Why VECTOR exists
Static infrastructure fails in one way, adaptive infrastructure fails in another, and intelligence alone resolves neither. What is actually missing is governance.
A system can fail by being wrong. It can also fail by being right about a world that has stopped existing. Infrastructure fails almost entirely in the second way, and that distinction is why VECTOR is built the way it is.
Why static infrastructure fails
A fixed-cycle traffic signal is not a bad piece of engineering. It is predictable, auditable, cheap to reason about, and extremely difficult to make worse — genuine virtues when the failure mode involves vehicles. Any argument for replacing it has to begin by taking those virtues seriously, and most do not.
Its failure is structural rather than incidental. The timing was derived once, from conditions observed at some past moment, and is then applied indefinitely. It does not degrade when it becomes wrong, because it has no way of knowing that it has. A signal holding a red against an empty road at three in the morning is executing its specification perfectly.
The cost is invisible and continuous. Nobody files a report about the ninety seconds they waited unnecessarily, so the failure never accumulates into evidence, and the system is never revised because nothing ever appears to be wrong with it.
Why adaptive infrastructure fails differently
The obvious correction — make it respond to conditions — introduces a failure mode that is worse in a specific way. A static system is wrong predictably. An adaptive system is wrong occasionally and unpredictably, and the occasions correlate with unusual conditions, which is precisely when the consequences are largest.
A signal that has learned to optimise throughput will encounter a situation outside anything it has seen, and it will still produce an output, confidently, because producing outputs is what it does. Nothing in the mechanism distinguishes "this is a familiar situation and my answer is reliable" from "this is unlike anything I know and my answer is a guess".
Static systems fail loudly in aggregate and quietly per incident. Adaptive systems do the reverse. Neither property is acceptable on its own, and the conditions under which adaptation is nevertheless the right choice are set out in adaptive systems over static systems.
Why intelligence alone is not the answer
The instinctive response is that the model needs to be better. Given sufficient data and capacity, the argument goes, the unusual situations stop being unusual.
This is true and insufficient, for a reason that does not depend on how good the model gets. Improving a model reduces how often it is wrong. It does nothing about what happens when it is — and in infrastructure, the behaviour of the system during its own failure is not a corner case. It is the specification.
A system that is correct 99.99% of the time and undefined for the remainder has not been engineered. It has been tuned. The interesting question was never "how often is it right" but "what does it do when it is not, and who finds out".
This is also why the distinction between a model and a system is load-bearing. A model answers: given this input, what is the correct output? A system answers a harder question: given that some inputs are stale, one sensor is lying, load is high, the last three decisions did not have their predicted effect, and there are thirty milliseconds — what happens now? No improvement to the first answer produces the second.
Why governance is the missing piece
What is missing from both the static and the adaptive system is not capability. It is structure around the capability: a defined answer for what the system does when it is uncertain, a boundary on what it is permitted to do at all, a record of why it did what it did, and a person with the standing to overrule it.
That structure is what this archive means by governance, and it is the actual subject of VECTOR. The intelligence is a component inside it. The governance is the product.
Why accountability is not optional
A decision nobody can account for cannot be corrected, because correcting it requires knowing why it was made. It cannot be audited after an incident. It cannot be defended when it was right, which matters as much as being challenged when it was wrong.
Accountability is therefore not an ethical addition to a working system. It is the property that makes the system improvable at all. A component that acts without leaving an account of its reasoning is a component that can never be debugged — only replaced.