Definition
An institutional capability deliberately maintained so the institution can still question, replace, or withdraw a system: retained expertise, a working manual path, independent records, a viable alternative supplier. Preserved capacities are named, owned, and funded even while idle, and they are kept honest by withdrawal rehearsal.
It matters because these are the first capacities efficiency removes. Their erosion is a governance event rather than a saving, since it is the mechanism by which a system consumes the conditions of its own contestability, destroys institutional reversibility, and ends in institutional capture. Registered capacities also decay quietly through capacity depreciation, which is why exercise dates rather than register entries are the evidence.
Tests that show it is in place
- Preserved capacities are listed with an owner and a budget line, separate from the system's own budget.
- Any proposal that reduces one is reviewed as a governance change, not as an efficiency measure.
- Each capacity is exercised on a cadence so that its continued existence is demonstrated.
Evidence to keep
- Artifact: A preserved capacity register with owners, budget lines, and exercise dates.
- Behavior: A cost reduction rejected or amended because it would erode a preserved capacity.
- Metric: Share of registered capacities exercised within their declared interval.
References
- Laws for engineering delegated intelligence · Doctrine