Contribution
A first-hand account of where the Viable System Model came from, of the method Beer claims makes it a scientific model rather than an analogy, and of its use as a diagnostic: organizational “pathology” read off as the failure of one of five subsystems. It is a reflection on the model, not a statement or test of it.
Key insight
That an organization exists says nothing about whether it is viable. Beer’s test is structural: whether it has the regulatory functions any independently viable system needs, at each level of recursion, and whether they are connected with requisite variety. A system can lack one and go on existing for a while, reacting without foresight.
Assumptions
- Viability can be characterised by invariances that hold across brains, firms, states and cells; mapping “across these boundaries” is legitimate (p. 5).
- Ashby’s Law of Requisite Variety is the central law of regulation (p. 7).
- A system’s elements can be treated by their function under role constraints, not their identity, so people’s free will does not break the model (p. 9).
- Boundaries, the number of subsystems and the identity of System Five are nominated by the observer (p. 11).
Key results
- Recursion (pp. 4, 10, 16): “In a recursive organizational structure, any viable system contains, and is contained in, a viable system”. The topology must be an exact replica at each level.
- Five subsystems (p. 10), necessary and sufficient for any system “capable of maintaining its identity independently of other such organisms within a shared environment”. System One’s elements are its “autopoietic generators” and are themselves viable systems. The whole interacts with its environment through its Systems One and Four “and not otherwise”.
- Identity by closure (p. 11): System Five closes the system and “only this determines an identity”. Allende told Beer that System Five was “the people”.
- Requisite variety in three places (p. 7): blocks of variety, channels and transducers. These become the first three principles of organization (Appendix 2).
- Pathologies (pp. 12–13), quoting Brain and Strategy (1983): removing any of the five functions from a subsidiary may kill it; System Five may fail to represent the whole to the larger system; without Two, the units of One fight; without Four, Three and Five collapse into each other. Beer’s gloss is that Five people are usually promoted from Three and “dive down” into its problems, leaving a metasystem that “responds reactively” and is “incapable of planning and foresight (Four) and will and judgment (Five)”.
- Takeovers (p. 12): autonomy “does not mean zero interference”, and the cohesion of two recursions inevitably takes up the new subsidiary’s variety.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Five subsystems are necessary and sufficient for viability in any system | weak | asserted; validation explicitly not attempted here, referred to the books |
| C2 | Every viable system contains and is contained in a viable system | moderate | the model’s central theorem; argued, not demonstrated here |
| C3 | The VSM is a scientific model, not an analogy | moderate | the homomorphic-mapping methodology (pp. 5, 14), with no test reported |
| C4 | Organizational breakdown can be diagnosed as failure of a subsystem, quickly | weak | the author’s consultancy experience, recalled |
| C5 | Without System Four, Three and Five collapse and the system only reacts | weak | quoted diagnostic and anecdote |
| C6 | The model has been confirmed by others across many organizations and organisms | weak | a list of users and theses; “No systematic archive of applications has been kept” (p. 18) |
Concepts
- viable — “capable of independent existence” (pp. 3–4).
- variety — the number of distinguishable states of a system (p. 6).
- requisite variety — “Only variety can destroy variety” (p. 7); a regulator short of it must attenuate or amplify.
- recursion — the nesting of viable systems within viable systems.
- metasystem — Systems Three, Four and Five together; its collapse is the “decerebrate cat”.
- yo-yo technique — testing the model by mapping fresh systems down to it and back (p. 5).
Connections
It restates the model of Brain of the Firm (LIT-746, unread) and The Heart of Enterprise, and rests on Ashby’s law and Sommerhoff’s directive correlation.
Aubin’s viability (THEORY-134). The two uses of “viable” differ. Aubin’s is a property of a constraint set relative to a dynamics: a set is viable if from each of its states some evolution stays in it, and the set is fixed independently of the dynamics. Beer’s is a property of an organization’s regulatory structure: whether it has the five subsystems, connected with requisite variety, that keeping an identity independently needs. Aubin’s theory has no subsystems or recursion; Beer’s has no constraint set or kernel. They meet at two points, as the record reads them. Both separate being somewhere now from being able to stay: Beer’s “The fact that the societary system is there does not guarantee that it will always be there” (p. 12) is a state outside a viability kernel, in Aubin’s terms. And both leave the hard part outside the theory: Aubin takes the constraint set as given, and Beer says the boundaries and the identity fixed by System Five are nominated by the observer. No relation is declared.
Argyris (THEORY-145). Beer reports that danger points “deliberately concealed out of embarrassment or self-serving” tend to signal themselves to the diagnostician (p. 12). Argyris’s defensive routines are the opposite claim: the concealment is itself made undiscussable. The collapse of Five into Three, a metasystem that reacts and cannot judge, resembles an organization confined to single-loop correction. That parallel is the record’s; Beer does not draw it.
Ostrom (THEORY-149). Ostrom’s nested enterprises and Beer’s recursion both make nesting a structural mark of what lasts. Ostrom’s is a correlate of survival in common-pool institutions; Beer’s is asserted as a necessary condition for any viable system. No relation is declared.
Bearing on the record
Source of THEORY-153. For the owner’s essay:
- §6.1 (stability is not viability). Beer draws the distinction in organizational terms: systems that are there but not “effectively there” operate “at the margins of viability” and grow “less and less viable in front of everyone’s eyes” (p. 12).
- §8.3 (recursive self-governance). Recursion is the model’s central principle, though Beer’s recursion is structural nesting, not the essay’s revision of standards by procedures that are themselves revised.
- §7.1 (corporate endorsement). Beer locates identity in System Five’s closure and says nominating its components is hard because “who are ‘we’?” has different answers for shareholders, employees and others (p. 11). That names the essay’s question without answering it.
- §6.3 (extraction from above). The takeover passage says the higher recursion inevitably takes up the subsidiary’s variety, whatever is promised about its autonomy.
Limitations
By its own account the paper neither expounds nor validates the model. The evidence is the author’s consultancy, recalled, and a list of other users. The claim that five subsystems are necessary and sufficient sits beside the admission that “five” is arbitrary. The copy lacks one page and the figures, including Figure 4, the diagram the VSM section is built around.
Open questions
What result would count against the model: Beer invokes falsifiability but names no observation the model forbids. How System Five is to be identified in a contested organization other than by the observer’s imputation, which Beer calls “probably idiosyncratic”.