Falsifiers: What Would Break the Space Forecast

The thesis

Five Claims and a Rule, and What Breaks Each

The thesis states its falsifiers up front. Each destination tests the thesis rather than illustrating it.

Thesis falsifiers
Claim	Falsified if	When we know	Rated
The solar system is one industrial system	Every outpost closes on Earth supply alone; no off-Earth node supplies another	By 2075	Modeled
Constraints, hazards and resources fix what each place becomes	A destination's role flips on a resource no survey has found	Ongoing	Modeled
Duration decides who funds it	An investor outside the small sponsor set carries an asset above $700M through a 20-year payback	Ongoing	Modeled
Almost nothing exports bulk material to a distant buyer	Bulk material from beyond cislunar space sells at a profit on Earth	By 2075	Modeled
Mars is a redundancy program, outside the economic frame	Mars is funded by trade rather than transfer before it is independent	By 2075	Modeled
Launch price fell once; in-space travel did not get cheaper	Heliocentric Δv costs fall the way launch did	2040s	Modeled

The claims

Specific Claims With a Named Test

One row per claim. The earliest resolve inside this decade.

Claim falsifiers
Claim	Falsified if	When we know	Rated
Propellant transfer makes the Moon and Mars affordable	Ship-to-ship cryogenic transfer never closes; outposts stay resupply-fed	Starship tests	Building
Launch cost floors near $100/kg	The floor settles far above it and the step reverts to a slow curve	Block 3 cadence	Projected
Radiator area sizes an orbital data center	Measured AI1 specific power beats the radiator model	2027	Building
Orbital compute is the first large customer	Orbital compute proves uneconomic and the cadence case loses its anchor	2030s	Building
The landed hull is a free parts bin on the Moon and Mars	Cargo hulls fly home and are reused	Starship cargo plans	Modeled
Goal-level autonomy works past the Moon	Mars machines need a human per task despite the 4-24 minute delay	First Mars outpost	Modeled
Discovery rate follows compute spend	No lag correlation between compute on one rung and discovery on the next	2030s	Modeled
Partial gravity permits development	0.38 g proves hostile to development; outposts succeed, settlement fails	Rodent centrifuge, then ring	Modeled
Mars development has no threshold above 0.38 g	A partial-gravity rodent study shows a threshold response above 0.38 g	Rodent centrifuge	Modeled
Food sizes a Mars settlement's power plant	Light-to-food efficiency well above 3%	Ground farm trials	Modeled
Terraforming Mars is an import problem	Mobilizable CO₂ far above about 20 mbar	Mars surveys	Modeled
Mars shares six of nine critical inputs with Earth	Concentrated fissile ore found on Mars	Mars prospecting	Modeled
Political sovereignty can arrive before independence	Independence is reached before the first adult who cannot come home	Before 2075	Modeled
Early population growth is limited by transport and training, not births	A large off-Earth population forms before 2060 outside employer or state training pipelines	Before 2060	Modeled
Radiation policy favors an older workforce	Dose rules, countermeasures or shielding remove age as a factor in who goes	Policy and medicine	Modeled
Repeat rotation dominates early lunar labor	Transport or health effects make repeat lunar tours less attractive than long residence	2030s-2040s	Modeled
Mars favors retention over rotation	Cadence, cost and health improve enough for Mars to run on rotation	2040s-2050s	Modeled
The first birth is a political discontinuity	Institutions treat locally born people as evacuable program participants	First off-Earth birth	Modeled
Scheduled beam access becomes a liquid market	A large array behaves as a single beam and the market becomes a queue	Beam array design	Modeled
Compute migrates off Earth at 15-20% a year for 45 years	The rate falls below 15% before 2050	2040s	Modeled
The plant output ratio sets the build-out rate	The ratio is small for the key processes; the 2075 picture is late	Analogue plant study	Modeled
No second independent human instance by 2075	Mars passes the independence threshold before 2075	2075	Modeled
Funded science doubles as operational infrastructure	A Dark Ages array or lens mission is under construction before an optical clock network	Around 2045	Modeled
Primary mirrors stop near 9 m	A monolithic primary larger than about 9 m flies before 2050	Before 2050	Modeled
Measurement precision binds first	A formation interferometer gets science-grade fringes before picometer metrology flies	2030s-2040s	Modeled
Constellations force astronomy off the ground	Rubin loses under 1% of pixels to satellites and drops no science case	Mid-2030s	Modeled
A biosignature stays ambiguous for decades	A candidate reaches community consensus within five years	On first claim	Modeled
Physics goes to space only for baseline	A first-rank experiment flies for vacuum, cold or free fall alone	Ongoing	Modeled
LISA holds to schedule	LISA slips past 2040 or is cancelled	2035-2040	Building
The microgravity materials market is small	Space manufacturing for Earth customers tops $5B a year before 2040	Before 2040	Modeled
The Mars sample cancellation was structural	A US Mars sample receiving facility is under construction before 2032	Before 2032	Observed
Closed ecology is still a research problem	A human-scale closed system runs three years without atmospheric resupply	Ongoing	Modeled
Proxima b settles the M-dwarf habitable-zone question	The ELT or LIFE finds no atmosphere and the target ranking moves	Before 2050	Modeled
A 0.2c sail is a thin-film problem	Absorptivity below 10^-5 across the Doppler band proves unreachable	Lab, this decade	Modeled

Soonest

The Tests That Report First

AI1 prototype data in 2027 tests the radiator model behind Compute Leaves Earth. Starship's ship-to-ship propellant transfer and the Block 3 catch test the two claims everything beyond low Earth orbit is priced on. Chang'e-7 and Chang'e-8 test the lunar water figures, which rest on one measurement. Tianwen-3's sample return in 2031 tests whether a state program with a fixed date outlasts a commercial one with a moving date.

The experiments behind each row are on the Experiments Board; the dates are on the Timeline.

China

Where China Decides the Answer

Three falsifiers resolve on Chinese hardware first.

Falsifiers resolved by Chinese programs
Claim tested	Program	When	Rated
Lunar polar water grade	Chang'e-7 · CNSA	After its postponed launch	Building
Mars ascent and return without a US program	Tianwen-3 · CNSA	2028 launch, 2031 return	Building
Duration tolerance of a state program	Mengzhou and Lanyue · CMSA	Landing before 2030	Building
Orbital compute economics	Three-Body Computing Constellation · ADA Space	Operating since 2025	Observed

Contested

Deliberately Not Claimed

The load-bearing uncertainties, tagged with the confidence they actually earn.

Modeled

A falsifier that resolves against the thesis revises the horizons. The honest response to a failed claim is to move the year or drop the claim, not to defend it.

Modeled

Some claims cannot fail quickly. Duration tolerance and the one-system claim resolve over decades; they are listed so the reader can watch them, not because a result is near.

Record

Provenance and Data Sources

Confidence: Observed for flown, fielded or measured - Building for hardware and programs in development - Projected for announced plans - Modeled for original analysis
Status date: Q4 2026
China: Chang'e-7; Tianwen-3; Mengzhou and Lanyue; Three-Body Computing Constellation
Weakest link: Claims with no dated test