The five axes, and where each comes from
Each material is scored on five independent failure axes, each drawn from an earlier layer: can’t scale = by-product-dominant (companionality ≥ 66); can’t diversify = one country makes ≥ 60% (production share, or trade concentration where no tonnage exists); can’t recycle = ≤ 10% end-of-life (recycling); demand surging = ≥ ~2.5× by 2040 (demand); no slack = a world market under 50,000 t/year (production).
On reading it: the axes are deliberately coarse thresholds, not a weighted composite — the point is how many independent safeguards are gone at once, not a false-precision index. Two axes (demand, thin market) are blank where the underlying data doesn’t cover a material, shown as n/a rather than assumed. Because hard cutoffs hide metals sitting just under a line, a companion Monte-Carlo page propagates uncertainty through all five axes (20,000 draws) and reports P(hardest) as a gradient — it confirms gallium and germanium are robust and surfaces the borderline cases this binary view erases. Inputs: the five layer JSONs → synthesis.json.