Question 3: Which Function Is Missing?
The short version
Any system that maintains its existence through time (a company, a hospital, a government department, a school, a community organisation) must perform five distinct functions. They are structural requirements, architectural facts about what any persistent system must do. Remove any one and the system develops a predictable failure mode: the specific failure that absence of that function produces.
The Viable System Model, developed by Stafford Beer from the architecture of the human nervous system, identifies these five functions and shows how they relate to each other. Its power as a diagnostic tool is specificity: it does not tell you that the system is “dysfunctional” in general. It tells you exactly which function is missing and what class of failure that absence produces.
The third diagnostic question is: which of the five necessary functions is absent or broken in this system, and what is the system unable to do as a result?
Why this question is non-obvious
Most organisational diagnosis focuses on performance (what are the outputs?) or culture (how are people behaving?). These are legitimate questions, but they are downstream of the structural question. A system with the wrong architecture will produce poor performance and dysfunctional behaviour regardless of how capable or motivated its people are, because the structure determines what kinds of action are possible.
The diagnostic question is “what is structurally absent?” rather than “what went wrong?” The first leads to blame and remediation. The second leads to architecture and design.
A system missing its intelligence function will fail to adapt to a changing environment. Every individual within it could be excellent, and the system would still fail, because no one’s job is to model what is coming and feed that intelligence into policy. The failure is not personal. It is architectural.
A system missing its coordination function will have its operational units working against each other, destroying collective performance through local optimisation. Again, each unit could be performing well by its own metrics. The failure emerges at the level of the whole, from the absence of a function that manages interactions.
The VSM makes this diagnostic. Once you know which function is absent, you know what to build, and you know why all the interventions aimed at performance and culture have failed to address the root cause.
The five functions
Beer derived the VSM from the architecture of the central nervous system: the most sophisticated viable system we know. The five functions correspond to distinct neurological structures. Beer argued that any system which needs to maintain its existence in a complex environment will, through selection pressure, converge on this architecture. Systems that lack any of the five functions are less viable and tend either to fail or to develop informal substitutes.
System 1: Operations
The set of primary activities that define what the organisation actually is. In a hospital: wards, theatres, diagnostic units. In a company: the business units that generate products or services. In a government: the delivery agencies. In a school: the classrooms.
Each System 1 unit is itself a viable system: it has its own internal management, its own operational variety, its own interface with its environment. This is the recursion property of the VSM: the same five-function structure appears at every level of scale.
System 1 units should be as autonomous as possible: capable of managing their own variety without constant reference to higher levels. The more they can self-manage, the more their variety is available for local regulation rather than consumed in upward reporting.
System 2: Coordination
The anti-oscillation mechanism. When multiple System 1 units share resources, customers, schedules, or physical space, their activities can interfere. One unit takes on a commitment that creates a problem for another. One unit’s success draws resources away from another. Without coordination, these interferences accumulate into systemic instability, not through malice, but through the normal operation of units optimising locally.
System 2 does not direct or command. It provides the shared standards, scheduling systems, communication protocols, and resource allocation mechanisms that allow System 1 units to operate in proximity without generating interference.
Its absence is recognisable: the left hand does not know what the right hand is doing. Departments make competing promises to the same customer. Projects fight over the same staff. Agencies solve each other’s problems without knowing it, or create each other’s problems without noticing. The units are working; their interactions are destructive. The failure is in coordination, not individual performance.
System 3: Operational Management
The function that looks at the collective of System 1 operations and manages them for the benefit of the whole. System 3 allocates resources across units, sets accountability structures, optimises synergies between units, and ensures that the overall operational picture is coherent.
Critically, System 3 has a direct audit channel, designated System 3*, that bypasses normal reporting and provides real-time ground truth about operational reality. Normal reporting is always filtered: it passes through people who have reasons to present information selectively, in formats that serve their interests. System 3* is the mechanism by which management independently checks whether what it is being told matches what is actually happening.
Without System 3*, management is governing a model of operations (the model that official reporting constructs) rather than operations themselves. In stable, honest environments these may coincide. In stressed, complex, or politically charged environments they diverge significantly, and management’s decisions are systematically based on the wrong information.
System 4: Intelligence
The function that looks outward and forward. System 4 scans the environment for signals of change (technological, competitive, regulatory, social, ecological) and models what those changes mean for the system’s future viability. It feeds that intelligence into System 5 for policy decisions.
System 4 is the future-facing half of the organisation. System 3 manages the inside now. System 4 manages the outside then. The balance between them (maintained by System 5) is the organisation’s fundamental adaptive mechanism.
System 4 is the most commonly missing function in organisations and institutions. It is easy to defund in the short term because its outputs are not operationally immediate. Strategy documents exist in most organisations, but genuine System 4 (ongoing environmental scanning, modelling of emerging conditions, real feeding of intelligence into policy) is rare.
Its absence is recognisable: the organisation is always surprised by the future. It responds to crises rather than preparing for shifts. Strategic planning happens annually and is immediately overtaken by operational pressures. The environment changes; the organisation does not adapt until the change has already produced a crisis.
System 5: Identity and Policy
The function that holds the system’s identity: what it is, what it is for, what values govern its behaviour. System 5 sets policy at the level of the whole system, not individual operations. It is the answer to the question: “when System 3 and System 4 disagree (when present operational demands conflict with future strategic requirements), how do we resolve that tension?”
System 5 is not the chief executive. It is the function the chief executive performs when they are acting as the custodian of the whole system’s identity and purpose, rather than managing operations or reading intelligence reports. Many organisations have individuals in System 5 positions who spend almost all their time in System 3 mode (managing operational detail), leaving the identity and policy function effectively vacant.
System 5’s failure mode is capture: a powerful System 1 interest takes over the identity function and bends the system’s purpose toward its own continuation. When this happens, what the system believes itself to be (and therefore what it optimises for) diverges from its stated purpose. The gap between the mission statement and the operational reality is the signature of System 5 capture.
How to apply the question
Five diagnostic moves, each targeting a specific function:
1. Find who is doing System 4. Ask: whose job is it to scan the environment and model the future? Not to write the annual strategy document: to continuously monitor signals of change and bring them into policy decisions. If you cannot name a person, a team, or a clear process, System 4 is absent.
Then ask: does what they produce actually reach System 5 and change policy? A System 4 function that generates intelligence that no one acts on is structurally equivalent to no System 4 at all. The channel from System 4 to System 5 must be open and operational.
2. Check the System 3/System 4 homeostat. System 3 and System 4 should be in productive tension. System 3 says: we need to maintain current operations. System 4 says: we need to adapt for what is coming. System 5 holds that tension and resolves it through policy.
If System 3 dominates (operational pressures always consume the resources, attention, and decision-making capacity that should go to System 4), the organisation is losing its adaptive function in real time. This is the standard failure mode in public institutions under resource pressure: every pound of slack that might have supported System 4 gets consumed by System 3 demand.
If System 4 dominates (strategy without operational grounding), the organisation loses touch with its actual capabilities and constraints. Plans are made that operations cannot deliver.
3. Map the System 2 gaps. List the System 1 units. For each pair of units that share resources, serve overlapping populations, make commitments that affect each other, or compete for the same inputs: ask: what is the coordination mechanism? Is there a shared protocol, a scheduling system, a resource allocation process that prevents interference?
Pairs with no coordination mechanism between them are System 2 gaps. Each gap is a site of predictable interference: the question is only when it will manifest and how badly.
4. Check System 3: find the unofficial truth.* Ask front-line workers, not managers, what is actually happening operationally. Ask customers what they experience. Ask the people who absorb the residual variety (identified in Question 2) what they have to improvise around daily.
The gap between what front-line workers describe and what management reports is the measure of System 3* failure. If management is surprised by what front-line workers say, the audit channel is broken.
5. Test System 5 for capture. Ask: when there is genuine conflict between the organisation’s stated purpose and the interests of its most powerful operational units, which wins?
If the operational interest wins reliably, System 5 is captured. The organisation has effectively redefined its purpose around the continuation of its most powerful unit, regardless of what its documents say.
The failure modes, by function
System 4 absent: the permanently surprised organisation
The organisation has no one consistently scanning the environment and modelling what is coming. It responds to crises because it did not see them approaching. Competitors arrive, technologies shift, regulatory environments change, demographics move, and the organisation encounters each as a surprise rather than a prepared-for transition.
This is the condition of most national health services with respect to demographic ageing. The information has been available for decades. There is no structural System 4 that converts that information into ongoing policy adjustment. The crisis arrives (too many elderly patients, not enough provision) as if it were unexpected.
System 2 absent: the coordination disaster
Each System 1 unit is performing, by its own metrics. Their interactions are destroying collective performance.
A technology company’s sales team makes product commitments that the engineering team cannot deliver. The sales team’s metrics are excellent. The engineering team’s delivery record deteriorates. Customer satisfaction falls. No one is failing individually. The System 2 mechanism (the shared process that would align what sales promises with what engineering can build) does not exist.
A government response to a social problem involves housing, health, employment, and social services agencies. Each agency delivers its own service adequately. The gaps between agencies (the situations that require coordination across boundaries) are not managed by anyone. The people who fall into those gaps experience the system as unresponsive or incoherent.
System 3 absent: the management surprise*
Management receives reports that accurately describe the world as the reporting system models it. The reporting system models a simplified, politically curated version of operations. Management makes decisions based on that model. Those decisions produce surprising results because they were applied to the model, not to operations.
A hospital’s performance dashboard shows acceptable waiting times. Front-line nurses know that the waiting time figures are met by reclassifying waits. Management is making decisions about capacity and resource allocation based on figures that do not reflect what is happening in the wards. The audit channel that would give management unmediated access to operational reality does not exist.
System 3/System 4 homeostat broken: the strategically frozen organisation
The organisation produces strategy documents. They are written by a planning team that consults widely, synthesises external intelligence, and produces recommendations. The recommendations are reviewed at the annual strategy retreat. Operational pressures then consume all available resource and attention for the following year. Next year’s strategy document notes that last year’s recommendations were not implemented and recommends them again.
The problem is structural: System 4’s outputs have no reliable path into System 5 decisions, and System 3 has first claim on all resource. The organisation’s adaptive function is intact on paper and absent in practice.
System 5 captured: the mission-drift organisation
A university’s stated purpose is the creation and dissemination of knowledge. Its most powerful System 1 unit is its research enterprise, which generates prestige, funding, and international standing. Over decades, hiring, promotion, resource allocation, and institutional identity are progressively aligned with research performance. Teaching (the primary means of knowledge dissemination) is treated as a cost centre. The mission statement still says “teaching and research.” The operational logic says research, and teaching is what researchers do when they cannot get out of it.
System 5 has been captured by the most powerful System 1 interest. The identity of the institution has drifted to match the interests of its most powerful unit. The gap between the stated mission and the operational reality is the mechanical output of System 5 capture.
The recursion check
The VSM is fractal: the same five-function structure appears at every level of scale. A national health system has Systems 1–5. Each hospital within it has its own Systems 1–5. Each ward within the hospital has its own Systems 1–5.
This has an important diagnostic implication: a function that appears to be present at one level may be absent at another. A government department may have excellent System 4 at the national level (a strong policy intelligence unit) while the same function is completely absent at the regional delivery level. The regional offices are executing policy with no mechanism for sensing what is changing in their specific environments and feeding that back into local adaptation.
When applying Question 3, specify the level of recursion you are examining. The diagnosis at one level does not transfer to other levels. Each level must be checked independently.
What the answer tells you
The VSM diagnosis produces a specific architectural prescription, not a general recommendation for improvement but a statement of what needs to exist.
If System 4 is absent: Designate resource, authority, and process for ongoing environmental scanning. Create a direct channel from System 4 to System 5. Protect it from System 3 consumption. The System 4 function does not need to be large: it needs to be real, consistent, and connected to where decisions are made.
If System 2 is absent: Design the coordination mechanism for each identified gap. The mechanism should be as lightweight as possible (shared protocols, scheduling systems, resource allocation rules) rather than new management layers. The goal is anti-oscillation, not direction. System 2 should not tell units what to do; it should prevent them from doing things that create problems for each other.
If System 3 is absent:* Create direct audit channels. Regular visits to front-line operations. Skip-level conversations. Real-time operational dashboards that bypass summary reporting. The goal is to give the System 3 function unmediated access to what is actually happening, in addition to official reporting.
If the System 3/System 4 homeostat is broken: Protect System 4 resource from System 3 consumption structurally, not through exhortation but through budget ring-fencing, scheduled System 5 time dedicated to System 4 inputs, and accountability for strategic adaptation as a distinct performance dimension.
If System 5 is captured: This is the hardest intervention, because System 5 capture is a power and governance question, not an organisational design question. The mechanism that would reclaim System 5 (an independent board, multi-stakeholder governance, external accountability) will be resisted by the interests that have captured it. The diagnosis is clear; the intervention is political.
Before moving to Question 4
Questions 1, 2, and 3 together give a structural diagnosis: what the system is actually controlling, whether it has the capacity to control it, and which architectural functions are present and absent.
This is enough to know what needs to be built. But building it into an existing system is not the same as building a new one. The existing system has evolved its own feedback dynamics: its own loops, its own ways of stabilising around the current state, including the absent function. Adding a System 4 into an organisation that has operated without one for years will encounter the feedback structures that developed to fill that gap, or to route around it. Simply constructing the missing function is necessary but not sufficient.
Understanding those feedback dynamics (where they are amplifying the problem rather than correcting it) is the subject of Question 4.
Reference
The theoretical foundation for this question is the Viable System Model, developed by Stafford Beer.
Primary sources:
- Brain of the Firm (1972, second edition 1981): the neurological derivation of the VSM
- The Heart of Enterprise (1979): the organisational application
- Diagnosing the System for Organisations (1985): the practical diagnostic manual
The VSM was operationally demonstrated at national scale in Project Cybersyn (Chile, 1971–1973). It has since been applied to businesses, hospitals, UN agencies, governments, and educational institutions.