Research programme

Read it as a derivation, not a pipeline. A claim at any layer that cannot be traced upward is unfounded: a contract argued on its own authority is a preference with a guard on it. A layer that nothing below it implements is decoration: a philosophy with nothing under it is an essay. The chain is a test rather than a diagram. For any piece of work, name its layer, name what it answers to, and name what realises it. If either neighbour is missing, that is the finding.

My own programme is strong at the bottom and was, until recently, fragmented at the top. That is the opposite of the usual shape, where a manifesto sits above nothing that runs. The lower layers are governed engineering built in commercial practice over six months, with controls that must be shown to fail before their passing counts as evidence. What was missing was a single account of why any of it is true.

The published work is that account. It is not a separate activity from the engineering. The position paper's central claim — that a checking mechanism inherits a presupposition it does not check, so two materially different states of the world produce one observable — is the engineering problem statement generalised. The engineering reached it first, in its own terms. Publishing is how that layer becomes checkable by people who are not paid by me.

The inversion is worth stating plainly because it cuts both ways. A philosophy written after the controls exist can be checked against them, which is an advantage. It can also be a rationalisation of decisions already taken, which is the standing risk and the reason the evidence for each claim is graded separately rather than asserted together.