Interrogations
Why the last cut is the one that ends the recovery
The budget line nobody could justify, removed, with performance improving right after.
The unit at full utilization, circulated to the others as the benchmark.
The recovery plan that assumes somebody has spare capacity to run it.
Spare capacity has no output attached to it, which is why every review finds it indefensible.
The pattern
What the cut removed was not output. It was the inventory of alternatives — the number of ways the work could still be completed once the usual way stopped working. That inventory never appeared in a deliverable, so it was never defended.
Utilization then reaches its maximum, and waiting stops behaving proportionally to load. Disturbances that used to be absorbed locally now require a decision. The operation still delivers the same result. It has only lost the ability to return.
The system that cannot return is the same system on the dashboard, reporting the same numbers.
The cost
- Unbounded waiting — past a certain load, delay stops tracking the work that caused it
- Lost release — the small failures were the pressure valves, and they were optimized away first
- Escalation by default — nothing is absorbed locally, so everything travels upward
- Coupled failure — one node with no reserve transfers to the next, which has none either
- Single path — one supplier, one process, one way of doing it, and nothing to fall back on
The question is not how lean the operation is. It is how many ways it still has of doing the work when the usual one stops.
The path
Count the alternatives. Not the headcount and not the budget: the number of distinct ways each critical operation could still be completed tomorrow if the normal route were unavailable. That count is the reserve, whatever the ledger calls it.
Slack is paid for in advance and never shows up as a return. That is why it is always the first thing cut.
The reserve was the option. The option was the strength.