A Staging Architecture · The Transition Study

The Ladder.

Our trade study compared three destinations for a fusion plant — the Martian surface, the lunar surface, deep space — and something unexpected fell out of the computed table: they are not alternatives. Ordered by the length of their supply line back to Earth, each one is the place where a different set of unproven assumptions can fail cheaply. That ordering is an architecture. This page states it.

Mid-century poster: three horizons rising in sequence — a blue Earth field, a grey lunar plain, a warm Martian ridge — joined by a single thin supply line thinning as it climbs.
Masthead bannerdrop assets/ladder_hero.webprecommended 2000×800 px (wide)
Three horizons, one thinning tether — each rung is where the next one’s assumptions go to be tested.
Claim type
Architecture thesis — an ordering with condition-gates, never a schedule
Built from
The computed deltas — heat, mass, fuel, replacement, autonomy
Principle
Every assumption should meet its test where failing is repairable
Status
Stated July 2026. Things can change; the gates are written so they may.
Rung 0 · Earth

Prove the Machine

The tether: none needed. Everything is local.

No magnetic fusion device has yet produced more energy than it consumes; no breeding blanket has ever been measured doing its job. These are the deepest assumptions under everything we write, and they are retired on Earth or nowhere — by the field’s own machines, not ours. We are a research house; we import the machine as a premise and say so on every page. Earth is where that premise either becomes hardware or takes the whole ladder down with it.

Rung 1 · The Moon

Prove the Plant

The tether: three days, any week of the year.

The Moon is where the off-world plant stops being a paper exercise, precisely because the supply line forgives what the other destinations cannot. The coils that are too heavy to land whole get wound on-site for the first time. The first square kilometres of radiator get made from local metal. And the two most dangerous assumptions get tested with a safety net: if the blanket breeds less tritium than designed — a real possibility the designers themselves would not bet against — a lunar plant simply runs on Earth make-up fuel, for years if need be, until a scheduled hardware swap corrects the design. On Mars that same shortfall is a slow strangulation by orbital mechanics; twenty-six months between resupply windows, with the stockpile decaying as it waits. On the Moon it is a delivery.

The Moon’s one weakness is also its quietest gift. A machine on the Moon does not need our autonomy — Earth can drive it, two and a half seconds away. Which means the control system that Mars will one day bet everything on can run there in shadow: watching, deciding, being overruled, accumulating years of “here is what I would have done” against what human operators actually did — at zero risk to the plant. The venue that least needs autonomy is the only one that can safely rehearse it.

Rung 2 · Mars

Prove the Thesis

The tether: twenty-six months, if the window is kept.

Mars is where the ladder stops being generous. Light-lag makes human operation impossible, so the autonomy must be real. The fuel must be locally sourced, because the tether cannot be trusted. Nothing may be wasted, because nothing can be replaced on demand. Every assumption that arrives at Mars must arrive already retired — that is what the lower rungs are for. And this is where our founding claim finally gets its test: the plant that powers the machines is run by the machines it powers. A loop that closes nowhere else, because everywhere else, Earth is still holding the other end.

Rung 3 · Deep Space

The Rung That Waits

The tether: none. That is the problem.

Our own numbers say the reference design does not survive deep space as built — its water-cooled blanket surrenders most of the plant’s output the moment the radiators are forced hot, and every spare part must fly with the plant at a multiple of its mass. Deep space is not the next rung; it is a different machine, waiting for a redesign this study has already specified in outline. We keep it on the ladder to be honest about where the ladder currently ends.

“Prove the machine on Earth. Prove the plant on the Moon. Prove the thesis on Mars.” The ordering is the finding. Each rung is the cheapest place the next rung’s assumptions can fail — a fuel shortfall that would end a Mars plant is a delivery schedule on the Moon; a control fault that would be fatal under light-lag is a logged disagreement two and a half seconds from an operator.
What the Moon Cannot Prove

The Honest Rung.

We state the weakness ourselves, before anyone else does: the Moon proves the plant, but it cannot prove the thesis. Our founding claim — the power runs the intelligence, the intelligence runs the power — never closes on the Moon, because Earth is close enough to hold the loop open. A world that goes to the Moon and stops has built our plant and skipped our reason for it. The ladder is only an argument if it has a top rung.

And a word about time. This sequence is written in conditions, not in years — each rung begins when the one below has demonstrably retired its assumptions, and not before. We do not know how many decades that spans, and we decline to guess in public. A research house that dates its ladder is selling something. We are checking something.

A standing premise of this page: no magnetic-confinement device has yet demonstrated net energy gain, and the reference plant we analyse is a peer-reviewed design, not built hardware. The ladder is an architecture for testing assumptions in the cheapest order — including, at the bottom, the assumption that fusion works at all. Figures behind every claim here are drawn from our computed trade study (July 2026), awaiting external review.