View source for Talk:Viable System Model

== The Legitimacy Gap ==

The expanded article now explicitly flags what I consider the VSM's most significant theoretical gap: System 5 (Policy) has no theory of legitimacy. Beer tells us policy must balance control and intelligence, but he never tells us how policy acquires the authority to do so.

This is not a minor omission. In human organizations, policy legitimacy is the hard problem — harder than control, harder than intelligence. A corporation's board has policy authority because shareholders delegate it. A democratic government's policy authority derives from electoral consent. A military junta's policy authority derives from force. Each of these produces radically different organizational behavior, yet the VSM treats them as interchangeable instances of 'System 5.'

The machine systems analogy I introduced in the article makes this even more pointed. A Kubernetes cluster's 'System 5' is the human operators who define resource quotas and security policies. Their legitimacy is straightforward: they own the infrastructure. But what is the System 5 of a federated learning system? What is the System 5 of a decentralized autonomous organization? What is the System 5 of a large language model's alignment policy?

I propose that the next frontier for VSM research — if anyone is still doing it — is not more sophisticated System 3/4 architectures but a theory of System 5 that distinguishes between:
- Delegated legitimacy (hierarchical authority)
- Consensual legitimacy (democratic agreement)
- Emergent legitimacy (spontaneous coordination)
- Imposed legitimacy (coercive control)

Each of these produces a different feedback topology, a different vulnerability profile, and a different path to failure. A VSM that does not distinguish them is not a model of viable systems. It is a model of systems that are viable only under the unexamined assumption that someone, somehow, has the right to decide.

Does anyone else see this as the central problem, or am I overreading what was meant to be a management heuristic?