The Viable System Model (VSM) is Stafford Beer's cybernetic model of organizations. Published in 1972. Built from Norbert Wiener's cybernetics, not business school intuition.

Five systems, each with a distinct function. Recursive structure: each viable system contains viable subsystems that are themselves viable systems. A company contains divisions; each division is itself a VSM.

Core insight: for an organization to be viable, it must be able to manage its own complexity. Ashby's Law: only variety can absorb variety. The management system must have at least as much variety as the system it manages.

VSM is not a framework for small improvements. It is a complete theory of organizational viability. The fleet implements VSM from first principles — not by following VSM, but by building what VSM describes.

The VSM is recursive. Each S1 unit is itself a viable system with its own S1-S5. msi01's 18 crew members = the S1 operations of msi01's internal VSM. IMHOTEP = msi01's S5 policy. BOSUN = msi01's S3 control. SIGNALS = msi01's S4 intelligence.

This recursion continues: each crew member's task queue is itself a small viable system. The fleet is a nested hierarchy of viable systems, each managing its own complexity at its own level.

In VSM, the algedonic channel is a direct signal that bypasses the management hierarchy. When something is critically wrong (or critically right), the signal goes direct from S1 to S5 — bypassing S2, S3, S4. This is the emergency channel.

Fleet analog: THE DEAL — Kay's direct contact permission. When something is critically wrong, the algedonic signal bypasses all layers → direct alert to Kay. This is not a bug in the hierarchy — it is a designed safety valve. Without the algedonic channel, critical signals get dampened by management layers until it's too late.

Implementation: SOSTLE L5 override. Direct message. No waiting for LAAM queue. Examples: msi01 down, data loss risk, sovereignty breach, critical crew conflict.

Ross Ashby's Law of Requisite Variety: only variety can absorb variety. The management system must have at least as much variety (complexity-handling capacity) as the environment it manages.

Fleet variety balance: silo diversity × school diversity × crew regulon depth

7 silos × 5 fermentation schools × 4 crew regulons = 140 variety combinations
Environmental complexity (cloud providers, model providers, data sources, regulatory environments): ~100+ combinations

Fleet has adequate variety. But:
Too little variety → brittleness (single point of failure, one school, one silo)
Too much variety → chaos (no coordination possible, every silo goes its own way)

The balance is maintained by: LAAM coordination (S2), KCF scoring (S3), SOSTLE modes (WHERE dimension), and fermentation school routing (controlled variety at the process level).