Settlement & Population

The crossing

A Base Becomes a Settlement When It Can No Longer Be Emptied

The threshold is arithmetic, the decision is a procurement choice, and nobody announces either.

Everything about a Martian outpost is reversible while every person on it can be brought home. Cancel the program, fly them back, leave the hardware. The day the population exceeds the seats available in the next departure window, that stops being true — and ending the program from then on means deciding what to do about people who cannot be retrieved.

The crossing is not marked because it is not an event. It is the absence of one: the second ferry that was not ordered, the tank farm that was not doubled, the extra window of node consumables that got trimmed from a budget. The first irreversible act of the Mars program is a decision not to buy something, and it will be taken by people who are thinking about cost rather than about sovereignty.

It also arrives earlier than the milestone everyone watches for. The first child born on Mars changes what the settlement is; the first adult who cannot be brought home changes what Earth can do about it, and that comes first. Evacuation capacity is the honest measure of a program’s reversibility, and it is the one nobody publishes.

Two clocks

Evacuation Has Two Stages Running on Different Clocks

A settlement can be made safe long before it can be emptied, and the gap is up to 26 months long.

Getting off Mars is two separate problems. Surface to orbital node is continuous and propellant-limited: a ferry cycles as often as the tank farm allows, so evacuation capacity to orbit is a smooth function of stored propellant. Node to Earth is quantized to the 25.6-month synodic period and limited by vehicles, so departure happens on a schedule the planet sets.

The third row is the one nobody has costed. Twenty people held at a node for a full window is roughly 100 t of consumables open loop, or about 40 t with station-class closure — comparable to the ascent propellant that got them off the surface in the first place. It is the difference between a settlement that can get its people off the ground and one that can actually send them home.

It will be the first thing cut, and cutting it changes nothing visible until it matters. A program can trim node provisioning and pass every review, because the capability it removes is only exercised in the scenario nobody plans for.

The food wall

Food Binds Before Water or Air, and It Sets the Ceiling

Close water and oxygen and roughly 80% of what remains in the manifest is groceries.

At a realistic import budget of 500 to 2,000 t per window, shipping food caps the settlement at roughly 300 to 600 residents. Growing it moves the ceiling by an order of magnitude. That makes agriculture the binding closure item for a settlement — not water, not oxygen, both of which close earlier and more cheaply.

It also reverses a finding that holds everywhere else in this reference. For a mission with an end date, growing food never pays inside five years: the mass crossover sits at 3.0–3.4 person-years and a mandatory contingency store defeats it anyway. For a settlement, there is no end date and no resupply guarantee. Long-stay Mars is the first architecture in the series where a growth chamber is a production facility rather than a research instrument, and it arrives for scheduling reasons rather than for scientific ones.

Two gaps here are real and unfilled. The mass of an agricultural system — structure, growing media, water handling, crop-failure margin — has never been modeled, and lighting dominates only the energy side of the ledger. And crop failure as a settlement-scale risk has no treatment anywhere: a monoculture failure with no resupply inside 26 months is a different class of event from hardware failure, and the redundancy discipline applied to reactors and ISRU trains has never been applied to the farm.

Power per resident

Every Resident Carries About Ten to Twenty-Five Kilowatts

And food lighting is the largest single term, which is why the megawatt threshold arrives early.

Feeding one person is 2,500 kcal a day, or 10.5 MJ. At 1–3% light-to-edible photosynthetic efficiency and 50–70% LED wall-plug efficiency, that is 4–12 kW continuous per resident — derived from physics rather than from any particular farm design, and therefore hard to argue down. Add return propellant at 2.2–4.7 kW, life support and thermal at 1.5–3.0, and industry at 2–5, and the total is 9.7 to 24.7 kW per resident, continuous.

The correction that matters is in the first row. A propellant-only calculation puts the megawatt threshold at 200–450 residents; counting food, life support and industry puts it at 40–103. The threshold arrives about four times earlier than the propellant argument implies, and a 500-person settlement is a 10 MW program rather than a 1 MW one — a different reactor class entirely from the 40 kWe units now in development. Sunlight does not rescue it, because installed capacity is sized by the global dust storm regardless of average insolation.

The crewed settlement is a farm with a reactor, in the same way the uncrewed outpost is a power plant with an excavator.

Dependencies

Six Things Tie the Settlement to Earth, and Two Are Total

Independence is measurable. Score the dependencies and the answer stops being rhetorical.

A redundancy program’s acceptance criterion is independence, and independence can be scored rather than asserted. Six dependencies carry the weight.

A Mars settlement of any size plausible this century would not survive the loss of Earth. It would survive the loss of Earth’s markets, its politics and its governments — which is a real and valuable form of redundancy, and not the one the advocacy literature claims. It would not survive the loss of Earth’s industrial base, because two of the six dependencies are total and both have decades-long paths to closure.

The economy

Mars Has to Run the Colonial Sequence Backwards

Every terrestrial frontier exported first and became self-sufficient later. Mars must invert that, and the inversion is expensive.

Bulk material never leaves Mars. Ascent costs 4.10 km/s, there is no elevator and no material that makes one, and nothing mined here competes with the same material mined closer to its customer. The plausible export is devices, enabled by the halogen inversion — and the market for them is outward, because from Mars the asteroid belt is closer in energy terms than it is from Earth.

Two problems follow. The belt is a 2050s project at the earliest, so the customer does not exist until well after Mars needs one. And the relationship is circular: Mars names the belt as its export market while the belt names Mars as its depot customer. Neither can be the other’s anchor customer, because each is the other’s prerequisite — and the connecting route is worse than either route from Earth.

Mars is the worst-placed inner-system body for cadence precisely because it is close, and its prospective off-Earth trading partner is worse placed still. So the settlement is funded by transfer rather than by trade for its entire existence within any horizon this reference treats as real. That is not a criticism of the project; it is an accounting fact that should be stated rather than deferred, because a plan financed by transfer and a plan financed by revenue have different failure modes and different political vulnerabilities.

Governance

Autonomy Is Discovered, Not Granted

At 4 to 24 minutes each way, no decision needing a response inside about 48 minutes can be made on Earth.

Light time forecloses remote authority as a matter of physics rather than politics. Any decision with a response horizon shorter than roughly 48 minutes — which is most operational decisions and all emergency ones — is made on Mars, whatever the org chart says. Autonomy is not conferred at a milestone; it is discovered the first time something goes wrong quickly.

Two structural oddities sit underneath, and neither has a terrestrial precedent worth leaning on. Jurisdiction follows the registry, so a settlement assembled from modules registered to different states has overlapping legal regimes with boundaries running through hatchways. And every life-critical system is owned by somebody: a polity in which the oxygen is a corporate asset and the only exit is controlled by the same party has no working terrestrial analogue, because even company towns were bounded by the ability to walk out.

Morphology

One Settlement Is a Single Point of Failure for the Species Argument

The second settlement is not a growth milestone. It is the first moment the acceptance criterion is partially met.

Within a settlement, the layout clusters rather than scatters. The Paschen minimum makes Mars uniquely bad at high-voltage termination, so generation and load group into two or three clusters rather than spreading across a site — an electrical constraint that ends up shaping the town plan.

At the program level the same logic runs one scale up. A redundancy program whose entire off-Earth human population lives at one site has not bought redundancy; it has bought a second basket for eggs that are all still in the first one. The second settlement is the first moment the acceptance criterion is even partially satisfied — and it costs more than the first, because it cannot share the landing pad, the reactor, the ice mine or the machine shop without reacquiring the failure modes it exists to remove.

It should also be chosen differently. A weighted site score and a decorrelation criterion are anti-correlated by construction, so running the same screen twice produces two copies of the same place. The second site maximizes decorrelation subject to viability and will score worse — the full argument, including what a second site does and does not decorrelate, is on Mars Site Selection. The cost of that second settlement remains unpriced everywhere.

Analogies

Why the Historical Comparisons Keep Producing Confident, Contradictory Answers

Every analogy captures one or two features of the Martian case. None captures four.

Mars is the only case that combines all four of: a population that cannot leave, an environment that kills without technology, life-critical systems under external ownership, and a generation born into it. No historical case has more than two. That is why the analogies produce confident and contradictory conclusions — each is reasoning correctly from a case that shares half the structure, and the half it omits is different each time.

Record

Provenance and Data Sources