STRIXFORMES
Research · 5 nodes

Relational coherence

Can understanding and changing the relationships inside a system preserve an action you had already decided to take?

That is the question the whole programme comes from, and the short answer is: not always, and the conditions under which it can are the interesting part. Everything below is an attempt to say what those conditions are precisely enough to be wrong about.

PREPRINTDeposited and citable. Not peer reviewed.
SCOPEA formulation and a set of counterexamples. Not a new theorem.
VALIDATIONNo independent replication. No hardware test.

Four ways into the same question.

01 · UNDERSTANDWhere the question came fromThe original proposition, why its strong form is false, and what was left after it was narrowed into something answerable.02 · EXPERIMENTRun the viability modelChange what the observer is allowed to know, and watch the set of safe starting states move. The numbers are computed, not illustrated.03 · MATHEMATICSThe functionalWhat the formula says, what the sign of it means, and which parts of it are established work rather than mine.04 · EVIDENCEWhere knowing stops helpingSix counterexamples that refute the strong version of my own hypothesis. The results I trust most.

The formula

The state is not only physical. It carries the relationship configuration, the reserve that is actually usable, the commands still in flight, the set of hidden coefficients that have not been ruled out, and what the observer knows.

Zt=(xt,rt,et,qt,Θt,ℐt)
Physical state, relationship configuration, usable reserve, commands still in flight, the admissible set of hidden coefficients, and what the observer knows. Only the first of the six is what engineering usually calls the state, and the argument of the paper is that safety is a property of all six.

The functional then asks one question of that state: choosing the best policy you can actually implement, and letting everything you do not control go as badly as it is allowed to, what is the smallest safety margin you can still guarantee?

SFH(ℐ0)=maxπminωmint,ihi(Ztπ,ω)si
Choose a policy that respects what you actually know. Nature then chooses the worst it is allowed to do. Take the smallest safety margin that occurs, over every constraint and every step. That number is what you can guarantee — and each margin is divided by its own engineering scale first, because a temperature and a clearance cannot be compared until they are.

A value that is not negative means a safe policy exists for the whole horizon. A negative one means none does. For a finite, fully specified model that equivalence is exact, and it is a restatement of robust viability theory — not a result of mine.

What I am proposing is the naming and the organisation: that information freshness belongs beside information accuracy, that physical reserve is part of safety rather than a separate resource question, and that the action you intended is something to be preserved rather than dropped the moment a margin tightens.

What the sign does and does not mean

Where it stops working

The version of this hypothesis worth refuting is the strong one: that an observer who knew every parameter could prevent any accident. It is false, and it is false for reasons that have nothing to do with information.

Knowledge does not create corrective authority.

A system with no effective correction fails whether or not it is understood. Two safety constraints can be coupled so that no single action satisfies both. An exact reading can expire before the actuator responds. Looking can itself cost the margin that looking was meant to protect.

All six, in full

The record

Computing it when you cannot afford to

The exact functional grows combinatorially, so the paper derives a conservative one-step check that can be evaluated cheaply. Its useful property is that latency appears multiplied by drift rate: a measurement does not become wrong when it is taken, it becomes wrong while you are acting on it.

Passing it proves one step and nothing further. Treating a one-step certificate as a guarantee over a horizon is the most likely way to misuse any of this.

The certificate

STILL OPENWhat would make the Strixformes formula worth more than a name?

Every node

The research is part of the same graph as the systems and the philosophy, so each of these carries its relationships to the rest of the work rather than sitting in a separate section of the site.

Relational coherenceCan understanding and changing the relationships inside a system preserve an action you had already decided to take?exploringThe Strixformes functionalThe worst guaranteed safety margin a policy can still promise, given only what the observer actually knows.developingThe drift certificateA one-step check you can afford, for when computing the whole functional is too expensive.developingWhere knowing stops helpingSix ways a perfectly informed observer still fails. These are the results I trust most.stableRelational Viability Under Incomplete and Changing KnowledgeThe programme written up: the formulation, the propositions, the counterexamples, and an audit of what the earlier experiments actually showed.preprint
UNPUBLISHED · NOT PART OF THE PREPRINT

Where the work turned next

Position, total geometry, permission, expansion, invariants and the centre of a structure. None of these is published, none has a formal definition yet, and each page says how far it has got — a position I hold, an exploratory idea, or an open question. They are here because they grew out of the research, not because they share its standing.

Position comes before relationshipWhere something is decides what it can relate to, and which actions are permissible for it.exploringTotal geometryThe geometry as the boundary of permissible positions — not a drawn shape.exploringWhen does a contained thing fit?How a contained object's own geometry, its position and the conditions of the container decide whether it can exist there.exploringChanges of geometry, changes of permissionWhen the geometry changes, what is permitted changes — and coherence is whether the two move together.exploringWhat stays coherent when everything expands at once?The containing geometry and the relationships inside it expanding concurrently.seedTime, inside a structure, is a sequence of changesWithin a structure, time is the order in which its states change.exploringWhat survives every change of view?What an observer can rely on across observations — and whether knowing it changes the outcome.seedConstraintsA constraint is not only an obstacle. It is the shape of what is allowed.developingAccidents as relational outcomesAccidents as outcomes of interacting conditions, incomplete observation, available interventions and the continued viability of a system.developingWhere is the centre of a structure?If it is not simply its most connected part, the centre is set by the structure's relationships and parameters — and that still needs a definition.seed
OBSERVER ACTIVE
GRID 9 · TIMING 3.6 / 9