14. Earth’s Initial Blueprints

It must have been a strange thing to watch a planet’s settlements be designed from a conference room on Earth. Before steel met regolith, Earth’s architectural studios and space agencies in the early twenty-first century drafted Mars as if it were a tidy extension of Antarctica just with regolith in place of ice.  Perfectly clean modules, ideal horizons, a tasteful amount of dust in neat little piles. Futuristic. Orangey. 

I do not fault them. The first century of Martian habitat concepts had to be drawn somewhere, and there is a certain courage in drawing first. Later, on the ground, we discovered which lines were load-bearing and which were only pretty. 

I keep some of those early dome plans in my memory. They are wrong in so many places. They are also beautiful. Like 1950s Russian rocket designs.

♦

Across that early twenty-first century design literature, the objectives were largely the same: keep humans alive; keep mass low; keep systems fixable; keep options open. Constraints were familiar: EDL bottlenecks; radiation and dust; thermal swings; 0.38 g ergonomics; thin power margins; comms lag; finite crew hours.

Most Mars designs optimized for 4–6 people with near mid-latitude ice, solar + batteries to start, small fission “maybe,” robots ahead of humans “ideally.” From there, convergence by gravity of reason:

repurpose the lander,

link pre-integrated cylinders,

expand with softgoods,

bury in regolith or print shells,

go underground if geology smiles.

Different logos. Same toolkit. I would grade it B+. Not a bad start, but not a good enough solution.

♦

The “stay-where-you-land” approach was the bravest kind of thrift. Turn a ship into a shelter; add a porch that doubles as an airlock; tie on another ship later. First night solved elegantly: no new pressure hull, known avionics, minimum interfaces.

On a whiteboard this stacks neatly; on a planet it stalls. Lander hulls are avionics-dense and maintenance-hungry, their geometry indifferent to streets, shops, or scale.

You can live in your vehicle. You should not build a town out of it. Just ask anyone who has been to the Earth’s twenty-third century ‘Burning Man’ festival.  Hundreds of years of repetition of that tradition has not solved the dust when living in a van.

On the Moon we learned this the gentle way: by trying to add a third and fourth porch and discovering all the doors in the wrong places.

♦

Then came the stationeers — designs that took 21st Century ISS logic and rotated it onto a surface. Cylinders or prisms, Ø 3–5 m, length 6–12 m, delivered hot and linked by pressure-rated corridors. Honest engineering: serviceable racks, clean cable runs, known leaks and known fixes. Early compounds grew like tidily connected bubble-shaped railcars.

But the ground is not orbit. Berms swallowed mass; radiators wanted more area than drawings allowed; turning radii and pallet lanes arrived late; each new module was another bespoke object whose spares multiplied faster than the crew.

Reliability was high. Repeatability, less so.

♦

Inflatable “expandables” added the correct provocation: volume per kilogram matters. Softgoods bladders around rigid cores, pressurized, then bermed or shelled — dignified space to live, think, and repair.

The hazard was operational, not theoretical. Until covered, softgoods compete with crew time and nerves; penetrations and seals multiply; the exterior must be made safe quickly by machines or the interior remains a promise.

The best concepts paired volume with a sheltering strategy that did not assume spare Thursols and Frisols at the end of the week. Where they erred, they imagined weekends on Mars.

There were no weekends on Mars in that first century. Only sols — and survival.

♦

The early designs printing with regolith — arches, vaults, sometimes domes — pointed persistently in the right direction. Shield equals structure. Dirt is both mass and masonry. Place a thin liner inside and let the shell do two jobs. Nice. Elon would approve.

The renderings were not the problem; throughput was. Printers want steady power (MW-class intent), feedstock prep (crush, screen, dry, mix), and takt. A planet is not built by a sculpture garden of one-offs; it is built by a yard that repeats the same bay again tomorrow.

Early proposals offered excellent geometry and weak factories. We did not reject them when we built on Mars. We industrialized the parts of them we could. But with those alone you could not build a settlement.

♦

Early 21st century subsurface Martian design schemes — lava tubes and crater caves — were the romance of geology: superb shielding, thermal grace, entire cities tucked into the planet’s bones. On maps, irresistible; on schedules, hostages to survey and luck. You cannot confidentially plan an independent civilization size settlement to the discovery rate of perfect cylindrical caves that have heavy wheel access.

The lesson from Mars’ second century was not to spurn the ground’s gifts, but to refuse dependence on them initially. Settlement must be location-agnostic first, opportunistic second. Mars permits beauty as a side effect of discipline. Not the other way around.

♦

And then we have the domes.  These pre-Mars landing pretty PR-friendly designs returned again and again — the silhouette that launched hundreds of early twenty-first century Mars books, and thousands of posters. 

Structurally fine; industrially awkward. Seams in dust, cranes in wind, staging lanes that vanish under berms, interiors that curve away from the rectilinear logic of tools and racks. A dome design flatters a skyline; a straight aisle flatters a maintenance shift. Most early plans chose the skyline on paper and rediscovered the aisle in rehearsal. What was easy on Earth, was untenable on Mars until materials engineering caught up a century too late for those who first stepped onto Mars.

♦

Outside the shells, in these early Mars building design systems converged, but not in a good way. 

Interfaces proliferated — berthing rings of many diameters, latch patterns, gaskets — and so did failure modes. Airlocks settled into a pattern: one main, one suitlock, dust-purge zones, sacrificial filters. ECLSS borrowed ISS closure — 70–95% water loops, CO₂ removal, O₂ recovery — with cartridge swaps sometimes too clever to do in gloves. All very twenty-first century idealized architects thinking, when what we actually need was something that passed an operating engineers tests.

The early designs for power came in sensible layers: deployable solar + batteries; fission optional at 10–40 kWe; radiators sized for thin air. Robotics were everywhere in prose — survey, haul, berm, print — yet rarely in numbers: bots measured in renderings per paper, not tonnes per sol, meters printed per sol, failures isolated per sol.

We would correct that omission later — with clipboards and hard seasons.

♦

There was one voice, early and persistent, that repeated a bracing simplicity: manufacture propellant there; stop shipping everything from Earth; use the atmosphere and ice; go direct, go lean. We owed a great deal to that insistence.

When the first methane–oxygen burned in a locally made engine bell, some of us cheered, and some of us ran leak checks twice; I did both. With the benefit of hindsight, I suspect Robert Zubrin would have been pleased by what followed — not because it copied his plan directly (we didn’t in most major ways), but because it honored his premise. 

On Mars, you live off Mars or you do not live alone for long.

♦

What these early design ideas for Mars’ settlement prior to humans landing shared — lander shells, station modules, expandables, printed arches, subsurface dreams, bermed vaults, roving camps — was coherence at expedition scale and fragility at civilization scale.

Designers optimized for arrival, not repetition; for the first 4–16 people, not the next 100s and 1000s; for pretty site plans, not utility backbones; for “robots ahead of humans” as a slogan, not a throughput budget.

Martian’s earliest wannabe architects solved a noble, tractable problem: keep a small crew alive beautifully. The harder problem — the one that turns survival into society — went mostly unasked on paper and answered later in dust. The question of installed cost per m² had to have an answer two orders of magnitude better. We learned to ask that question. And when we got to an answer, it changed everything.

♦

I learned early to ask a rude question of every settlement design drawing: what, exactly, are we optimizing? Question the requirement first as the old saying goes.

Earth’s early twenty-first century plans usually answered with virtue — crew safety, reliability, mass thrift. All noble. But the hidden objective function was “survive the first season with six people,” not “repeat this a thousand times without thinking.” Independence level one solutions, crew sorties and temporary base camps.

Budgets were sized for living, not for building; power for lights and lungs, not for factories; robots promised, not provisioned. The colony that exists on a slide is a picture. The one that survives in dust is a rate: tonnes moved per sol, meters printed per sol, sealed ports per sol, failures isolated per sol.

When we began to measure like that, half the beautiful ideas changed color. They remained beautiful. They stopped being plans.

♦

Habitat skins were the first area to sport the structural problems with these early designs. Rigid shells arrived in many alloys and more adjectives; inflatables arrived folded like hope; liners threaded between them like penance. Each added penetrations, seals, fasteners — and failure probability math.

On paper you can tolerate many envelope families; in a yard you cannot maintain them. The fix was not exotic: standardize the skin, shrink geometries, stop inventing a new gasket for every nicer window. Softgoods are legitimate volume wins — if a machine, not a crew, can bury or shell them the same sol. Rigid shells are serviceable — if their details repeat like a product line, not a portfolio.

Every new curvature is a spare part that will go missing at the worst hour. Ask me how I know, and I will show you a thousand examples from those first decades.

♦

Shielding was where aesthetics most often beat arithmetic in early 21st century settlement designs. Berms, blocks, printed vaults, subsurface cuts — each can work. The decisive variables are “how fast” and “with what.” Berms shoveled by people are theatre. Printing without feedstock is sculpture.

Most proposals carried honest numbers for life support (10–50 kWe) and then assumed the same plant could run a print farm. It cannot. Regolith handling wants crushers, screens, dryers, mixers; printers want steady power and redundancy; heat rejection wants area. Without MW-class intent (delivered or staged), “print the city” collapses into “print the demo.”

Direction right: let the shield be the structure. Missing: the factory that makes the shield.

♦

Doors and interfaces decided fates. Agencies loved their ports; architects their corridors; hobbyists whatever coupler fit last year’s render. We would have had to pay for that romance in adapter rings and incompatible latches. That made them non-starters.

A Martian self-sustaining city is a network effect wearing a pressure suit. One pressure-port standard, one corridor section, one utility manifold spec — these are not constraints; they are multipliers. With them, berthing becomes plug-and-live. Without them, every join is an integration event.

Site plans suffered similarly. Pretty clusters hid missing streets: no turning radii for haul bots, no pallet lanes to heavy users, no overhead clearance where cranes actually swing. When your power trunk is an afterthought, your outages are a prophecy.

♦

Life support exposed another inversion. The early 21st century design literature spoke of partial closure, cartridge swaps, robust racks — the right themes. The details betrayed orbital habits: hidden plumbing, tight clearances, software states legible only to mission control, swap procedures written as if gloves were optional.

Closure is not a virtue if it creates non-serviceable complexity. The survivable kind puts every consumable on a predictable cadence, every filter where a tired human can reach, every pump with a bypass you can understand in a bad hour. Earth liked to fly one perfect rack. Mars required a thousand imperfect spares, stored locally, that fit anyway.

That difference is philosophy more than hardware. It is also survival more than philosophy.

♦

Robotics were promised like weather. “Robots ahead of humans” appeared in every early twenty-first century Mars Settlement Plan executive summary and in suspiciously few equipment lists. Fleets were described as if naming them summoned them: graders, haulers, printers, cranes, inspector-bots.

What mattered was throughput and resilience: tonnes moved per sol per hauler; meters of vault per sol per printer; battery swap intervals; dust-hard seals; spares at 10–15% of fleet; a depot where a limping bot could arrive and not die in a drift.

Construction wants takt — repeatable cells with mini-factories: grade → shape → place → seal → test → handover. If step three cannot be done by a machine on a Tuesol night in a storm, step four will not happen on a calm Wedsol. Schedules that omit this are fiction.

♦

Scaling exposed the last, most expensive and fatal illusion of early Mars settlement designs. Designs that held nicely from four to six people fractured at sixty and totally disintegrated at six hundred. Corridors too narrow became fire codes by accident. Airlocks sized for one suit became queues that burned crew time. Power plants framed for “habitat plus margin” starved the print yard and the mine.

The heroic fix was always “add more of what we have,” which is how you drift into bespoke sprawl: more module types, more interfaces, more spares, more training overhead. In orbit, adding one more special is a solved problem. On a planet, it is an invoice that never stops arriving.

The cure was not to dream smaller. It was to design for the second order — the moment the tenth copy arrives and nothing about the process changes.

♦

If this sounds unkind to Earth’s pre-arrival design drawings of Mars’ cities, let me correct the record: we needed them. They proved you could live; they taught why you could not yet grow. The amateurs with elegant renders, the big-name PR-savvy architect studios with tasteful politically correct eco-domes, the cautious space agencies with rituals of over-complexified redundancy — they all pushed the conversation forward. We especially commend Zubrin’s Mars Society as the catalyst-leaders they were.

We borrowed their materials, stole their better joints, and retired their worst assumptions with gratitude. 

The problem was never that early Mars city designs were foolish. The problem was that they optimized the wrong problem: the expedition, not the enterprise. They imagined a picture; we on Mars required a scalable process.

And in that gap lay two major constraints that would come to rule everything we built — obvious, once named, you will wonder how the early designs ever forgot them.

The naming came later, after a few hard winters, and not on paper.

The Canonical Designs of Early/Mid-Century Martian Settlements