Assumptions the model rests on

Each of these is built into the arithmetic. If one is false, the output is wrong in the specific way stated.

1. Decision latency dominates structural health. Backlog at boundaries carries half the composite penalty, the largest single share. If latency matters less than that, the score over-rewards removing queues relative to fixing incentives.
2. Authority misplacement compounds. A team without local authority is charged three times: its capacity is cut, it counts toward the escalation share and it drives the influence divisor when others depend on it. Its escalations also land as load on the authorities it resolves through, so a saturated centre registers as latency. If authority mattered less than this, the delegation moves the guide favours would be systematically overpriced.
3. Concentration is scale-relative (the prince band). All three authority charges are scaled by the population of the unit an escalation actually resolves in: the nearest enclosing unit holding a clean authority, falling back to the whole frame. The factor is attenuated to 0.3 up to 150 people, at parity by 200, then amplified with the log of the population and capped. This is Machiavelli read structurally: the Prince governs the small state well and the durable large state is a republic (or a federation of Dunbar-sized principalities), with the survivor cap conceding that rare princes persist at scale. If the Dunbar horizon is the wrong hinge (a hundred-person organisation that already needs a republic, a founder who genuinely scales command past a thousand), the score over-rewards small autocracies or over-punishes large ones in exactly that band. Contested ownership is never attenuated by scale.
4. Every dependency costs alike. Coupling counts edges, not their nature; a well-interfaced, cheap dependency and a daily-friction dependency weigh the same. Where cheap dependencies are common, coupled organisations read worse than they are.
5. Delay converts to load linearly. Arrivals inflate by 0.25 per unit of propagation delay. If the true effect is super-linear (oscillation, compounding rework), high-latency boundaries are under-priced, not over-priced.
6. Workload arrives uniformly. Every team receives the same arrivals per turn. If load concentrates on hotspots, the score is too kind: one drowning team is averaged away.
7. A shared unit can arbitrate its sovereigns. A dependency between two teams that both hold clean authority is priced as fragmentation only when they share no enclosing domain; any common unit counts as a roof under which the conflict can be resolved. Where declared units are nominal (a holding company that arbitrates nothing), fragmentation is under-priced.
8. The declared structure is the real one. The model scores the organisation you drew. If informal authority routes around the formal graph, the score describes an organisation that does not exist.