the chiasmic ontology has displaced.

Systems theory in the von Bertalanffy (1968) and Checkland (1981) tradition assumes definable boundaries, input-output relations, feedback loops connecting separate entities, and observers who model systems objectively. Each assumption has been undermined in the preceding sections. The chiasmic ontology requires mathematics capable of articulating boundaries as zones rather than lines, organisations without inputs or outputs because of operational closure, simultaneous feedforward and feedback, and the inseparability of observer and observed.

Luhmann (1995, p. 37) approaches the requirement with his formulation that “the system is the difference between system and environment.” The formulation still treats the difference as pre-existing the operation that produces it, where the chiasmic ontology and Barad’s intra-action analysis treat the difference as itself enacted in operation. The mathematical apparatus required to articulate the chiasmic features must accommodate this enactment.

Four formal apparatuses point in the right direction. Category theory (Mac Lane, 1971) articulates relations between relations, with morphisms primary rather than objects. Topos theory (Johnstone, 1977) provides internal logic and sheaf-theoretic local-to-global structure. Homotopy type theory (Univalent Foundations Program, 2013) handles paths and equivalences as constitutive rather than derived structure. Derived categories (Verdier, 1996) accommodate transformations with irreducible remainder. The four apparatuses operate predominantly in the structural-analogue register that §25.3 below makes explicit. They articulate the chiasmic features rather than constructively embedding chiasmic organisations into specific mathematical objects, because the chiasmic argument shows constructive embedding from any observer position would itself violate the chiasm it sought to formalise. The proof-register foundation supplied by the horizon-mismatch theorem (§24) establishes the floor of the cascade. The four apparatuses extend the structural argument upward.

Part IV: Mundane Complexity in Everyday Organisational Life#

§14. Methodological Note#

The narrative that follows is a phenomenological reduction rather than an anecdote. It examines an ordinary organisational moment through the theoretical apparatus developed in Parts II and III. The chiasmic, autopoietic, and recursive phenomena it identifies are the substance of daily organisational life rather than rare exceptions requiring special circumstances.

Every element of the scenario, including Sarah’s embodied knowledge, the bug report’s resistance to categorisation, the manager’s demand for metrics, Tom’s emergent debugging strategy, Mark’s marginalised support work, instantiates principles derived in Parts II and III. Practice and theory specify each other through the narrative; the chiasmic relation between them is itself an instance of the chiasm the paper has been articulating.

The scene is ordinary because the extraordinary is ordinary. A Tuesday morning at a software company contains the multiplicities the theoretical apparatus identifies.

§15. Tuesday Morning, 9:47 AM: A Bug Report#

Sarah, a backend developer at a mid-sized software company, receives a bug report from customer support. The customer’s API calls are failing intermittently. The situation seems simple enough on its face: debug the code, find the issue, deploy a fix. The narrative that follows traces what actually happens through the theoretical lenses Parts II and III have developed.

The bug report arrives through Jira. Sarah sees the notification because she is designated as the API team lead, a role that exists because the organisation drew a distinction (Spencer-Brown) between API team and not-API team. She also works on the payment system, which makes her simultaneously inside and outside the API team boundary. When she opens the ticket, she is the API team. When she switches tabs to her payment code, she is not. The boundary exists only through her enactment of it.

§15.1 The Information That Is Not#

The bug report contains what Bateson would recognise as news of difference: the API worked before, now it does not. But which differences do not make a difference here? The customer’s frustration level is not captured in the ticket. The support agent’s previous interaction with this customer is not linked. The fact that three other customers experienced similar issues but did not report them is invisible non-difference. Sarah receives a transformed, coded version of events that preceded it rather than information in any unmediated sense. The map (the bug report) is not the territory (the actual system fault).

Meanwhile the support agent, Mark, has already told the customer “our engineering team will fix this quickly,” creating what will become a double bind. Sarah must fix it quickly to satisfy the customer promise and also fix it thoroughly to prevent recurrence. She cannot do both. She also cannot comment on the contradiction because doing so would mean throwing Mark under the bus. The four conditions of Bateson’s double bind (1972, p. 208) are all present: primary injunction (fix quickly), secondary injunction (fix thoroughly), inability to escape the field, and prohibition on commenting on the contradiction.

§15.2 The Autopoietic Response#

Sarah’s response is not determined by the bug report itself; it is determined by her organisational structure (Maturana and Varela). The bug report is a perturbation rather than an instruction. Her internal structure, shaped by years of similar tickets, previous fixes, and learned patterns, determines her response. She immediately thinks “probably a rate limiting issue,” not because the ticket says so, but because her structural coupling with the system over time has produced this pattern recognition.

She sends a Slack message to Tom on the infrastructure team: “are we seeing elevated API traffic?” Tom is in a meeting but sees the notification. He is now simultaneously in the meeting (bodily present) and in the Slack conversation (digitally present). His manager is talking about fourth-quarter planning while he checks Grafana dashboards. The organisation exists in multiple overlapping presents that cannot be synchronised into a single temporal frame.

§15.3 The Chiasmic Moment#

Tom discovers that API traffic is up three-fold because a large customer has started using a new integration. The chiasm becomes visible at exactly this point: the internal infrastructure problem is caused by external customer behaviour, but the customer’s behaviour was enabled by the company’s own business development team who promised unlimited API calls. Inside becomes outside; the company created the customer behaviour that now appears as external pressure.

Sarah realises she needs to implement rate limiting, but this will affect the customer causing the problem. She must communicate with sales, who sold the unlimited plan. Sales reports to a different vice president who has different OKRs. Sarah’s world (engineering efficiency) and their world (customer satisfaction) operate with different logics, different truth values. What is “true” in engineering (we need rate limits) is “false” in sales (we promised unlimited). The two organisational subsystems operate with incompatible codes, in exactly the form Luhmann’s social-systems theory predicts.

§15.4 The Observer Creating Reality#

The vice president of engineering, David, asks for a status update. In preparing her response, Sarah creates the reality she is reporting. She writes: “we have identified the root cause as insufficient rate limiting.” But was that really the root cause? Or was it the sales promise? Or the infrastructure capacity? Or the business model? The act of observation, writing the status update, creates “the root cause” from a field of possible causes. Von Foerster’s recognition holds with full force at the keyboard: Sarah constructs the reality she will subsequently report as discovered.

David reads the update and thinks “we need better monitoring.” He has now created a different reality, one in which the problem is observability rather than capacity or promises or business models. Each observer creates a different environment through their categorical operations on the same underlying field.

§15.5 Recursive Blind Spots#

Sarah cannot see her own pattern matching as construction; she takes it as discovery. David cannot see his framing as construction; he takes “better monitoring” as identifying what is real. Sales cannot see how their promises shape what becomes a problem; the promises arrive in their consciousness as customer needs. Mark in support cannot see how his interaction shapes what gets reported; the reports arrive as just what the customer said.

Each participant operates with blind spots that are constitutive rather than incidental. The blind spots are not failures of attention or competence; they are the consequence of the participant’s situated position in the field, which makes some aspects of the field visible by making others invisible. The pattern is Merleau-Ponty’s chiasm operating at organisational scale: every visibility is doubled with a counter-invisibility that makes it possible.

§15.6 Learning Levels Colliding#

The automated test suite operates at Learning Zero, mechanical responses to predetermined inputs without learning. Sarah operates at Learning One, applying pattern recognition from past tickets to the immediate fix. Tom operates at Learning Two, questioning capacity planning at the framework level: should the infrastructure be designed for these traffic patterns at all? The framework-level question (should the company sell unlimited API calls?) belongs to Learning Three. Nobody asks the Learning Three question. The organisation has the cognitive capacity for it; what it lacks is the institutional standing for the question to gain traction. This is exactly the §3.4 structure: VSM-style organisations operate at Learning Two with occasional moments aspiring to Learning Three, but the question that would constitute Learning Three is unavailable from within the framework that would have to be questioned.

§15.7 Eigenforms Emerging#

“The way we handle scaling issues” emerges as a stable form across the operations: implement rate limiting, communicate with affected customers, log the incident in the runbook, postpone the deeper question about business model. The pattern stabilises through recursive operation. Each time the team executes the pattern, the pattern reinforces. After enough iterations, “the way we handle scaling issues” appears as an object in organisational reality, something with properties and characteristics that can be referred to and reasoned about. It is, in von Foerster’s terms, an eigenform: the stable product of recursive operation rather than an antecedent feature of the organisation.

§15.8 Re-Entry and Paradox#

The system attempting to observe itself creates Spencer-Brown re-entry. The monitoring system that watches API performance is itself making API calls, contributing to the load it measures. The Grafana dashboard showing system health affects system health by consuming resources. The Slack conversation about communication problems is itself a communication problem (too many channels, too many interrupts). Sarah’s fix, implementing rate limiting, will be deployed through the CI/CD pipeline that itself makes API calls that could be rate limited by the fix she is implementing. The system enters itself, creating the kind of paradox Spencer-Brown showed requires imaginary values for formal resolution.

§15.9 Structural Coupling Across Time#

While all this happens, the organisation and its environment are structurally coupled across multiple timescales. At milliseconds, the API and the customer request cycle. At minutes, Sarah and the debugging session. At hours, the ticket and the customer satisfaction metric. At days, the sprint and the feature delivery. At quarters, the OKR and the market position. At years, the technical debt and the capability erosion.

Each timescale has its own coupling dynamics, and they are not synchronised. A decision made at the quarterly scale (promise unlimited API calls) creates perturbations at the millisecond scale (API overload) that create responses at the hourly scale (debugging) that accumulate at the yearly scale (technical debt). The horizon mismatch developed formally in §24 appears here in concrete form: the regulator (the quarterly OKR-setter) operates at a temporal horizon of quarters; the substrate (the API system under load, the engineering team, the customer base, the technical debt) operates at temporal horizons spanning milliseconds to years. The regulator’s measurement frame cannot resolve the substrate’s full variety. The theorem’s inequality $V_M \leq V_R \leq N(T_R) < V_\Sigma$ is exactly what the team experiences as their fix at one timescale generates problems at others.

§16. The Ordinary Impossibility#

By 11:30 AM, Sarah has deployed a fix: rate limiting that excludes the large customer. The fix is not really a fix. It is a temporary stabilisation that creates technical debt (the architecture should handle this load), preserves the double bind (Mark will continue to promise quick fixes), defers the framework-level question (should the business model continue to sell unlimited capacity?), deepens structural coupling (the mobile team will now need to handle the cascade effects from rate limiting), adds to recursive blind spots (the monitoring will now treat rate-limited traffic as success), and operates at Learning One (the deeper question never gets asked).

The Jira ticket is marked resolved. The dashboard shows green. The KPI for mean time to resolution improves. Every theoretical lens applied in this section shows that the situation has become more complex, more entangled, more problematic. The metric improves while the substrate degrades.

§16.1 What KPIs Cannot See#

The KPIs show the resolution time of one hour and forty-three minutes, customer impact contained, two systems affected. The actual occurrence involves Sarah’s autopoietic pattern matching, three boundary reversals across the engineering-sales-customer field, two double binds requiring her bodily absorption, observer-dependent construction of “the root cause,” five recursive blind spots across the participants, the emergence of “how we handle scaling issues” as an eigenform, three re-entry paradoxes in the monitoring and CI/CD systems, structural coupling across six timescales, learning-level mismatches between automated tests and the questions that would constitute Learning Three, and the accumulation of technical debt in derived categories that have no metric. The first list is the surface visible to measurement. The second list is the substrate the measurement cannot resolve.

§16.2 The Mundane Reality#

This is not a crisis. This is Tuesday morning. Sarah has three more tickets to review before lunch. Tom has two more meetings. Mark in support has seventeen more customer messages. David has a board deck to prepare in which he will show that mean time to resolution is down 15% this quarter.

Every ticket, every meeting, every customer interaction involves all these phenomena simultaneously. They are not special cases or dramatic events. They are the mundane, everyday reality of organisational life. The complexity is generated by the ordinary operation of organisations as chiasmic, autopoietic, recursively coupled, paradox-generating systems, rather than by unusual circumstances or exceptional events.

§17. The Human Absorption of Impossibility#

The resolution time of one hour and forty-three minutes looks good on the dashboard. The ticket is closed. The KPI is green. The question for this section is how this actually happens, and what it costs.

Sarah skipped her morning break to debug the issue. She has been skipping breaks for three weeks because ticket velocity is being tracked. Her neck hurts from hunching over her laptop. She is drinking her fourth coffee of the day. When she implemented the rate-limiting fix, she knew it was a workaround. She knew it would create problems later. The ticket needed to be closed; the metric needed to be green.

The pattern is structurally similar to fighters cutting weight for competition. The double bind is direct: you must be strong enough to fight (requiring muscle mass and hydration) and light enough to qualify (requiring dehydration and muscle loss). Fighters resolve the impossibility by destroying their bodies in the weight-cut period, sitting in saunas until they pass out, wearing plastic suits while running, taking diuretics that damage their kidneys. Some die. Others suffer permanent organ damage. They make weight. The metric is achieved at the cost of the substrate that produces it.

Sarah is making weight in the same structural sense. She must be fast enough to close tickets (requiring quick fixes) and thorough enough to prevent future issues (requiring careful design). She resolves the impossibility by destroying her well-being: skipping meals, working weekends, implementing solutions she knows are incomplete. The ticket closes. The KPI turns green. Her body absorbs the contradiction.

Tom checked the Grafana dashboards while his manager was presenting fourth-quarter plans. He absorbed the double bind into his body: physically present in the meeting, mentally debugging infrastructure. Tonight he will dream about API endpoints. His partner will ask why he seems distant at dinner. He will not be able to explain that he is simultaneously at home and at work, that the boundary dissolved at 9:47 AM when Sarah’s Slack message arrived.

Mark in support has managed customer expectations through careful emotional labour, crafting messages that sound helpful while promising nothing. He has absorbed customer frustration into his body, metabolising their anger into careful phrases. His empathy is being extracted and commodified, measured as customer satisfaction while destroying his capacity to feel. The linguistic labour, the careful crafting of non-promises that sound like promises, exhausts him in ways that no metric captures.

The KPIs work because humans absorb the impossibility. The absorption operates through three modes simultaneously: somatic absorption (Sarah’s neck, Tom’s sleep, Mark’s empathy), social absorption (the boundary dissolution between work and home, the strained relationships at dinner), and biographical absorption (the cumulative erosion of capacities over months and years that no quarterly measurement registers).

§18. The Three Sites of Extraction#

When KPIs create impossible demands, three sites absorb the impossibility: the worker, the substrate, and the externality. The three-site framework articulates what is happening across the Tuesday-morning scenario in its underlying structure.

Worker-site absorption. Sarah works sixty-hour weeks to keep the tickets flowing. She makes it work by coding during meetings (partial attention to everything), working evenings (time taken from her life), taking shortcuts she knows will create future problems, building mental models too complex to document, and maintaining relationships that hold the system together despite the system itself. The KPIs look good. Sarah is looking for a new job. She will leave in four months and take irreplaceable knowledge with her. Her replacement will not understand why certain things are done in certain ways. The system will degrade slowly and invisibly, until it collapses without warning.

Substrate absorption. The team makes the KPIs work by marking tickets resolved when they are really worked around, implementing fixes that are technical debt, creating special cases that make the codebase unmaintainable, skipping documentation because velocity is measured and documentation is not, and building fragile solutions that work today and break tomorrow. The KPIs improve quarter over quarter. The codebase becomes increasingly brittle. Every change takes longer. Every fix creates new bugs. This quarter’s metrics look good. Next quarter is someone else’s problem.

Externality absorption. The organisation makes the KPIs work by pushing complexity elsewhere. Rate limiting pushes load to the mobile team. Quick fixes push problems to future quarters. Local optimisations push failures to interfaces. Departmental successes push costs to other departments. Current solutions push debt to future teams. The dashboards stay green. The debt is invisible. It accumulates in derived categories that have no metric. One day, adding a simple feature becomes impossible and no one can explain why. The organisation has painted itself into a corner one green KPI at a time.

The three sites operate simultaneously rather than in alternation. Worker burnout is co-occurring with quality destruction is co-occurring with externality accumulation, and the metric improvement that drives all three is the same metric improvement.

§19. Tuesday’s Hidden Cost#

Sarah has handled four tickets today. The metrics show four resolutions, average resolution time of one hour and twelve minutes, customer satisfaction score of 4.3 out of 5. The actual cost includes seventeen micro-decisions that will haunt the codebase in months to come, three double binds absorbed and manifest as shoulder tension, forty-three context switches without achieving deep focus on any of them, a promise to Tom to review his infrastructure proposal (which she will not have time to do), and accumulating mental models that exist only in her head.

Tom has attended three meetings while monitoring seventeen Slack channels. He has agreed to commitments in meetings while debugging issues in Slack. His brain maintains parallel tracks that cannot be synchronised. He will wake at 3 AM thinking about API rate limits.

Mark has managed customer expectations through careful emotional labour. He has absorbed customer frustration into his body, metabolising their anger into careful phrases. His empathy is being extracted and commodified, measured as customer satisfaction while destroying his capacity to feel.

The KPIs show tickets resolved, customer satisfaction trending up, system uptime at 99.9%. The same KPIs do not show three people approaching burnout, technical debt approaching critical mass, knowledge residing in exhausted brains and undocumented mental models, and systems maintained by human sacrifice.

§20. The Logical Impossibility#

That the KPIs work does not make them logical, sustainable, or even short-term viable. The KPIs operate by extracting unsustainable human effort, hiding accumulating complexity, pushing problems to invisible futures, destroying quality while measuring quantity, and creating the pressure that ensures the degradation continues.

The arrangement is not logical. It is not sustainable. It is barely short-term viable. The dashboard is green so it must be working. Tomorrow Sarah will quit, or the system will crash, or the technical debt will make further change impossible. Or all three. The KPIs will look good right up to the moment they do not. Then everyone will wonder why no one saw it coming.

What the KPIs cannot see is human exhaustion accumulating, technical debt compounding, knowledge evaporating, quality degrading, and complexity accumulating. The metrics work by not measuring what matters. They succeed by defining success as what can be measured while ignoring everything else. They function by forcing humans to absorb impossibility into their bodies, their relationships, their lives.

Sarah’s resolution time of one hour and forty-three minutes is both true and a lie. True: the ticket closed. Lie: everything else about what happened, what it cost, what it means, and what it is building toward.

§21. The Tracking Impossibility#

To track what actually happened on Tuesday morning, one would need to map autopoietic responses (impossible: internal states are inaccessible to external measurement), trace boundary reversals (impossible: boundaries are enacted in operation rather than fixed for measurement), identify all double binds (impossible: most are unspeakable within the discourse that generates them), track observer constructions (impossible: requires observing the observation), map blind spots (impossible: one cannot see what one cannot see), detect eigenforms (impossible: they appear as natural reality rather than as recursive products), resolve re-entry paradoxes (impossible: requires Spencer-Brown’s imaginary values for formal treatment), synchronise time scales (impossible: different scales have different logics), measure learning levels (impossible: Learning Three is the level at which the measurement framework would have to be questioned), and calculate accumulating remainder (impossible: it exists in derived categories with no metric structure).

The ten impossibilities are the empirical manifestation of the horizon-mismatch theorem from §24 applied to the substrate of Tuesday morning. The substrate’s variety exceeds the measurement frame’s resolution. Every element of the morning is invisible in principle to the apparatus that is supposed to manage the morning.

§22. The Mundane Conclusion#

By lunch, Sarah is eating a sandwich while code-reviewing a colleague’s pull request. The API issue is resolved without being solved. The organisation has adapted without learning. The complexity has increased while appearing managed. The KPIs look good while measuring nothing real.

This is not a malfunction; it is normal operation. Organisations work through this complexity rather than despite it. The impossibility of tracking does not mean organisations cease to function. It means they function through processes that exceed any possible measurement system.

The question is not how to measure organisational reality but how to navigate skilfully through unmeasurable complexity. The answer cannot be metrics, dashboards, or KPIs. It must be something that engages the simultaneous operation of all these phenomena rather than fighting against them, which is the philosophy of cultivation §27 below sets out in detail.

Tomorrow Sarah will discover that the rate-limiting fix causes cascade failures in the mobile app. The mobile team’s retry logic, combined with the rate limiting she deployed today, will create an unexpected resonance pattern. The cascade will require coordination across three teams, each operating with different assumptions, different priorities, different worldviews. The CEO will ask why a simple bug fix is affecting mobile performance. No one will be able to explain that there are no simple bug fixes in structurally coupled, autopoietic, chiasmic systems with recursive blindness and emergent eigenforms. They will create another dashboard to track cross-team dependencies. It will not help. The organisation will survive, adapt, continue, through the mundane complexity of organisational life that exceeds any possible formalisation. The organisation will persist because it is unmeasurable, maintaining itself through processes that no mathematics, from arithmetic to ∞-topoi, can capture.

This is Tuesday. This is Wednesday. This is every day.

Part V: The Mathematical Apparatus#

§23. The Constituent Cannot Total the Process#

§23.1 What the Argument Is Not#

The argument of this part is easy to mistake for a claim it is not making, and the misreading disarms it, so the boundary is drawn first. The argument does not hold that organisations cannot count. A firm can count units shipped, headcount, cash on hand, revenue booked in a quarter, defects per thousand, days from order to delivery. These quantities are real. They are not approximations of something truer that the firm has not reached. When a company records forty million in sales for a quarter, the forty million happened. No part of this paper denies it, and any reading that takes the paper to deny it has mistaken the target.

The economists’ complexity argument is also not the argument here, though it sits nearby and is worth separating out. Hayek (1945) held that the market’s information is distributed across millions of agents, that no central mind can assemble it, and that the price system economises on the knowledge any single participant needs. The Austrian case is epistemic. It says the calculator is too small for the object: the object is vast, the knowledge is dispersed, and the central planner falls short for want of reach. The complexity economists who followed (Arthur, 1994; the broader Santa Fe programme) sharpened the point with non-linearity and path dependence, but kept its shape. The system is too complex to compute from outside.

The present argument runs underneath that one. The trouble is not that the calculator is too small for the object. The trouble is that there is no outside from which the calculation could be performed, because the would-be calculator is a constituent moment of the process it proposes to total. This is the lesson the second-order cyberneticists drew and the lesson Beer’s own architecture could not absorb without dissolving. The firm does not observe the market and then act in it. The firm is one of the processes through which the market continuously constitutes itself. A retailer setting a price is not reading a market price and responding to it; the price it sets is one of the constituting acts out of which the market price, that statistical artefact, is afterward assembled by an observer who is also inside. The same holds within the firm. Management does not stand outside the organisation observing its state and adjusting inputs. Management is one of the organisation’s own processes, and its observations are operations of the organisation upon itself, in exactly the recursive sense §5 and §6 developed.

The wave cannot measure the ocean’s displacement including its own contribution to the displacement, because the measurement would have to complete before the wave’s own motion completed, and the wave’s motion is part of what is being measured. The organisation is in this position with respect to itself and its market. The incalculability is not a size mismatch between a small calculator and a large object. It is the structural impossibility of a constituting process totalling, from within, a whole that includes its own constituting activity as an unfinished term.

The transition from participant to constituent is not a matter of degree, and the distinction can be drawn without appeal to size. An agent participates in a market when it responds to conditions the collective has established; it constitutes the market when its own acts generate the conditions to which others respond. Four marks separate the two. The first is control of interface: the agent determines how others encounter the economic space, what is visible to them, what alternatives are surfaced, so that others meet the market as the agent presents it rather than directly. The second is control of alternatives: the agent’s position ensures that competing options are absent, absorbed, or rendered unviable, so that exit is formally available and practically empty because there is nowhere to exit to. The third is insulation from exit: network lock-in, data accumulation, and the absence of interoperability ensure that the agent’s offering can degrade without proportionate loss of participation, suspending the discipline that departure would otherwise impose. The fourth is the capacity to set conditions rather than respond to them: the agent does not take prices, it makes them; it does not move within rules, it writes the rules others move within. When these four obtain together, the agent has crossed from participant to constituent.

The crossing shows itself in the sign of the feedback. Below it, perturbations to an agent’s position are self-correcting: raise prices and customers leave, let quality slip and a competitor takes the demand, and the agent is returned toward the collective condition. This negative feedback is the discipline the cybernetic tradition and the market tradition alike rely on. Above the crossing, the same perturbations become self-reinforcing: the agent with the largest base draws more participants, the agent processing the most exchange builds the most complete model of the space, the agent mediating the most flow captures the rents that fund further mediation. The feedback that disciplined the participant now entrenches the constituent. The attractor has changed from distributed condition to monopolistic constitution, and the change is a change of kind rather than of magnitude, because the very mechanism that held the participant in place is what now consolidates the constituent’s hold. A regulator inside such a system is regulating conditions it is itself generating, and the deeper the consolidation, the less its measurements report on anything external to its own constituting activity. This is the precise sense in which the constituent cannot total the process: at the moment its measurements would matter most, they have become the most fully circular.

§23.2 The Two Things a Metric Cannot Do#

Two distinct incapacities follow, and they must be kept apart because they hold for different reasons and a defender of metrics will try to collapse them.

The first incapacity concerns people. A workforce is not a stock of interchangeable units whose output sums. It is a tissue of relationships, accumulated judgement, tacit coordination, and mutual adjustment built up over years, the substance Part IV showed Sarah, Tom, and Mark absorbing into their bodies on an ordinary Tuesday. The relationships are not a quantity with measurement error. They are not a quantity. A person’s standing in a team, the trust that lets one engineer accept another’s verdict without re-deriving it, the half-sentence in a corridor that prevents a fortnight of wasted work: these are constitutive of the team’s capacity and they have no scalar. One can count the engineer’s commits. One cannot count what the engineer is to the team, because what the engineer is to the team is a relation in a continuous process, not a magnitude in a state vector. This is the sense in which people in a complex system are incalculable, and the claim is held without qualification.

The second incapacity concerns trajectory. Even granting every real number the firm books, the metric cannot register what the firm is building or dismantling in the act of booking it. The quarter’s profit is a real number. Whether that profit is the yield of a renewable process or the harvest of a substrate now spent is not a further number the firm forgot to compute. It is a question about the firm’s position on a trajectory, and the trajectory is constituted by processes operating at horizons the quarter cannot see. A firm can post record efficiency in the same quarter it consumes the last of an engineering substrate twenty years in the building, and the efficiency figure and the consumption are not in tension within the measurement frame, because the frame has no term for the consumption. The number climbs. The substrate is gone. The instrument reads success.

The two incapacities are independent. The first would hold even if trajectories were legible, because relations are not magnitudes. The second would hold even if people were interchangeable units, because the trajectory question is about horizons rather than about persons. A defender of metrics who concedes the first sometimes tries to recover the second by promising better instruments, longer dashboards, leading indicators. The horizon-mismatch theorem of §24 closes that recovery in the proof register: the second incapacity is not a gap in current instruments but a structural feature of any regulator operating at a horizon shorter than its substrate.

§23.3 The Projection: Real Part, and a Component That Is Not a Number#

The relationship between what the metric reports and what the firm is can be stated precisely, and the precision matters because the imprecise version invites the dismissal that there is more to measure.

Let the firm’s condition at a moment be written as an ordered pair

where the first component collects the firm’s scalar magnitudes, revenue, output, headcount, the quantities that do admit a number, and the second component collects the firm’s constitutive process: its relations, its position on trajectories unfolding at horizons longer than the measurement window, its ongoing participation in the constitution of its own market. The measurement frame is a projection

The notation borrows the form of the complex number deliberately, and the borrowing must be qualified at once so that no more is claimed than holds. The complex number’s imaginary part is itself a real magnitude, a member of the same field, recoverable from the modulus and argument. The claim here is sharper and stranger than the complex case. The second component is not a scalar of the same kind as the first, waiting on a finer instrument that would read it off the same axis. It is a process, relational and constitutive, and the reason the projection $\pi$ discards it is not that $\pi$ has too few channels but that $\pi$ ranges over a single scalar axis and the second component does not lie on that axis at all. This does not make the second component structureless or unnameable. It can carry rich structure, with structure that has its own invariants, as §27 sets out when it names that structure cohomologically. What it cannot do is sit on the one-dimensional axis $\pi$ projects onto, or be recovered there by refining $\pi$. A projection onto the real axis discards the imaginary part because the imaginary part lies off the axis; in the complex case one rotates the axis and recovers it. Here no rotation brings the constitutive process onto the scalar axis, because the constitutive process is not displaced from the axis by an angle. It is of another order than the scalar. The analogy to the complex number holds in form, the metric reports one component of a two-component condition, and breaks in substance, the discarded component is not a scalar recoverable by re-aligning the scalar frame but a structured process that no choice of scalar frame renders as a scalar.

This is the precise sense in which the metrics measure something imaginary. They do not measure nothing; $\mathrm{Re}$ is real and the firm’s forty million is in it. They report $\mathrm{Re}$ and present it as $z$. The error is not in the reading of $\mathrm{Re}$, which is accurate, but in the identification of $\pi(z)$ with $z$, the treatment of the projection as the whole. An organisation steering by $\pi(z)$ steers by a real shadow of itself and takes the shadow for the body. The shadow can grow while the body fails, and nothing in the shadow registers the failure, because the failure lives in the component the projection does not range over.

The cohomological articulation of §27 names the structure of the discarded component for the cases where it admits topological description: the higher cohomology where local data fails to glue into global, the obstruction that has no representation in $H^0$. The naming is structural-analogue, in the sense §25.3 makes explicit, and it does the work of characterising the shape of $\mathrm{Im}$ without claiming $\mathrm{Im}$ is computable. What secures the existence of $\mathrm{Im}$, prior to any characterisation of its shape, is the horizon-mismatch theorem, to which the argument now turns.

§24. The Horizon-Mismatch Theorem#

§24.1 The Claim and Why It Carries the Paper#

The theorem establishes, in the proof register and without any appeal to topos theory, derived categories, or homotopy type theory, that a regulator operating at a temporal horizon shorter than its substrate cannot capture the substrate’s variety within its measurement frame. Its corollary, holding wherever the firm goes on persisting, locates the uncaptured variety in the absorbing terms rather than leaving it to vanish. The theorem is the floor on which the rest of Part V stands. If it holds, the existence of the discarded component $\mathrm{Im}$ of §23.3 is secured by proof rather than by phenomenological assertion, and the cohomological apparatus of §§27-28 is freed to characterise the component’s shape without bearing the burden of establishing that there is anything to characterise. If the floor gives way, the apparatus above it is decoration. For this reason the theorem is stated and defended at length, and the defence is built to survive a hostile reader who grants nothing that is not forced.

§24.2 Premises#

The theorem rests on three premises, each of which is either a structural identity or an elementary fact about finite measurement, and none of which is an empirical conjecture about organisations.

The first premise is Ashby’s law of requisite variety (1956, p. 207), in the form Beer adopted for organisational regulation (1985, p. 89): a regulator can absorb disturbance variety only up to the regulatory variety it commands. Variety here is the count of distinguishable states. The law is a structural identity rather than an empirical hypothesis. It says that if a disturbance can present more distinct states than the regulator can distinguish and respond to differentially, then some distinct disturbance states are mapped by the regulator to the same response, and control over the difference between those states is lost by definition. Ashby’s law is not in dispute among cyberneticians; it is the founding constraint of the field, and Beer built the VSM on it. The present argument takes it as given precisely because the tradition the argument addresses already takes it as given. A defender of metrics who rejects Ashby’s law has abandoned the cybernetic ground on which managerial control theory itself stands.

The second premise is the finite resolution of any measurement frame within a finite window. A measurement frame is whatever apparatus the regulator uses to distinguish states of the substrate: its instruments, reports, dashboards, sampled metrics. Any such frame has finitely many channels, each resolving finitely many discrete levels, sampled at a finite rate over a finite observation window. This is not a contingent limitation of present technology. It is a property of measurement as such: an instrument that made infinitely many infinitely fine distinctions in finite time would transmit unbounded information in bounded time, which Shannon’s source-coding theorem (1948) forbids for any channel of finite capacity. The premise is therefore secured by information theory, not assumed.

The third premise is the horizon relation. The regulator operates at a characteristic horizon $T_R$: the window over which it observes before it must act, the reporting period, the planning cycle, the sprint, the quarter. The substrate, the workforce, the codebase, the customer relationships, the market position, the accumulated engineering, operates at a characteristic horizon $T_\Sigma$ over which its own variety unfolds, and for the substrates that matter $T_\Sigma \gg T_R$. A codebase accumulates structure over years; a quarter cannot see it. A workforce’s tacit coordination builds over the careers of its members; a sprint cannot see it. The horizon relation $T_R \ll T_\Sigma$ is not assumed as universal; it is the condition under which the theorem bites, and §24.5 identifies exactly the organisational substrates that satisfy it.

§24.3 Statement and Proof#

Theorem (Horizon-Mismatch). Let $\Sigma$ be a substrate whose variety $V_\Sigma$ unfolds over characteristic horizon $T_\Sigma$. Let $R$ be a regulator coupled to $\Sigma$, operating at horizon $T_R$, with a measurement frame resolving at most $N(T_R)$ distinguishable states within any window of duration $T_R$. Let $V_M$ denote the variety the frame actually captures and $V_R$ the regulatory variety $R$ commands. Then

and whenever $V_\Sigma > N(T_R)$, which obtains generically when $T_R \ll T_\Sigma$, the frame strictly understates the substrate:

The difference $V_\Sigma - V_M$ is the variety the frame leaves unregistered. The corollary below establishes where it goes once the further condition of persistence is added.

Proof. The first inequality, $V_M \le V_R$, holds because the regulator cannot act differentially on distinctions its frame does not capture; the variety available for regulation is bounded above by the variety measured. The second, $V_R \le N(T_R)$, is the resolution bound of premise two: the frame distinguishes at most $N(T_R)$ states in its window, so the regulatory variety predicated on that frame cannot exceed $N(T_R)$. To bound $N(T_R)$ explicitly, let the frame have $k$ channels each resolving $r$ levels, sampled with period $\tau$. A single measurement act distinguishes at most $r^k$ states. Over a window $T_R$ the frame takes at most $T_R/\tau$ such acts, giving $N(T_R) \le r^{k}\cdot(T_R/\tau)$ when states are read independently and $N(T_R) \le r^{k(T_R/\tau)}$ when the temporal sequence itself carries state. The sampling period $\tau$ is the frame’s temporal resolution, the finest interval at which it can register a change in substrate-state; a later paper in this series carries this resolution as a separate quantity, the model-cycle, alongside the action-cycle $T_R$ and the substrate’s replenishment-cycle, where the same theorem is restated for the political economy of extraction. Either bound is finite, and by Shannon’s theorem neither can exceed $2^{\rho T_R}$ where $\rho$ is the channel capacity of the regulator’s information pathway. Thus $N(T_R)$ is finite and fixed by the regulator’s internal constitution.

Now $V_\Sigma$ is bounded by no regulator-internal quantity. It is set by the substrate’s own configuration space unfolding over $T_\Sigma$. When $T_\Sigma \gg T_R$, the substrate passes through state-distinctions that mature over durations longer than the regulator’s entire window, and these escape the frame because they are not complete within any single window the frame observes. A codebase’s architectural decay, a team’s erosion of trust, a customer base’s slow disaffection: each is a variety that exists only over $T_\Sigma$ and has no representative within $T_R$. Hence $V_\Sigma > N(T_R)$, and the strict inequality follows.

A regulator might try to escape the bound by integrating across many consecutive windows: sixteen quarters of reports, read together, to recover the decay that no single quarter shows. The escape relocates the regulator rather than defeating the bound. A regulator that integrates to a horizon of sixteen quarters has become a regulator operating at that longer horizon, and the same inequality applies to it at $T_\Sigma$, where it can no longer act on the quarter-scale disturbance it was built to handle. The dynamical regime changes with the horizon: the fast regulator corrects fast disturbance and cannot see slow decay; the slow regulator sees slow decay and cannot correct fast disturbance. No single regulator occupies both horizons at once, because the horizon is what fixes which disturbances it can resolve and which it can act upon. The integration manoeuvre is taken up again in §24.6, where the contradiction in demanding both horizons of one instrument is drawn out.

The theorem as stated establishes the measurement gap: the frame understates the substrate, and the understated variety $V_\Sigma - V_M$ goes unregistered. Where the unregistered variety goes is a further question the theorem does not settle on its own, and the corollary below settles it under one added condition.

Corollary (Absorption under Persistence). Suppose the coupled system persists: the firm continues to hold its essential variables in range, continues to ship, continues to meet the metric the regulator reports on. Ashby’s law, applied now to the whole coupled system rather than to the regulator alone, then forces a consequence. The disturbance variety $V_\Sigma$ has to be met by matching regulatory variety somewhere, or the essential variables would leave their range and persistence would not obtain. The regulator supplies at most $N(T_R)$ of the required variety. The remainder is supplied by the other elements coupled to the substrate, because there is nowhere else in the coupled system for it to come from: the people operating the substrate, the substrate’s own material reserve, or systems downstream to which disturbance is passed. The displacement follows as the contrapositive of Ashby’s law: if the firm persists and the regulator is short of requisite variety, some other element made up the difference. Read across many such firms and many such periods, the displacement hardens into a constraint, the variety a formal system cannot hold is conserved against the demand for coherence and reappears wherever the coupled system can still absorb it. The constraint, and the full accounting of where the absorbed variety lands and what its absorption costs the bodies and substrates that take it on, is the work of a later paper in this series. Here the corollary establishes only that the variety must land somewhere, and the cannibalisation case of §24.5 reads which element takes it on and at what price. $\quad\blacksquare$

§24.4 What Absorption under Persistence Means#

The corollary is the part of the argument that converts an epistemic result into a material one, and it is where the force gathers, so it is drawn out.

The unregistered variety has to be met somewhere once the firm goes on persisting. This is not rhetoric; it is Ashby’s law read as a contrapositive over the whole coupled system rather than over the regulator alone. If the substrate generates more distinguishable disturbance than the regulator can meet, and the firm holds together rather than coming apart, then by the same law that grounds the regulator’s own operation the difference is met by the other terms coupled to the substrate. There are three such terms, and the metabolic cost their absorption carries, traced through the bodies and substrates that bear it, is followed in detail by a later paper in this series; the present argument establishes the three sites and leaves the accounting of their cost to that work.

The first absorbing term is the people. When the measurement frame cannot represent the variety of what holds the work together, that variety is met by the workers as unmeasured labour: the evenings, the held-together relationships, the judgement exercised in the gap the metric cannot see. Sarah making weight is the corollary read on a body. The metric reads her resolution time; the variety her resolution time cannot represent is met by her skipped breaks and her decision to leave in four months.

The second absorbing term is the substrate itself, as accumulating degradation. When the variety is not met by people it is met by the thing, as wear that does not appear in the window in which it is generated. The codebase takes on the debt. The architecture accretes the special cases. The degradation is real, and it accrues from the moment it is generated, but it matures over $T_\Sigma$ and so stays out of view across every $T_R$ until the horizon at which it surfaces all at once, as the sudden incapacity Part IV described, the day a simple feature becomes impossible and no one can say why.

The third absorbing term is the externality, the variety discharged to coupled systems outside the regulator’s boundary: the other team that inherits the rate-limited load, the future quarter that inherits the deferred problem, the commons that inherits the discharged cost. The regulator’s boundary is drawn so that these lie outside it, which is precisely why its frame does not register them and its metric can improve as it discharges them.

The corollary has a consequence that the cannibalisation case of §24.5 turns on. Because the unmet variety is met by the absorbing terms rather than vanishing, an improvement in the measured component can be produced by moving variety onto those terms. Efficiency, read as a rising ratio of measured output to measured input, can be manufactured by taking variety off the measured input and placing it on the unmeasured absorbers. The metric does not simply miss the cost. The metric improves by incurring the cost, because the cost falls in the component the metric does not range over.

§24.5 Cannibalisation Read as Efficiency#

The theorem now meets the case you raised, and the case is not an illustration bolted on; it is what the corollary says, read at the horizon of a firm consuming its own substrate.

A firm may carry an engineering substrate built over twenty years: a codebase, yes, but more to the point a workforce whose tacit coordination, accumulated system knowledge, and mutual trust constitute the firm’s actual capacity to build. The substrate’s variety unfolds over $T_\Sigma$ of years to decades. The regulator, the quarterly operating review, the efficiency programme, the headcount ratio, operates at $T_R$ of a quarter. By the theorem, $V_\Sigma > N(T_R)$: the substrate carries variety the quarter cannot resolve.

The firm undertakes an efficiency drive, reducing the slack, the apparently unproductive time in which the substrate was in fact maintained: the senior engineers’ availability to answer questions, the documentation time, the redundancy of knowledge across people, the headroom that absorbed shocks. Measured productivity rises: the same output, fewer measured inputs, the ratio improves quarter on quarter. The frame reports efficiency, and the report is accurate as a reading of $\mathrm{Re}$, since output over headcount has indeed risen.

What the frame cannot report is that the slack it removed was the mechanism by which the substrate maintained itself, and that its removal has begun to consume twenty years of accumulated coordination as current output. The variety formerly met by the slack is, by the corollary, now met by the two remaining terms: the people, who do the same work with the maintenance margin gone, which the firm also reads as efficiency because their increased load appears only as steady output against reduced input; and the substrate, which begins to degrade as the maintenance that the slack performed ceases, the degradation maturing silently over $T_\Sigma$ and invisible across every quarterly $T_R$ in which it accrues.

The firm is on a slope toward incapacity and the instrument reads the slope as ascent. Careful measurement does not avert this; the structure of the theorem produces it. The efficiency gain and the substrate consumption are the same event viewed in the two components of §23.3: in $\mathrm{Re}$, output per head rising; in the discarded component, the trust and knowledge and maintenance-margin that constituted the firm’s capacity being burned as fuel. The two are not in tension within the frame because the frame projects onto $\mathrm{Re}$, where only the gain appears. The cannibalisation is legible only at $T_\Sigma$, by which horizon the substrate that would have been needed to recover has already been spent, and the workers who carried the accumulated knowledge have already, like Sarah, begun to leave, taking the unmeasured component out of the door with them.

A firm in this position does not discover its error by measuring better. Every measurement it can take is a reading of $\mathrm{Re}$ within $T_R$, and within that frame the error does not exist. The error is the identification of $\pi(z)$ with $z$, sustained over the horizon across which the discarded component matures into incapacity. By the time the incapacity surfaces into the measured component, as missed deliveries, as the simple feature that has become impossible, as the exodus of the people who knew why things were done as they were, the substrate that would have been required to respond is the substrate that was consumed to produce the efficiency that the response would now have to reverse. The slope is not observed until it is the floor.

§24.6 Why Better Instruments Cannot Close the Gap#

A defender concedes the cannibalisation case and proposes the obvious remedy: measure the substrate. Add leading indicators, track technical debt, survey engagement, instrument the trust. The remedy does not work, and the theorem says why, so the breakdown is structural rather than a matter of insufficient diligence.

Every proposed instrument is a measurement frame, and every measurement frame is subject to the theorem. An instrument that tracks technical debt operates at some horizon $T_{R’}$ and resolves some $N(T_{R’})$ states; if the debt it must track matures over $T_\Sigma \gg T_{R’}$, the new instrument stands in the same relation to the substrate as the old one and understates it by the same structure. One can make $T_{R’}$ longer, but a regulator whose horizon approaches $T_\Sigma$ has, by the third premise, ceased to be a regulator operating faster than its substrate; it now acts on year-scale information and cannot regulate quarter-scale disturbance, which is the opposite breakdown. The horizons cannot both be short enough to regulate and long enough to see the substrate, because they are the same horizon. This is the precise content of the claim that the incalculability is structural and not technological: there is no instrument because the demand on the instrument is internally contradictory.

And beneath the horizon argument lies the deeper one of §23.1. Even an instrument operating at $T_\Sigma$ would still be an instrument deployed by a constituent of the process, and the component it is asked to capture, the relational and constitutive process of §23.2, does not lie on the scalar axis any frame projects onto. The trust between two engineers is not a low-frequency signal awaiting a slow enough sensor. It is not a scalar signal at all. The horizon-mismatch theorem shows the gap cannot be closed from within the measured component even for the variety that is in principle scalar; the projection argument of §23.3 shows that the residual does not lie on the scalar axis in the first place. The two results bound the firm from both sides. What is scalar cannot be captured at the horizon that matters; what matters at that horizon is not scalar.

§25. Strategic Deployment of Mathematical Sophistication#

§25.1 The Vulnerability the Apparatus Addresses#

The chiasmic ontology developed in Part III, grounded phenomenologically and demonstrated empirically through the Tuesday-morning narrative of Part IV, faces a vulnerability of position. It appears insufficiently rigorous compared to the metric-driven, algorithmic approaches that dominate contemporary management discourse. When executives demand KPIs, dashboards, and metrics, phenomenological arguments about reversibility and intertwining may seem insubstantial in the institutional register where decisions are made.

The vulnerability is more than rhetorical. Power (1997, p. 97) documents in The Audit Society that organisations increasingly require quantification and measurement as a condition of legitimacy. Meyer and Rowan (1977, p. 340) identify the structural pressure as institutional isomorphism: organisations incorporate the practices and procedures defined by prevailing rationalised concepts of organisational work. An argument that abandons the institutional register surrenders the field; an argument that engages the register must do so with apparatus the register recognises.

The chiasmic model requires protection from reductive demands. The protection comes not through better arguments alone but through mathematical demonstration of the structural limits of formalisation itself. The paper deploys advanced mathematical apparatus for diagnostic rather than representational purposes: to articulate the structural features any adequate formalisation would need to capture, and to establish in the proof register that the simplest formalisations underlying KPIs and dashboards cannot meet the conditions Ashby’s law imposes.

§25.2 Mathematics as Diagnostic Apparatus#

The strategy uses mathematics diagnostically rather than representationally. Representational mathematics attempts to model and capture the phenomenon. Diagnostic mathematics reveals where and why such capture breaks down. A physician uses diagnostic tools to understand a patient’s limits without becoming the patient, and the present argument uses mathematical formalism to mark its own boundaries without claiming to formalise the chiasmic phenomena it discusses.

The approach has direct precedent in mathematical logic. Gödel (1931) used arithmetic to prove arithmetic’s incompleteness. Turing (1936) used computation to prove computation’s limits. Arrow (1951) used preference logic to prove preference aggregation’s impossibility. The present argument follows the same tradition: using formal apparatus to establish the structural limits of formal apparatus when applied to organisational phenomena.

The strategy constitutes an immanent critique. The argument uses the opposition’s preferred language (mathematical rigour) to demonstrate the inherent limits of that language when applied to organisational reality. The argument is not an external critique from a softer discipline that management science can dismiss as lacking rigour. The argument is mathematics demonstrating, with mathematical precision, that formalisation pursued with rigour reveals its own structural limits. The master’s tools cannot dismantle the master’s house, in Audre Lorde’s (1979) formulation, but they can prove the house was never built on the foundation its architects believed it rested on.

MacKenzie (2006, p. 89) captures the feature the strategy exploits: “mathematical models gain authority not from accuracy but from sophistication.” The argument leverages the authority strategically, using mathematical rigour to protect phenomenological insight from the dismissive moves the institutional register makes against work in softer disciplines.

§25.3 Two Registers#

The mathematical apparatus this paper deploys operates in two distinct registers. The first register supplies proof. The second register supplies structural analogue. Both registers do work, but the work each does is different, and a reviewer or consultant encountering the apparatus deserves to know which is which.

In the proof register, mathematical claims are demonstrated by deduction from premises that hold for any system meeting the relevant conditions. The horizon-mismatch theorem of §24 above is the paradigm case. Ashby’s law and finite-cardinality arguments yield the variety-absorption inequality without requiring any commitment to what specifically the substrate is. The Novikov-Boone result on undecidability of the word problem in finitely presented groups is another proof-register result, applied in §24 below.

In the structural-analogue register, mathematical apparatus is deployed as articulation. Sheaf cohomology articulates the phenomenon of local solutions that do not extend to a global one. Derived categories articulate the phenomenon of transformation producing irreducible remainder. Higher inductive types articulate the phenomenon of identity emerging through paths rather than through pre-given containers. In each case the mathematics names what the phenomenology has identified. The naming is rigorous in the formal sense, with the mathematical object’s properties matching the structural features the phenomenon exhibits. It is not rigorous in the sense that the mathematical object has been constructed from the phenomenon through a justified embedding.

The distinction matters for three reasons. First, the cascade argument depends on the proof register at its foundation. If the floor is hand-waved, the cascade collapses. The horizon-mismatch theorem secures the floor; from the floor the cascade extends upward through structural-analogue articulations. Second, the distinction matters because reviewers and consultants with mathematical training will probe the apparatus. A paper that claims more than it can defend invites dismissal of even the parts it can defend. Better to mark the structural analogue explicitly and let it do the work it can than to dress it as proof and have the dressing torn off. Third, the distinction matters because the paper’s commitment is to mathematics as reasoning rather than as modelling. Reasoning includes both proof and structural articulation. Modelling is precisely what the paper rejects. The structural-analogue register does not propose that organisations are topoi or are chain complexes; it proposes that organisations exhibit features that topos-theoretic and homological language can name with precision.

§26. The Word Problem and Algorithmic Undecidability#

The second proof-register result applied in the cascade is the algorithmic-undecidability result of Novikov (1955) and Boone (1958).

Theorem (Novikov-Boone). There exist finitely presented groups whose word problems are algorithmically undecidable. That is, there exist groups $G$ with finite presentation $\langle S \mid R \rangle$ such that no algorithm decides, for arbitrary words $w_1, w_2$ in $S$, whether $w_1 = w_2$ in $G$.

The result is standard combinatorial group theory from the mid-twentieth century. Its application to organisational categories proceeds in two steps. The first step is a proof-register result. Consider an organisational category $\mathcal{C}$ whose objects are organisational states and whose morphisms are the available transitions between states. The nerve $N(\mathcal{C})$ is the simplicial set whose $n$-simplices are sequences of $n$ composable morphisms. The geometric realisation $|N(\mathcal{C})|$ is a topological space, and its fundamental group $\pi_1(|N(\mathcal{C})|)$ admits a presentation derived from the morphism structure. When the morphism structure of $\mathcal{C}$ is rich enough to encode the generators and relations of a Novikov-Boone group, $\pi_1(|N(\mathcal{C})|)$ contains a subgroup with unsolvable word problem, and algorithmic equivalence-checking between sequences of morphisms in $\mathcal{C}$ is undecidable. This step follows directly from the embedding theorems of combinatorial group theory.

The second step is a structural-analogue claim. Organisations of even moderate complexity meet the conditions for such encoding. The justification is not a constructive proof of the embedding for any specific organisation. The justification is structural: organisational categories combine the cardinality (large state space), the composition structure (sequential operations), and the relational density (non-commuting transformations) that the standard embedding theorems require. The structural-analogue claim is that the conditions are met by organisations as a class. Constructive proof for any particular organisation would require fixing the category $\mathcal{C}$ explicitly, which the chiasmic argument of Part III shows is itself not possible without observer-imposition.

The consequence is that strategic-planning procedures requiring algorithmic equivalence-checking, including procedures that ask whether two sequences of organisational operations lead to identical states, are undecidable in the proof-register sense for any sufficiently rich organisational category, and undecidable in the structural-analogue sense for organisations as encountered.

The cascade does not require this result to operate. The horizon-mismatch theorem of §24 secures the foundation. The Novikov-Boone result articulates an additional dimension of breakdown: even if the horizon mismatch were somehow overcome, algorithmic procedures over the resulting structure would still break down by undecidability. The two modes of breakdown are independent.

§27. Cohomological Obstruction Theory: A Structural Articulation#

The apparatus that follows operates in the structural-analogue register. Sheaf cohomology is deployed to articulate, with the formal precision of standard algebraic geometry (Grothendieck, 1957; Hartshorne, 1977), the phenomenon that the chiasmic argument has identified: organisational intelligence resides in the obstruction to uniform extension of local solutions across the organisation’s relational field. The intuitive face of this apparatus was given in 3.2, where the impossibility of flattening a curved surface onto a plane without distortion stood as the geometric form of the map-territory relation. The patching obstruction below is that impossibility stated exactly: where the flat map cannot represent the globe without relocating distortion, the local sections cannot be glued into a global section without an obstruction that the higher cohomology measures.

Principle (Intelligence as Obstruction). Organisational intelligence appears where uniform solutions do not extend. In the language of sheaf cohomology, the higher cohomology groups $H^n(\mathcal{F})$ of a sheaf $\mathcal{F}$ over a site measure exactly such obstruction to extension. The structural analogue is direct: where $H^0$ captures globally-extensible sections (uniformly applicable solutions), $H^n$ for $n \geq 1$ captures obstructions to global extension. The principle states that organisational intelligence corresponds to the higher-cohomology content rather than the $H^0$ content.

The articulation is structural-analogue rather than constructive embedding. The correspondence between organisational structure and the sheaf-theoretic object is not a constructive embedding because the chiasmic argument shows constructing the embedding would require committing to a specific sheaf on a specific site, a commitment that cannot be made from any observer position. What the principle captures is rigorous in the structural sense: the formal features of sheaf cohomology (sequence exactness, the local-to-global spectral sequence, the connecting homomorphisms relating $H^n$ to $H^{n-1}$) exhibit the structural features the phenomenology identifies (irreducible local-global gap, propagation of obstruction through the relational field, dependence of higher obstruction on lower).

The implication for organisational practice is direct. Measurement instruments designed to capture $H^0$ content (uniformly applicable metrics, standardised KPIs, dashboards with fixed dimensional resolution) capture precisely the content that is not the intelligence. They project onto the trivial cohomology and discard the higher cohomology where the intelligence resides. When management consultants propose breaking down silos or imposing departmental KPIs, they propose operations that destroy higher-cohomology content while preserving $H^0$ content. The intelligence is then no longer present in any measurable register.

The Leray spectral sequence provides further structural articulation. The standard result $E_2^{p,q} = H^p(X, \mathcal{H}^q(\mathcal{F})) \Rightarrow H^{p+q}(I^\bullet)$ articulates the entanglement of cohomological levels: organisational intelligence cannot be computed layer by layer because the differentials $d_r$ entangle the levels essentially. McCleary (2001) provides the standard reference. The structural-analogue claim is that organisational reform programmes which proceed level by level (first restructure leadership, then departments, then teams) do not register the entanglement that the spectral-sequence structure articulates.

§28. Derived Categories and the Necessity of Remainder#

The apparatus of derived categories (Verdier, 1996; Gelfand and Manin, 2003) is deployed here in the structural-analogue register to articulate the phenomenon of transformation producing irreducible remainder.

Principle (Necessary Remainder). Every organisational transformation produces a non-trivial cone: a residual structure that cannot be transferred into either the prior state or the subsequent state without loss. In the language of derived categories, the distinguished triangle

captures the structure: $C^\bullet$ is the cone of the morphism $A^\bullet \to B^\bullet$, and it is non-trivial whenever the morphism is not an isomorphism. The principle states that organisational transformations correspond to non-isomorphism morphisms, and the cone $C^\bullet$ corresponds to the irreducible remainder.

The articulation is structural-analogue. The cone construction is well-defined, the non-triviality condition has precise meaning, and the propagation of remainder through composed transformations follows the formal rules of the derived category. The analogue is not constructive in the sense of producing $A^\bullet$, $B^\bullet$, $C^\bullet$ as complexes for any specific organisation. The chiasmic argument of Part III shows why such construction is not available from any observer position.

The phenomenological content the principle articulates is direct. Transformation programmes that promise “seamless transitions,” “frictionless adoption,” or “lossless migration” promise the impossible. The derived-category articulation does not prove this directly; it names with formal precision the structural feature that the phenomenology identifies. The Fourier-Mukai apparatus (Huybrechts, 2006) extends the same articulation to knowledge transfer between organisations: every transfer of practice involves transformation, and transformation produces non-trivial cone, and the cone is precisely the part of the practice that does not transfer.

§29. Higher Inductive Types and Chiasmic Boundaries#

The apparatus of homotopy type theory (Univalent Foundations Program, 2013) is deployed here in the structural-analogue register to articulate the chiasmic identity that Merleau-Ponty’s late ontology identifies: the touching-touched, inside-outside, observer-observed structure of §9.

The articulation proceeds by treating organisational boundaries as higher inductive types. A higher inductive type is a type defined not only by constructors that build elements (as in ordinary inductive types) but also by constructors that build paths between elements. The constructor

posits, as a primitive of the type, that inside and outside are connected by a path. The path is not derived from prior identity of the two; it is constitutive of the type itself.

The structural analogue captures the chiasmic feature with precision. Where ordinary set-theoretic and category-theoretic formalisms treat boundaries as separations between pre-existing entities, the HoTT formalism treats them as paths constitutive of the entities they connect. The Univalence Axiom (Voevodsky, 2006) extends the articulation: identity and equivalence are themselves equivalent, so organisations that operate equivalently are operationally identical. Claims about “culture” or “values” that have no operational manifestation correspond to types with no path-constructors and are therefore empty in the formal sense.

The articulation is structural-analogue rather than constructive embedding. HoTT has precise rules for forming higher inductive types, the path-constructor formalism is well-developed, and the Univalence Axiom has its standard consequences. The articulation does not construct the type $\text{OrgBoundary}$ for any specific organisation, because the chiasmic argument shows that observer-imposed construction would itself violate the chiasm it aims to formalise. The articulation captures the structural feature without claiming to instantiate it.

§30. The Cascade Footing#

With the proof-register foundation established by the horizon-mismatch theorem of §24 and the Novikov-Boone result of §26, and with the structural-analogue articulations of §§27-29 explicitly marked as such, the cascade argument proceeds on solid ground. The cascade does not claim that ∞-topoi have been proved insufficient for organisations in the strict sense of constructive impossibility. The cascade claims four things, distinguished by register.

In the proof register: for any regulator operating at horizon shorter than its substrate, the substrate’s variety exceeds the regulator’s measurement frame (the horizon-mismatch theorem of §24); and, under persistence, the unregistered variety is met by the absorbing terms (the corollary of §24). In combinatorial group theory: the word problem is undecidable for finitely presented groups of sufficient richness (Novikov-Boone, §26), a result proved about groups, not about organisations.

In the structural-analogue register: the claim that organisational state-categories encode such groups, and so inherit the undecidability, is an application argued structurally rather than by constructive embedding (§26); and organisational intelligence exhibits the formal features that sheaf cohomology, derived categories, and homotopy type theory articulate, namely local-global obstruction (§27), transformation remainder (§28), and path-constitutive identity (§29).

The consequence: mathematical formalisations operating below the level of §§27-29’s articulations cannot capture the features the articulations name. Elementary formalisms (arithmetic in KPIs, basic real analysis in OKRs, linear algebra in balanced scorecards) operate well below this level. The cascade therefore concludes that the inadequacy of elementary formalisms is structural and not a contingent shortfall of present technique, for the reasons the proof-register results establish and the structural-analogue articulations clarify.

The argument is one theorem about variety-absorption with its persistence corollary, one proof-register theorem about algorithmic undecidability with a structural-analogue application to organisations, and three structural-analogue articulations of features that any mathematical formalisation would need to capture. The cascade rests on what is proved at its base and on what is articulated at its higher levels, with the join between them marked at each step. The consultant or reviewer who attempts to attack the apparatus must engage either Ashby’s law (structurally the regulatory premise of cybernetics itself), the Novikov-Boone embedding theorems (standard mid-twentieth-century combinatorial group theory), or the structural-analogue articulations (explicitly marked and not overclaimed). The attack surface is narrow.

§31. Creating Protected Space#

The mathematical apparatus creates protected space for phenomenological investigation by making reductive demands provably misdirected. Any critique that engages the apparatus must engage the mathematics (requiring years of study to assess), ignore the apparatus (appearing anti-intellectual within the institutional register the critique would have to operate in), or dismiss the apparatus (conceding the field). The trilemma is exhaustive: there is no fourth move available to a critique that wishes to remain within the rationalist register that motivated the demand for metrics in the first place.

Inside the protected space, phenomenological investigation can proceed without defending itself against reductive interruption. The chiasmic intertwining of organisation and environment can be examined for what it is. The reversibility and flesh structures of Part III can be developed. The temporal depth across multiple horizons that Part IV demonstrated empirically can be characterised without being reduced to scalar metrics. The organisational becoming that the genealogy from Beer through second-order cybernetics to chiasmic ontology has traced can be the substance of further investigation without each step being challenged on its measurement compatibility.

As Ashby (1956, p. 207) formulated the regulatory principle that the horizon-mismatch theorem of §24 sharpens: “only variety can destroy variety.” The mathematical variety of the apparatus this paper deploys destroys the reductive simplicity of the institutional demand for metrics. The argument is not that organisations should not be reduced to metrics. The argument is that the horizon-mismatch theorem and the Novikov-Boone result in the proof register, supplemented by the structural-analogue articulations of §§27-29, together establish the lower bound: any regulator operating at horizon shorter than its substrate cannot capture the substrate’s variety in any measurement frame, and algorithmic equivalence-checking over sufficiently rich organisational categories is undecidable. The structural-analogue apparatus articulates what any mathematical formalisation would need to capture and demonstrates that the most expressive available formalisms operate in the structural-analogue register only. The elementary formalisms underlying KPIs and dashboards fall short by both registers.

Part VI: Strategic Adjustments and Empirical Grounding#

§32. Methodological Declaration#

This paper employs a methodological strategy that requires explicit acknowledgement. The paper uses mathematical impossibility results to establish rigorous boundaries of formalisation in organisational theory rather than to model organisations. The approach follows a distinguished tradition in mathematics itself, where impossibility theorems have proven as valuable as constructive results. Lakatos (1976, p. 9) captures the position in Proofs and Refutations: “mathematics does not grow through a monotonous increase of the number of indubitably established theorems but through the incessant improvement of guesses by speculation and criticism.” The argument of this paper represents one such criticism, using mathematical rigour to demonstrate the limits of mathematical application to organisational phenomena.

The method has four steps. First, deploy the advanced mathematics available in the contemporary field. Second, articulate the structural features any formalisation would need to capture (marking the apparatus as structural analogue rather than constructive embedding, per §25.3). Third, establish in the proof register that the elementary formalisms underlying institutional metrics cannot meet the conditions Ashby’s law imposes (per §24). Fourth, create protected space for non-reductive organisational understanding (per §31).

The approach is rigorous in its application of mathematics to its own limits, which is what Gödel (1931) did for arithmetic and what the present argument attempts for organisational theory. A serious critique might argue that organisations do not necessarily instantiate the mathematical properties the apparatus invokes, and that the paper has not proven organisations are topoi. The response is threefold. First, the paper does not claim organisations are topoi; it claims that if mathematics could capture organisations, topos theory would be among the most promising frameworks given its capacity to handle multiple truth values (organisational ambiguity), internal logic (organisational reasoning), and sheaf conditions (local-global properties), and the structural-analogue framing of §§27-29 makes this explicit. Second, if organisations do not instantiate advanced mathematical properties, they certainly do not instantiate the elementary properties assumed by KPIs (scalar addition), percentages (ratio measurement), averages (statistical aggregation), and rankings (total ordering). Third, the burden of proof lies with those claiming mathematical capture is possible. The paper has shown the most advanced attempts fall short at the level of constructive embedding while remaining usable as articulation. Defenders of elementary metrics must explain why arithmetic succeeds where topos theory operates only at the structural-analogue register.

§33. Empirical Grounding: Three Documented Cases#

The mathematical argument of Part V claims that the horizon-mismatch theorem of §24 and the structural-analogue articulations of §§27-29 hold across organisational contexts. Three publicly documented cases instantiate the argument: a digital-transformation lawsuit between a Fortune 500 company and a global consultancy, a healthcare scandal that prompted the largest NHS public inquiry of the post-war period, and the only published empirical test of the architecture-organisation mirroring relationship in the software industry. The cases are deliberately chosen across contexts (commercial technology, public healthcare, open-source software) and methodologies (court documents, statutory inquiry, peer-reviewed quantitative study) to test whether the argument holds invariantly across them.

§33.1 The Hertz-Accenture Digital Transformation (2016-2019)#

In January 2015, Hertz Corporation hired a new Chief Information Officer and committed to a comprehensive digital transformation of its enterprise IT environment. At that time, Hertz operated approximately 1,800 IT systems, six database vendors, and thirty separate rental-processing systems, according to documentation produced in subsequent litigation and a MuleSoft conference presentation of 2018 (Dolfing, 2022). In March 2016, Hertz engaged Accenture under a consulting services agreement that had been in place since 2004. Accenture recommended a five-layer digital platform architecture: a front-end layer using the Angular JavaScript framework, a content-management layer using Adobe Experience Manager, a microservices layer, an enterprise-services bus, and an integration layer connecting to legacy back-end systems. The recommendation and its consequences are set out in Hertz Corporation v. Accenture LLP, the complaint filed in the United States District Court for the Southern District of New York on 19 April 2019, with the amended complaint filed 20 June 2019 (Case No. 1:19-cv-03508).

The project was governed throughout by the standard agile-at-scale measurement apparatus: story-point velocity, sprint completion rates, integration milestones, user-acceptance test pass rates. By the standard metrics, the project was tracking acceptably through 2017. In September 2017, Hertz signed the statement of work for the first phase, with go-live scheduled for December 2017. The go-live was postponed one month later. By late 2017, Hertz had removed Accenture from the project and engaged IBM to complete the work. Hertz never deployed the system Accenture had been paid approximately USD 32 million to build.

The Hertz complaint alleges, in Count 2, that Accenture engaged in deceptive and unfair trade practices, specifically that Accenture elected, without Hertz’s authorisation, to design the system in ways that locked Hertz into ongoing dependency on Accenture for maintenance; staffed the engagement with personnel lacking the experience Accenture had represented; and delivered code that did not satisfy basic requirements for security, accessibility, or performance. The complaint reproduces internal Accenture communications, obtained in discovery, in which Accenture personnel acknowledged the project was coming apart while public reporting through the agile-velocity framework remained on track. Approximately 2,000,000 documents were reviewed in discovery; approximately 500,000 were placed in evidence (Dolfing, 2022).

The case instantiates three of the paper’s claims with documentary precision. First, the horizon-mismatch theorem of §24: Hertz’s measurement frame operated at sprint-level horizon (two-week cycles, story-point velocity) while the substrate the project was reorganising operated at the horizon of Hertz’s 1,800 legacy systems, six database vendors, and thirty rental-processing systems, accumulated across two decades. The regulator’s variety was bounded by what the agile measurement frame could resolve; the substrate variety exceeded that frame by orders of magnitude that the metrics could not represent. Second, the N-th Order State of §8.1: the four features are documented in the Hertz pleadings. Internal communications show recursive acknowledgement that the framework’s reporting did not capture project reality. Third, the three-mechanism reform foreclosure of §8.3: every reform proposal that arose within the project was processed by absorption (rephrased into agile-compatible velocity targets), attenuation (engineers’ concerns dismissed as not following the methodology), or decomposition (hybrid proposals split, with the agile-compatible component implemented and the structural critique discarded).

The case ended through external shock rather than internal reform. The shock was provided by Hertz’s new Chief Marketing Officer, hired in late 2017, who appears from the public record to have brought sufficient organisational standing to override the agile-velocity reporting framework. Within months of her arrival, Accenture had been removed. The reform required someone with institutional authority operating outside the project’s measurement frame, exactly the pattern the impossibility result of §8.3 predicts.

The Hertz-Accenture case demonstrates that the three-mechanism foreclosure is not peculiar to public-sector organisations or planned economies. It operates with full force in commercial technology projects backed by Fortune 500 capital and the largest consultancy firms in the world.

§33.2 Mid Staffordshire NHS Foundation Trust (2005-2009)#

Between January 2005 and March 2009, the Mid Staffordshire NHS Foundation Trust operated under intensified performance-management arrangements designed to secure Foundation Trust status, which the Trust obtained in February 2008. The arrangements required the Trust to meet specific operational targets: maximum accident-and-emergency waiting times of four hours, financial-balance requirements, and a range of performance indicators tracked through the Healthcare Commission’s annual health-check ratings.

The Healthcare Commission investigation, published March 2009, found that the Trust’s mortality rate had been substantially elevated throughout the period under review. Press reports during the public inquiry cited figures of between 400 and 1,200 excess deaths. The Francis Report itself, the Mid Staffordshire NHS Foundation Trust Public Inquiry chaired by Robert Francis QC, three volumes totalling 1,782 pages published 6 February 2013 (Francis, 2013), concluded that it would be unsafe to attribute a specific number of avoidable deaths from the available statistical data alone, focusing instead on documenting the conditions of care and the institutional mechanisms that permitted those conditions to persist.

The Francis Report’s findings instantiate the three-mechanism reform foreclosure of §8.3 with documentary specificity that no constructed case could match. The Report reviewed evidence from more than 250 witnesses and approximately one million pages of documentary material. It found that nursing staff, doctors, patients, families, and external regulators had all attempted at various points to raise concerns about the standard of care at Stafford Hospital. The Report documents how those concerns were processed.

The first mechanism, absorption, appears throughout. Concerns articulated in the Trust’s own performance-management vocabulary were processed as routine quality-improvement matters. Specific examples in the Report include incident reports filed by nursing staff that were classified as low-severity and absorbed into the Trust’s monthly clinical-governance dashboard, where they appeared alongside routine items and lost institutional visibility (Francis, 2013, Volume 1, Chapter 8).

The second mechanism, attenuation, dominates the Report’s findings. Nurses and patients who articulated concerns in non-target-compatible language (clinical judgement, dignity, basic patient care) lacked the institutional standing required for uptake. The Report documents that the Trust’s senior management consistently prioritised attainment of the four-hour A&E target, financial balance, and Foundation Trust assessment criteria over the clinical concerns raised by ward-level staff. Concerns that did not translate into the target framework were dismissed as anecdotal, isolated, or attributable to individual nursing-staff inadequacy rather than systemic conditions (Francis, 2013, Volume 2, Chapter 18).

The third mechanism, decomposition, operates on the families’ campaigning. Julie Bailey’s Cure the NHS campaign, founded after the death of her mother at Stafford Hospital in 2007, articulated concerns in a hybrid register: specific clinical harms (target-compatible language) and a broader claim that the institutional culture was producing those harms (non-target-compatible language). The Trust’s response, documented in the Report, addressed the specific clinical harms through individual incident investigations while declining to engage with the systemic claim. Local managers handled the specific deaths; the systemic critique was not taken up. The hybrid was decomposed exactly as the theorem of §8.3 predicts.

The Mid Staffordshire case also instantiates the homeostatic triad of §8.2. Credentialism operated to attenuate ward nurses’ concerns relative to management’s metric-based reports, because management’s reports met the Healthcare Commission’s data requirements while nursing concerns did not. Performance amplification followed the predicted pattern: as the gap between target attainment and clinical reality widened, the Trust intensified its target-focused performance management rather than questioning the targets themselves, and the Foundation Trust assessment was pursued through the intensification. Source isolation marginalised staff who persistently raised concerns through transfers, performance-improvement procedures, and the cultural classification “not a team player.” The Report names specific individuals subjected to these mechanisms (Francis, 2013, Volume 3, Chapter 22).

This state ended through external shock that arrived in two stages: first, the Healthcare Commission’s independent investigation of 2008-2009, conducted without the Trust’s authorisation; second, the public inquiries chaired by Robert Francis. Both shocks required institutional authority external to the Trust’s measurement-and-incentive frame. The first Francis Report (2010) was conducted under the Inquiries Act 2005 by an authority the Trust could not absorb, attenuate, or decompose. The reform that followed was substantial: the Trust was dissolved in 2014, with services transferred to neighbouring trusts. The pattern matches the prediction of §8.3: internal reform was foreclosed by the three mechanisms, and change required perturbation of sufficient magnitude to exceed the regulatory frame’s absorption capacity.

The Mid Staffordshire case demonstrates that the state this paper diagnoses operates in public-sector contexts with statutory accountability requirements, professional regulation, and decades of established quality-improvement practice. The professionalism of clinical staff and the existence of statutory regulators proved insufficient to defeat the three-mechanism foreclosure once the target-attainment regime had stabilised.

§33.3 Conway’s Chiasm in Software Architecture#

Section 11 of this paper introduced Conway’s Chiasm: the proposition that organisational structure and product architecture co-constitute one another, rather than one being the cause and the other the effect. The proposition is testable. MacCormack, Baldwin, and Rusnak (2012) provide the test.

The MacCormack-Baldwin-Rusnak study appeared in Research Policy volume 41, issue 8, pages 1309-1324 (October 2012). The paper takes advantage of a natural experiment in the software industry: identical functional requirements being implemented by organisational forms at opposite ends of the coupling spectrum. At one end, commercial software firms with tightly-coupled organisational structure (shared employment, hierarchical authority, co-located development). At the other end, open-source communities with loosely-coupled organisational structure (distributed contributors, no shared employer, asynchronous communication). The mirroring hypothesis predicts that the products developed by loosely-coupled organisations will exhibit more modular architecture than the products developed by tightly-coupled organisations.

MacCormack and colleagues construct matched pairs of software products that fulfil the same function but were developed under the two organisational forms: matched pairs in word processing, operating-system kernels, database management, and other categories. They measure product modularity through Design Structure Matrix analysis, which quantifies the level of coupling between a product’s components and the propagation potential of design changes across components, with the underlying method developed in Baldwin and Clark (2000).

The empirical finding (MacCormack, Baldwin, and Rusnak, 2012, p. 1318): in every matched pair examined, the product developed by the loosely-coupled organisation exhibited significantly higher modularity than the product developed by the tightly-coupled organisation. The magnitude of the difference reached a factor of six in propagation cost between component pairs. The mirroring hypothesis was supported in every pair, without exception.

The finding instantiates Conway’s Chiasm rather than Conway’s Law alone. Conway’s Law (1968) is a one-way claim: organisation structure determines product architecture. The MacCormack-Baldwin-Rusnak finding is consistent with Conway’s Law and cannot establish its directionality from cross-sectional data. What the finding does establish is mirroring: organisational coupling and product coupling are statistically inseparable. Subsequent work by Colfer and Baldwin (2016) reviews the broader literature and finds that the mirroring relationship persists across context, methodology, and time period. Where mirroring breaks down, it does so through organisations consciously decoupling their structure from the product they are building, in order to achieve the modularity the structure-imposed coupling would otherwise prevent (Sosa, Eppinger, and Rowles, 2004). The conscious decoupling itself confirms the underlying mirroring relationship; it is the mirroring’s mechanism being overridden by deliberate organisational design.

The case supports Conway’s Chiasm in the form stated in §11: organisation and architecture co-constitute one another through ongoing mutual specification, with neither term temporally prior. The universal support for mirroring in the matched-pair study and the documented difficulty of decoupling architecture from organisation in practice are inconsistent with treating organisation and architecture as separable entities that can be measured, optimised, or transformed independently. The proposition that an organisation can transform its architecture while holding its organisational structure constant, or restructure while leaving the architecture in place, is exactly the proposition the mirroring literature shows does not hold empirically across more than fifty years of software-industry observation.

The §31.3 case differs in evidentiary register from §§31.1 and 31.2. The Hertz and Mid Staffordshire cases are documentary: court filings and statutory inquiry reports. The MacCormack-Baldwin-Rusnak case is quantitative empirical work in a peer-reviewed journal. The convergence of the three cases across three widely different evidentiary registers is itself an empirical argument: the mechanisms the paper identifies operate across contexts that share no domain features beyond being organisations.

§33.4 Summary#

Three cases, three registers, one underlying pattern. The Hertz-Accenture litigation demonstrates the three-mechanism reform foreclosure operating in commercial digital transformation under conditions of maximum measurement intensity. The Mid Staffordshire NHS public inquiry demonstrates the same three mechanisms operating in public healthcare under conditions of statutory regulation and professional accountability. The MacCormack-Baldwin-Rusnak study demonstrates the chiasmic co-constitution of organisation and product architecture as a quantitative empirical regularity across the software industry. The convergence across the three cases is the empirical argument. The mechanisms identified in Parts II through V and formalised in §§8 and 23-27 are not artefacts of one institutional context; they appear, with the structural features the paper specifies, across contexts that share no domain features beyond being organisations of human action coordinated by measurement and incentive.

§34. Constructive Elements: What Chiasmic Understanding Enables#

The impossibility of calculation does not entail impossibility of intelligent action. It calls for a shift from a philosophy of control to one of cultivation and navigation. The following pairs map reductive approaches to chiasmic alternatives.

Where the reductive approach measures boundaries, the chiasmic alternative recognises transformation zones: the shift moves from lines to gradients, with the practical consequence that the organisation can work with permeability rather than enforce walls. Where the reductive approach quantifies intelligence, the chiasmic alternative senses obstruction patterns: the shift moves from metrics to topology, preserving the conditions that enable adaptation. Where the reductive approach calculates change, the chiasmic alternative anticipates remainder: the shift moves from planning to expecting debt, preparing for what cannot be eliminated. Where the reductive approach optimises locally, the chiasmic alternative respects entanglement: the shift moves from silos to fields, recognising that changes reverberate across the entire structure. Where the reductive approach controls variance, the chiasmic alternative maintains requisite variety: the shift moves from reduction to preservation, with variety understood as the condition of adaptation rather than its enemy. Where the reductive approach defines objectives, the chiasmic alternative enacts purposes: the shift moves from targets to emergence, with objectives discovered through pursuit rather than fixed in advance.

The pairs translate into concrete leadership practices. Instead of enforcing strict departmental silos with defined handoff procedures, the organisation can create formal boundary-spanning roles whose responsibility is the quality of interaction between teams. The success of such roles is assessed not by ticket closure rates but by the richness of knowledge transfer and emergence of shared understanding across supposed boundaries. A Customer Experience Translator sitting between support and engineering, responsible for ensuring each team understands the other’s constraints and possibilities, is the kind of role the chiasmic ontology supports.

Instead of sprint retrospectives focused on velocity improvement, the organisation can hold sessions where teams collectively map recurring friction points, discovering that delays cluster around specific handoffs and revealing architectural rather than performance issues. The resulting patterns, not individual complaints, become organisational intelligence made visible through obstruction-mapping practice. Management’s role shifts from measuring output to cultivating the capacity to sense and navigate these obstruction topologies.

For every major initiative, the organisation can allocate a formal remainder budget, typically 20-30% of project capacity, pre-dedicated to addressing unforeseen consequences, cleanup, and system reintegration. The allocation is honest accounting of what change actually costs in chiasmic systems, rather than the conventional “slack” framing that treats remainder work as inefficiency. A platform migration that allocates three engineers for remainder work, whose explicit job is handling the emergent problems that were not in the plan (unexpected dependencies, legacy workarounds now surfacing, integration debt), implements the recognition the derived-category articulation of §26 supplies.

Instead of static quarterly objectives, the organisation can use directional vectors: teams assessed on demonstrated progress along a direction, not achievement of a fixed point, allowing adaptation to discoveries made en route. “Move toward deeper customer understanding of our value proposition” supports the kind of qualitative judgement the chiasmic ontology recognises, where “increase conversion rate by 15%” forces the team into the absorption-attenuation-decomposition cycle of §8.3.

The organisation can match the diversity of internal responses to the complexity of challenges faced, allowing multiple context-specific ways of working rather than imposing uniform processes. Different product teams using different methodologies (some Kanban, some Scrum, some ad-hoc) based on their specific coupling with customers and technology embodies what Ashby’s law has been pointing at: regulatory variety matched to environmental variety, deployed at the practice level rather than the metric level. Leadership’s role is ensuring adequate knowledge transfer between approaches, rather than enforcing uniformity that destroys the variety the practices need.

When teams consistently work around the official deployment process, the chiasmic recognition is that the workaround likely encodes critical knowledge about system behaviour that the official process ignores. The response is to study the workaround rather than punish it. The pattern recognition is the substance of organisational intelligence; the official process is a partial map that has become outdated.

The practices are not techniques to be implemented like software updates. They are stances, orientations, ways of attending to organisational reality that acknowledge its chiasmic, autopoietic nature. They cannot be reduced to checklists because they are modes of engagement with incalculability itself.

§34.1 From Control to Cultivation#

The shift from reductive to chiasmic organisation is a philosophical shift rather than a methodological one. It requires leadership to relinquish the fantasy of total knowledge and embrace a different relationship with organisational reality. Leaders in chiasmic organisations cultivate sensing capacity rather than demanding reports: they create conditions where signals about organisational reality can emerge and be heard, rather than filtering everything through predetermined metrics. They work explicitly with remainder: after every major change, they ask not “did we hit our targets?” but “what emerged that we did not anticipate, and what does that teach us about the system?” They navigate by obstruction: when work becomes difficult, they ask “what is the system telling us?” rather than “how do we remove this obstacle?” because some obstructions are the system’s immune response protecting essential complexity. They protect boundary intelligence: they recognise that the most valuable organisational knowledge often exists at interfaces (between teams, between organisation and environment) and invest in roles and practices that make boundary work visible and valued. They enact purpose through experimentation: they understand that organisational purpose is not discovered through strategic planning but emerges through action, and they create fast feedback loops and preserve the capacity to change direction based on what is learned.

None of these practices are measurable in the conventional sense. Their success is visible in system behaviour: increased adaptability, reduced burnout, better retention of essential knowledge, more real innovation. These outcomes resist quantification because they are chiasmic, existing in the intertwining of organisation and environment, observer and observed, measurement and measured. The practices are rigorous in the application sense; they apply careful attention to what actually exists rather than demanding it conform to what can be calculated. Spinoza’s formulation captures the position: “no one has yet determined what the body can do” (Ethics, III, P2 Scholium). No one has determined what organisations can do when freed from the illusion of mathematical control.

§35. Preparing the Counterargument#

The argument anticipates six characteristic objections, each addressed below.

“But we still need metrics; we cannot manage without measurement.” The objection concedes the argument while attempting to reject it. The paper does not claim organisations can exist without tracking or awareness. It demonstrates that formalisation at the level of KPIs and dashboards destroys the very intelligence it claims to measure. The horizon-mismatch theorem of §24 establishes the structural reason: any regulator operating at horizon shorter than its substrate cannot capture the substrate’s variety in any measurement frame the regulator deploys. The objection assumes measurement approximates organisational reality where the argument shows measurement transforms that reality, destroying what it claims to capture. Saying “we need metrics” after engaging this argument is structurally analogous to saying “we need phlogiston” after the oxygen theory of combustion: the need is grounded in a misapprehension of the phenomenon.

“This is too theoretical for practical application.” The objection demonstrates the blindness the paper diagnoses. Part IV’s Tuesday-morning phenomenology shows that chiasmic, autopoietic, and recursive phenomena are not rare theoretical exceptions but the ordinary substance of daily organisational life. Every bug report, every architectural decision, every attempt to categorise work for reporting contains the complexity the paper describes. The complaint that the argument is too theoretical typically means “I cannot recognise my own organisational reality in this description,” and the inability is internal to the conventional frameworks the objector has internalised. Metrics, KPIs, and objective measurement create the blindness that makes the argument seem theoretical. The objection proves the thesis: conventional management frameworks generate their own justification by destroying the capacity to recognise what they destroy.

“But companies using metrics succeed; they cannot all be wrong.” The objection confuses correlation with causation in the way second-order cybernetics warned against. Organisations succeed through the unmeasured remainder, through the shadow systems, informal networks, workarounds, and tacit knowledge that develop around official metrics, rather than because of the metrics themselves. Part IV documents the dynamic in detail. The actual work happens in the spaces metrics cannot see, and “success” is in any case measured how, if not by the metrics whose adequacy the paper has put into question. The objection assumes the legitimacy of the measurement framework to defend the measurement framework, which is circular reasoning that the cascade argument severs. Organisations persist through their autopoietic, chiasmic capacity to recreate themselves through processes that exceed formalisation. Metrics do not enable this persistence. At best they are noisy. At worst they degrade organisational intelligence by forcing energy into performance maintenance rather than adaptive response.

“Surely some measurement is better than none.” The objection commits a category error by assuming metrics approximate organisational reality with some degree of inaccuracy. The cohomological argument of §25 articulates the structural feature: measurement at the elementary formalism level annihilates the higher-order structure (the $H^n$ groups) where organisational intelligence resides. The structural-analogue articulation supports the substantive claim: measurement of $H^0$ content is not measurement with error bars; it is the systematic discarding of the higher cohomology where the intelligence lives. The analogy is direct: saying “some measurement is better than none” is structurally analogous to saying “some bloodletting is better than none” when treating bacterial infection. The intervention operates on a different ontological level than the problem, and at that level the intervention damages rather than approximates.

“Organisations are not really topoi or cohomological structures.” Precisely. The objection misunderstands the argumentative direction. Organisations are not topoi; they are more complex than topos theory can capture. The cascade argument does not claim organisations are topoi. It claims that even if they were, even if humanity’s most advanced mathematical tools could represent them, capture would still fall short in the constructive-embedding sense (per the §25.3 distinction). Therefore attempting to measure them with elementary formalisms (arithmetic in KPIs, real analysis in averages) is structurally inadequate, with the inadequacy quantified by the horizon-mismatch theorem of §24. The objector has accepted the argument while misunderstanding its direction.

“We can optimise locally even if we cannot measure everything.” The objection misunderstands the topological nature of organisational reality. In chiasmic systems, elements are not independently modifiable; they exist only through their relations. Mathematically, the feature is articulated by the fact that in derived categories and ∞-topoi, morphisms (relationships) are not external connections between pre-existing objects; they constitute the objects themselves. The cohomological structure of §25 articulates the consequence: if organisational intelligence resides in higher cohomology groups $H^n$, then these groups are defined by exact sequences where each element is simultaneously the kernel of one map and the image of another, and changing any single element forces changes to propagate throughout the entire sequence. There is no local in a structure where locality is defined relationally.

The spider’s web analogy captures the feature in compressed form. The web vibrates as one because it is one: one differentiation, one continuous intertwining where the distinction between local and global is itself produced by the weaving rather than antecedent to it. The web cannot be touched at one point without the entire structure moving, not because of mechanical linkage between independent points but because the points are nothing more than nodes in the pattern of tensions that the whole web maintains. Local optimisation in such a structure is not an adjustment of one variable in a multivariate function; it is the perturbation of a node in a coherent sheaf, forcing the entire structure to reorganise around the perturbation. The optimisation succeeds locally (the velocity metric increases) through global reorganisation (work gets redefined to be measurable, architectural debt accumulates, shadow systems proliferate). The local-versus-global distinction does not apply at the ontological level the chiasmic argument operates at; it is an artefact of the measurement frame that the horizon-mismatch theorem of §24 has put into question.

§36. Conclusion#

Three connected lines of argument run through the paper. The historical analysis has traced the genealogy from Beer’s cybernetic insights through second-order critique to phenomenological alternatives, and has shown this to be a necessary evolution through the internal logic of the positions taken. Each phase revealed limitations in the previous, driving theoretical development through internal necessity rather than external fashion.

The philosophical development has established the chiasmic ontology as the account of what organisations are once the observer-stance has become untenable. Organisations are patterns within fields they co-create, exhibiting reversibility, autopoiesis, and emergent intelligence that exceed formal capture in the constructive-embedding sense. The chiasmic ontology recovers what the second-order critique displaced and supplies the framework within which the mathematical apparatus operates.

The strategic and mathematical contribution has created protected space for non-reductive organisational understanding through the proof-register foundation supplied by the horizon-mismatch theorem of §24 and the Novikov-Boone result of §26, supplemented by the structural-analogue articulations of §§27-29. The argument deploys the advanced apparatus to establish structural limits on what mathematical formalisation can do with organisational phenomena, while remaining honest about which register supplies which result.

§36.1 The Honest Paradox#

The paper uses mathematics to establish the structural limits of mathematics when applied to organisational reality. The procedure has precedent: Gödel used arithmetic to prove arithmetic’s incompleteness, Turing used computation to prove computation’s limits, and Arrow used preference logic to prove preference aggregation’s impossibility. The present argument follows the same tradition, using categorical mathematics and information-theoretic constraints to demonstrate that organisational phenomena exceed categorical capture in the constructive-embedding register while remaining articulable in the structural-analogue register. As Wittgenstein (1953, Part I, remark 309) put the philosophical task, “what is your aim in philosophy? To show the fly the way out of the fly-bottle.” The mathematical apparatus shows the way out of the metric-fly-bottle by deploying the institutional register’s preferred language to mark its own limits.

§36.2 The Space Created#

The space the apparatus creates supports phenomenological investigation without reductive interruption, recognition of chiasmic intertwining as structural rather than metaphorical, respect for organisational intelligence as emergent from obstruction, acceptance of transformation remainder as necessary feature, and engagement with organisational reality on its own terms. The space is now secured, since demands for KPIs, impositions of dashboards, and the institutional pretence of optimisation face mathematical results in the proof register (the horizon-mismatch theorem and Novikov-Boone application) and structural-analogue articulations in the higher register. The protection operates at technical, philosophical, and strategic levels.

§36.3 Future Directions#

The immediate next step is the development of a fully worked chiasmic praxis built upon the principles set out in §32. The development is not a call for new methodologies to be marketed to organisations; it is an invitation to practitioners to develop skilful navigation within acknowledged incalculability. A chiasmic toolkit cannot provide algorithms for control, but it can provide heuristics for cultivation: ways of attending to organisational reality that honour its autopoietic, chiasmic, and incalculable nature. Such a praxis cannot be formalised in the conventional sense because formalisation in the constructive-embedding sense is what the paper has shown is structurally inadequate to organisational phenomena. The praxis can be cultivated through communities of practice, through apprenticeship in sensing obstruction patterns, through developing the embodied skill of working with remainder rather than against it. The work that follows from this paper is inhabitation of its insights rather than implementation of its findings.

Three companion papers extend the argument in directions this one has not pursued. The first develops the thermodynamics of organisational extraction, formalising the three-site absorption (worker, substrate, externality) implicit in Part IV’s Tuesday-morning narrative. The second develops the dissolution of orders, treating the breakdown of organisational orderings as a structural feature of late-modern institutional life rather than an exceptional crisis. The third develops the membrane architecture, articulating the chiasmic ontology of organisational boundaries through the formal apparatus of biological membranes. Together with the present paper, the four constitute a post-cybernetic series on organisational ontology.

Empirical studies grounded in the chiasmic model would investigate boundary zones, reversibility patterns, and emergence dynamics in actual organisations. The methodological challenge is direct: any empirical methodology that imposes the observer-stance the chiasmic argument displaces will fail to capture the phenomenon it seeks to study. The work points toward participatory and phenomenological research methodologies developed in the traditions of Schön (1983), Lave and Wenger (1991), and Stiegler (1998-2001). The philosophical implications of the argument extend beyond organisational theory to other domains where formalisation is treated as the default register of legitimate knowledge: legal reasoning, economic measurement, educational assessment, public-health metrics. The structural features the paper identifies are not peculiar to organisations; they appear in any social phenomenon where the measurement apparatus is part of what is measured.

The thesis can be stated in compressed form, with which the argument concludes. An organisation cannot calculate the field it is part of co-creating through the act of calculation. The attempt to do so generates a stable configuration in which the inadequacy is collectively known and institutionally maintained. This state absorbs the impossibility into human bodies, organisational substrates, and discharged externalities. The argument’s empirical anchor is the three documented cases of §31. The argument’s structural anchor is the horizon-mismatch theorem of §24. The argument’s phenomenological anchor is the Tuesday-morning narrative of Part IV. The three anchors converge on a single point. The cost of insisting on calculation where chiasmic relation is the structure is paid by people whose lives, capacities, and relationships absorb what the metrics cannot register. The argument of this paper is that the cost is structural and consistent across institutional contexts, and the response is not better metrics. The response is the recognition that what the metrics cannot register is, in fact, where the work is done and the cost is paid.