A decision-dependence memory system for capital allocation.
Invariant does not predict what will happen. It shows what would matter if something did.
Investment organisations repeatedly face a hard question:
This question is rarely answered structurally. Most teams rely on narrative, recollection, or committee consensus. That fails because belief does not crystallise at a single moment — it hardens through reuse.
Invariant is built on a single primitive: dependence under reuse, elevated only by counterfactual consequence.
Invariant is not a trading system, an approval workflow, compliance tooling, explainability software, or decision rationale capture.
It does not optimise strategies or recommend actions. It is an evidence layer, not a governance layer.
An artefact becomes load-bearing only when:
Artefacts that persist without being tested are explicitly labelled: persistent but untested. This is not failure — it is information about current exposure.
Fictional but mechanically realistic.
| Component | Probe | Status |
|---|---|---|
| Dataset A | Removal / substitution | None |
| Cost Model | Alternative cost simulation | Partial |
| Decay parameter | Retuning sensitivity | Partial |
Invariant survives success, politics, and time because it does not ask for virtue, acknowledgement, or confession.
It allows people to do what they always do — reuse what works — and then faithfully records the consequences of that behaviour.