Expansion Stage

E0.77

200 W self-directed, of 30 kW total

Price of a world

12missions

36 missions spent, 3 worlds held

Edition 1 · August 2026

This is the permanent address for Edition 1. It is not edited after publication — corrections are appended here and logged in the archive.

More than 99% of humanity's power beyond Earth orbit is executing instructions written on Earth.

The autonomous fleet is two rovers.

What the number means

The Expansion Stage counts power operating beyond Earth orbit that runs without Earth in the decision loop. Not total power — autonomous power. A radio signal takes twenty minutes to reach Mars on a good day and four hours to reach Neptune, so beyond a certain distance a machine that cannot decide for itself is not exploring. It is waiting.

The scale is logarithmic, and each stage is a thousand times the one below it. E1 is a kilowatt of self-directed hardware — enough to say we can explore without waiting on Earth. E2 is a megawatt, enough to operate a world on its own judgment. Today the figure is 200 watts, which puts us at E0.77: past the first watt, not yet at the first kilowatt.

The second metric asks a different question. The price of a world is how many missions it takes to establish presence somewhere new — total missions spent, divided by the number of bodies where we currently operate hardware. At one mission per world, expansion would be limited only by power. At twelve, we spend an order of magnitude more effort than the ideal before anything starts working.

Neither number is a scorecard for science. A year with a spectacular discovery and no new capability shows a flat line, and that is correct for a measure of capacity.

The arithmetic

E = log₁₀(P_auto) / 3

P_auto
— watts operating without ground in the loop
3
— orders of magnitude between stages

m = M / B

M
— missions spent establishing presence
B
— bodies where we operate hardware now

E = log₁₀(200) / 3 = 2.301 / 3 = 0.77  ·  m = 36 / 3 = 12

What counts, and what does not

A system's electrical power counts if it conducts its primary operations at a body without ground in the loop. Fault protection alone does not count — reacting to a problem is not conducting operations. Autonomy during cruise does not count either, because navigating between worlds is not exploring one.

SystemPowerSelf-directed
Perseverance~100 WYes — drives roughly 90% of its traverse autonomously and selects its own science targets
Curiosity~100 WYes — same capabilities
Mars and lunar orbiters~12 kWNo — ground-sequenced
Voyagers, New Horizons~700 WNo — fault protection only
Probes in transit~14 kWNo

P_auto ≈ 200 W of roughly 30 kW total.

WorldMissions to establishEstablished
Moon~24ALSEP, 1969
Mars~7Viking 1, 1976
Jupiter~5Galileo orbit insertion, 1995

Once a world is established its price is fixed. Supply runs and later traffic cost nothing on this metric — returning to a world you already hold is a sign of success, not a cost.

The record

YearWorlds heldPrice
1970124
198017
200026
201025.5
2026312

The low point was 2010, when we held Mars and Saturn. Saturn took four missions to establish, cheap next to the Moon's twenty-four, and holding a cheap world pulled the average down. Cassini ended in 2017 and Saturn left the count. The price has since more than doubled, because the Moon returned to the portfolio and the Moon was expensive.

What would move it

EventEffect on the Expansion Stage
Europa Clipper operating autonomously at Jupiter from 2030P_auto roughly quadruples
A single kilowatt-class autonomous surface systemCrosses E1
An orbiter granted autonomous observation authorityFirst non-surface entry
Both current rovers endingReturns to zero

Four autonomy milestones in twenty-nine years — Sojourner in 1997, the Mars Exploration Rovers in 2004, AEGIS in 2010, Perseverance in 2021 — all of them on the Martian surface. Nothing in orbit anywhere, nothing beyond Mars, and nothing on the Moon.

Limitations

Data quality, methodological weaknesses and open questions in this edition.

Verified

Generator outputs on Curiosity and Perseverance, near 100 watts each. Voyager output near 225 watts each. Queqiao-2 at 1,350 watts. Published mission catalogs. Perseverance's autonomous traverse fraction and the AEGIS deployment, both documented in Planetary Exploration 3.0.

Estimated

The power subtotals for the Moon, Mars and deep space are assembled from individual mission ratings. They are reliable to an order of magnitude, not to three digits. The Expansion Stage is a logarithmic measure, so an order-of-magnitude estimate still places the stage correctly — but the underlying watt figures should not be quoted as precise.

Under verification

The 2026 operating status of Juno and MAVEN, which affects both metrics. The exact composition of the 22-mission Mars campaign figure, which should be quoted from the source paper rather than reconstructed. The establishment dates used for M — choosing Luna 9 rather than ALSEP as the Moon's establishment point changes its price materially.

Binary autonomy test

The counting rule for autonomy is binary. A system either conducts operations without ground in the loop or it does not, which is crude for a capability that comes in degrees. Ono's group proposes a System Adaptivity Level scale, analogous to Technology Readiness Level, and explicitly leaves its definition as future work. When that scale exists, this metric should adopt it and the whole series should be recomputed.

Baseline year

This edition establishes 2026 as the baseline year. Constructing a defensible historical series for the Expansion Stage would require verifying the operating status and autonomy level of every past mission, and estimating it would undermine the point. Future editions build forward from here.

Exclusions

Earth orbit is excluded throughout. Satellites represent roughly 75 megawatts, far more than everything beyond Earth orbit combined, but they constitute infrastructure rather than expansion and would dominate the measurement. Also excluded: scientific return, funding, political will, human spaceflight, mission reliability and data volume.

Sources