Adaptive systems over static systems
Why the preference for adaptation is conditional rather than absolute, and what a system must have earned before it is allowed to adapt.
This principle is the one most easily misread, because stated baldly it sounds like a preference for sophistication. It is not. It is a claim about which failure is preferable, and it comes with conditions that most adaptive systems do not meet.
The principle: adaptation is preferred to a fixed schedule — but only for a system that can detect when its adaptation is no longer justified, and stop.
The argument for adaptation
A static system encodes a decision made once and applies it indefinitely. Its central defect is not that the decision was poor; it is that the system has no mechanism for discovering that conditions have moved. It cannot be wrong in a way it can notice, which means it cannot improve and cannot alert.
An adaptive system can be wrong in ways it can notice. That is the whole of the advantage, and it is a large one — but notice that the advantage is about detection, not about accuracy. A system that adapts and cannot evaluate its own adaptation has taken on the additional failure modes and none of the benefit.
The conditions
Adaptation is justified only when the system can do three things, and a system missing any of them is better off static.
It must know when it is outside its competence. Conditions leave the range a model was built for. If that departure is not detectable, the system produces confident output in exactly the circumstances where confidence is unwarranted.
It must have somewhere to go. Detecting incompetence is only useful if there is a defined behaviour to fall back to — see governance before intelligence.
It must be able to account for what it did. An adaptive system that cannot explain a decision cannot be corrected after a bad one, so its adaptations accumulate without ever being audited.
Why the preference is not absolute
There are components in VECTOR that are deliberately static, and they are the ones that matter most in a crisis. Emergency shutdown does not adapt. Termination does not adapt. Their behaviour is fixed, simple, and reasoned about completely.
The rule that resolves this: adapt where being wrong is recoverable; be static where it is not. Sophistication is appropriate in proportion to how reversible the consequences are. An adaptive emergency stop is a contradiction — the entire value of the mechanism is that its behaviour is known in advance, including under conditions nobody anticipated.
Failure mode
The characteristic failure is adaptation during instability. A system that adjusts its own parameters while conditions are deteriorating amplifies exactly what it is trying to damp, and the resulting oscillation is difficult to diagnose because every individual adjustment was locally reasonable.
The defence is that adaptation is itself gated: the slowest components in the system are the ones that change it, and they decline to act while conditions are stressed.