Supply Network — the Materials Risk Engine
CorridorsWho gets hurt first — and did we call it?
The payoff layer: ripples from each risk point to who gets hurt first — and the engine's own frozen predictions pinned where they happen, graded honestly once their deadline passes.
For the kids: The ripple. When a spot turns red, watch the rings spread to whoever gets stuck first — and check the pins to see if the tower guessed it early.
Live from the engine: 8 frozen calls pinned · 9 matured of 105 — 9 graded (2 hit) — the track record has begun
Who feels it first — Cobalt:
Producers GLNCY benefit; consumers pressured: Battery-cathode + aerospace-superalloy users face input pressure.
Directional thesis, not a graded call — the graded calls are the pins.
Geography is descriptive (region centroids; schematic routes). Layers 5-7 are joined at render time from the radar / events / exposure / forward feeds — the map adds no numbers of its own.
A parallel engine that reuses the whole freeze/grade/neural-mesh substrate but points it at ONE tight, high-value niche: the EARLY signals of battery & magnet-mineral supply tightening and regional decoupling — resource nationalism, export controls, stockpiling, trade-route reversals — before the broader market prices them in.
Why this exists — in plain English, and for the kids
This is an early-warning system for the raw materials modern life runs on — the rare earths, lithium, cobalt, and graphite inside batteries, magnets, and electronics. A handful of countries control most of the supply. This engine watches for the moment one of them starts keeping its share — restricting exports, stockpiling, or routing it only to allies — which usually happens quietly, months before prices and headlines react.
The goal is discipline, not hot takes: every warning is frozen as a specific, measurable call, then graded against a dumb “nothing changed” baseline so we can prove the engine actually saw it coming. Those graded calls pile up into an auditable Risk Ledger. This is the lab's first attempt to point its whole freeze-and-grade machinery at a real, valuable job in the outside world.
Think about the special metals and minerals hidden inside phone batteries, electric cars, and fridge magnets. A few countries have most of them. This is a lookout tower that watches for when a country quietly starts hoarding its share instead of sharing — kind of like a kid sneaking all the best LEGO pieces into their own pile before anyone notices.
Here's the trick: it writes down “I think THIS metal is about to get hard to get” before it happens, then checks later if it was right. Getting it right over and over is how it earns trust.
The Story Map — watch how one material becomes a global risk
China mines about 68% of the world's Rare earths (Nd/Pr/Dy/Tb). Mining is the first step — but it's usually NOT the choke point. Watch what happens next.
For the kids: A treasure spot! About 68% of all the world's magnet metal comes out of the ground right here.
USGS/IEA magnitude (universe registry)
Where it's dug up · The treasure spots
Tap any dot or line on the map to learn what it is.
Same data as the big map above — universe shares, computed radar, exposure thesis, and frozen calls — shown one material and one step at a time. Geography is descriptive (region centroids; schematic routes). Layers 5-7 are joined at render time from the radar / events / exposure / forward feeds — the map adds no numbers of its own.
The Dependency Web — who depends on what, with receipts · 2026-09-09
Showing the 6 most-strained connected materials · 6 cited chokepoint links · 12 recent events · rebuilt from the ledgers every run — the graph is a picture of the evidence, never a source of it. Structure is cited; consequences stay hypotheses until frozen and graded. Why only one origin region? Every USGS-cited chokepoint edge in the ledger today originates there — that concentration IS the finding, not a display error; new regions appear the moment a cited chokepoint earns them.
The thesis (the moat)
The moat is not data (the signals are public and dateable) and not scale — it is an AUDITABLE, frozen-and-graded track record of tightening calls that nobody else keeps. Freeze-means-freeze is the product.
Starting slice: Battery & magnet minerals — the highest-signal-density slice (frequent, public, consequential export actions).
The engine loop — how it gets better over time
Each graded outcome updates the Neural Mesh's trust weights and, when it settles, the Materials Canon — so the next call is made by a slightly better engine.
Material × Region — where processing concentrates
The five primitives start here: each Material's controlling Region. Concentration is the leverage a Restriction signal exploits.
Current watchlist — the top forward concerns
Magnet-material controls widen from tech to finished magnets, squeezing EV-motor + defense buyers before substitutes scale.
Permit throughput dialed down enough to bite Western anode lines that haven't localized.
Defense-munitions supply becomes the visible edge of bloc decoupling.
DRC re-uses export suspension as a price lever, whipsawing battery makers.
Regional decoupling — read
4 materials computed from Comtrade; regionalizing (rising intra-bloc share): Rare earths, Graphite, Antimony, Lithium (carbonate)
Ripple map — who feels it first
Primitive 5: which downstream industries a squeeze in each watched material hits first.
Coverage roadmap
Expansion is EARNED by graded calls in the tight slice, not asserted by widening the universe. These are candidate future slices — explicitly not yet graded.
Research + risk-intelligence scaffold, not investment advice. Chokepoint shares are USGS/IEA-cited magnitudes; radar froth is computed (see provenance); the decoupling index is still a directional scaffold (Phase 5). Evidence-base events are RETROSPECTIVE (the base rate) — the engine's own forward calls live, graded, in the Risk Ledger.