Mars Colonization

Mars colonization means people living on Mars in hardware they brought or built: shielded from a measured radiation field, breathing recycled air, eating food grown or shipped, making propellant from the atmosphere, and doing all of it with no evacuation option for up to five hundred days at a time. The framing under test is that Mars is a plausible second home for humanity.

This slot has the richest measured record of any settlement subject and the reason is worth stating up front: the radiation dose has been measured, in transit and on the surface, by an instrument that flew rather than by a model; the soil chemistry has been measured by a wet-chemistry laboratory that landed; oxygen production from the atmosphere has been demonstrated, sixteen times, with a published gram count; the transport's flight record is public and countable; and the programme cost has been independently re-derived by two methods that converge. The correct posture for this page is markedly more established than the popular treatment, and precisely because of that, harder on the settlement claim. The boundary has moved from “we do not know” to “we know, and the number is above the limit”.

Three seams are honoured strictly. Artificial Gravity owns everything about what gravity does to a body — bone, muscle, vestibular tolerance, the AGBRESA trial, partial-gravity animal data, the modelled health floor — and this brief carries one paragraph and routes. Historical Space Colonization Concepts owns Biosphere 2 in full, including the oxygen decline and what the episode did and did not establish; this brief takes only the present state of the art in closure. And Terraforming owns any claim about changing the planet: nothing in this brief assumes or requires terraforming, and the two subjects are separated by orders of magnitude in both energy and time.

Established Start with the only in-situ resource utilisation ever demonstrated off Earth, because it is genuinely remarkable and because its own authors state its scale factor. The Mars Oxygen In-Situ Resource Utilization Experiment ran aboard Perseverance from 20 April 2021 to a final run on 7 August 2023: 122 grams of oxygen total, across 16 runs, at a peak rate of 12 g/hour — twice NASA's original goal — at a purity of 98% or better. Established The peer-reviewed account is more useful than the press release. Hoffman, Hecht and colleagues, covering the first seven operational cycles over sols 60–276, report nominal production of 6–8 g/hour from a 55 g/hour carbon dioxide intake, through a solid-oxide electrolysis stack of ten cells in series with a centre tap, a scandia-stabilised zirconia electrolyte and a nickel-based catalysed cathode at 800 °C. Atmospheric acquisition used a HEPA dust filter and a scroll compressor limited to 3,500 rpm with passive viscous flow control rather than true regulators, producing about 5% cathode pressure oscillation at the compressor frequency. Degradation is characterised: area-specific resistance rose from below 0.75 to about 0.9 Ω·cm² during burn-in and then “increases slightly with every cycle” from differential thermal stress, with the stack projected to meet requirements for more than 60 cycles. Established And the number the brief exists to carry: the authors state that a full-scale Mars propellant plant must be “hundreds of times larger”, producing 2–3 kg/hour against MOXIE's 6–8 g/hour — a factor of roughly 250 to 500. Their own list of what remains unproven is equally explicit: long-term continuous durability, thousands of hours of running rather than intermittent testing, autonomous control of a scaled system, and response to daily and seasonal atmospheric variation.

Frontier What MOXIE licenses and does not license should be stated flatly. It licenses: oxygen can be made from the Martian atmosphere by a flight-qualified device, and the electrochemistry, the degradation mode and the purity are known. It does not license: propellant ISRU at mission scale is solved. Established MOXIE made oxygen. It did not make methane, did not liquefy, did not store, and did not run continuously. Every Mars return architecture also needs methane by Sabatier reaction from CO₂ and hydrogen, which needs water — either brought as hydrogen or mined from subsurface ice — and no Martian water extraction has been demonstrated. NASA's 2005 ISRU roadmap put a Mars water extraction demonstration at 2013 and rated the capability “just initiated”; it is still not flown. Cryogenic liquefaction and storage of the product is a further unflown capability that MOXIE did not attempt.

Established The radiation record is the strongest measured block in the subject and it is not a model. The Radiation Assessment Detector aboard Mars Science Laboratory measured 1.84 ± 0.3 mSv/day dose-equivalent during the 253-day cruise, giving about 0.66 Sv for a round trip with current propulsion, and the instrument team's comment on shielding is unusually blunt: “even an aluminum hull a foot thick wouldn't change the dose very much.” On the surface at Gale Crater over 300 sols the same instrument measured 0.210 ± 0.040 mGy/day absorbed dose and 0.64 ± 0.12 mSv/day dose equivalent, at a mean quality factor of Q = 3.1 ± 0.3 from measured linear energy transfer, with unshielded organ exposure around 520 µSv/day falling to about 380 µSv/day with body self-shielding. Established The mission total is about 1.01 Sv for a round trip with 180-day transits each way plus 500 days on the surface.

Established Now the number it has to be compared against, which the popular treatment almost never supplies. The US National Academies' 2021 report Space Radiation and Astronaut Health recommended a universal career limit of about 600 mSv, based on risk of exposure-induced death for a 35-year-old female, replacing NASA's age- and sex-varying 3% REID standard with a single limit. The same report states that astronauts on long-duration Mars missions “are likely to be exposed to radiation levels that significantly exceed the proposed new standard” and that unless protection improves, “NASA would need to seek waivers to the radiation health standard to pursue these missions.” Established State the comparison arithmetically because it is the cleanest fact on the page: a measured 1.01 Sv mission against a recommended 600 mSv career limit is 1.7 career limits in a single flight. The instrument team's independent risk framing is that 1 Sv is associated with roughly a 5% increase in fatal cancer risk.

Established And the finding that stops shielding being simply a matter of adding mass. NASA's Advanced Radiation Protection Thick GCR Shielding Project reports that “a minimum in the dose equivalent versus aluminum shielding thickness may exist in the 20–30 g/cm² region”, beyond which dose rises, because thick shielding generates secondary neutrons and light particles. The project frames this explicitly against the conventional assumption that “increasing shielding thickness will decrease risk to crew health.” Interest running against the finding: this is NASA's own shielding programme reporting that its own primary lever has a floor and then reverses. Frontier Hydrogen-rich materials behave differently and the free-space areal densities — 6–7 t/m² of polyethylene, 10–11 t/m² of regolith for a 20 mSv/yr target — belong to O'Neill Cylinders. What this brief carries is the Mars figure: reducing surface dose “to levels as existing on Earth” requires habitat shielding of several hundred g/cm², which is a mass problem measured in tonnes per square metre of habitat roof.

Established The soil has been measured and the number is specific. Hecht and colleagues, from the Wet Chemistry Laboratory aboard Phoenix at Vastitas Borealis, report 0.4–0.6% perchlorate by mass, total dissolved salts around 10 mM, a pH of 7.7 ± 0.5 consistent with carbonate buffering, and cations dominated by magnesium and sodium. Established Note carefully what that paper does and does not say: it reports the chemistry and does not address habitability implications or potential human use of the soil. The toxicological inference is made elsewhere and should be attributed elsewhere. Frontier Wang and colleagues, in GeoHealth (2025), make it: a mean grain size possibly as small as 3 µm — small enough to pass lung defences and enter the bloodstream — with constituents of concern including silicates, iron oxides, beryllium, arsenic and perchlorates, and predicted effects including chronic respiratory disease, irreversible silicosis, and thyroid disease and severe anaemia from perchlorate. Their proposed countermeasures are iodine supplementation, habitat filtration and pre-mission preparation. Frontier The flags have to split here. The chemistry is established; the human dose–response is frontier and rests on inference from terrestrial mineral dusts, because nobody has been exposed. The authors' own caveat is that substantially less is known about Martian dust's health impacts than about lunar dust's — and the lunar figure itself rests on a ground-milled proxy whose representativeness is, in its source's words, “at present unknown”.

Established Landing is the starkest single engineering gap in the subject and the current popular treatment does not contain it. Munk and Cianciolo at NASA Langley: Viking-heritage entry, descent and landing technology — essentially unchanged since the 1960s and 70s — lands up to about one metric tonne on Mars. Human missions require landed payloads of 20 to 40 tonnes; human precursors need 5 to 10 tonnes of landed usable payload. That is a factor of twenty to forty against the largest mass ever landed on Mars. The candidate technologies are named with their maturity: supersonic retropropulsion at “a low Technology Readiness Level” because plume interactions are poorly characterised and stability effects unstudied; hypersonic inflatable aerodynamic decelerators, about 20% less massive than rigid aeroshells but with their own development challenges; and mid-L/D rigid aeroshells, higher TRL and heavier. The operative statement is that these methods “have not yet been demonstrated, and are not yet planned in future Mars missions.” Frontier What has changed since is worth stating fairly: supersonic retropropulsion is now routine in Earth's atmosphere on Falcon 9 booster entries, so the TRL statement should be read as dated rather than wrong. What has not changed is that nothing above about a tonne has been landed on Mars and no Mars-specific demonstration is manifested.

Established Closure is the wall, and the literature answers the question precisely rather than vaguely. In orbit, the ISS reached 98% water recovery in June 2023 through the Water Processor Assembly, the Urine Processor Assembly with distillation, the Brine Processor Assembly and advanced dehumidifiers capturing crew breath and sweat — against NASA's own stated requirement that beyond-LEO life support “need[s] to recover close to 98% of the water”. For water, the requirement is met. Established On the ground, with humans, closed, the record is a ladder: BIOS-1, one crew, 90 days, 90% gas closure; BIOS-3 in 1972, one to three crew, 180 days, 91%; BIOS-3 in 1977, 78–81%; BIOS-3 in 1983–84, two crew, five months, 95.4%; BIO-Plex Phase III, four crew, 90 days, about 75% air and 100% water; the Biosphere 2 Test Module, 100% closure for over 60 person-days; and Lunar Palace 1, three crew, 105 days, 60% of food with 100% of oxygen and water. The longest human closure on record is six months, in BIOS-3, with one to three occupants. Established And the explicit statement of what has never been done, from Porterfield and colleagues in npj Microgravity (2025): “to date, no nation has demonstrated a completely closed BLiSS system that integrates or even includes the essential requirements for regenerative operation, feedstock and nutrient recycling, and human waste processing.” The same paper notes that Lunar Palace 1 ran a crew of four for an entire year on closed atmosphere, water and nutrition and yet “failed to close the loop on waste recycling”; that MELiSSA is “the longest continuously running BLiSS research program in the world” but “never approached closed-systems human testing”; that waste recycling remains the most significant technical challenge; and that four to eight years of operational experience would be needed to reach full technology readiness for a fully bioregenerative habitat. Established Set that against the mission: a conjunction-class flight is roughly 900 days. The gap is not marginal — it is a factor of two to three in duration and a category difference in what “closed” means. For Biosphere 2 as programme history, see Historical Space Colonization Concepts; only the Test Module datum belongs here.

Established The transport has a countable flight record and an unattempted gating milestone. Through Flight 12 on 22 May 2026, Starship has flown 12 times with 7 successes and 5 failures or partial failures: total losses on Flights 1 and 2; a partial on Flight 3 with the booster under control and the ship lost at entry; the first tower catch on Flight 4; a booster catch on Flight 5; partials on Flights 6 and 7; then successes on Flights 8 through 11, with Flight 11 the last Block 2 vehicle; and Flight 12 the Block 3 debut with a Starlink deploy. Established The capability every Mars architecture depends on most has never been attempted by anybody. Ship-to-ship cryogenic propellant transfer — the same capability that gates the lunar Human Landing System — has not been flown. Flight 3 in March 2024 demonstrated transfer between two tanks on the same vehicle, which is a different thing. The demonstration “is expected to occur in 2026”, boil-off mitigation by insulation and vacuum jacketing appeared on the Block 2 S33 vehicle, and approximately ten tanker launches are required to fill a depot sufficiently for one lunar mission.

Established The cost estimates differ by a factor of ten and the spread is the finding, not either number. Zubrin's Mars Direct, as published: an Ares booster placing 121 t into a 300 km orbit and boosting 47 t toward Mars; two launches per mission with the Earth Return Vehicle first and the Mars Habitat Unit 26 months later; a crew of four; six-month transits; an 18-month surface stay; and an ISRU core in which 8 tonnes of hydrogen brought from Earth are reacted with atmospheric CO₂ over about ten months using a small nuclear reactor to produce up to 112 t of methane and oxygen against a 96 t requirement — a 14:1 mass leverage on the imported hydrogen. Mars Semi-Direct was costed at $55 billion over ten years; Zubrin's 1991 figure was $42 billion. Established That figure sits in a NASA-authored table beside two independent re-derivations. Jones at NASA Ames compiles the published estimates — SEI's 90-Day Study at $517 billion (1989), Zubrin at $42 billion (1991), Hunt and van Pelt at $53 billion (2002), a VSE analysis at $250–300 billion (2005), Price and colleagues at $110–120 billion, the NRC at $300–600 billion (2014), Garver and Sommerer both at $500 billion — and then re-derives the half-trillion figure two ways. Mass-based, using ISS cost data of $150 billion for 420 tonnes in orbit, or $0.36 billion per tonne, against a Mars mission needing 900–1,300 tonnes: $320–460 billion. Life-cycle, using GAO historical ISS data: $275–398 billion, split 73% development, 18% launch, 9% operations. His conclusion is that the estimate persists because it is “well established and widely known” and converges across methods, while the low estimates describe “austere and minimal missions”. Established His caution cuts both ways and this brief prints it: “thinking that we can do much better could produce very over-optimistic cost estimates”, and “we can hope but not plan that new ideas or new approaches can get humans to Mars for less than half a trillion dollars.” Print all three numbers — $42B, $275–398B and $320–460B — label the first advocacy and the other two independent parametric re-derivation from ISS outturn, and do not average them.

Established Planetary protection for crewed Mars is not a strict policy; it is an expired one. The COSPAR Panel on Planetary Protection heard NASA's Planetary Protection Officer in Cologne on 15 April 2025: NASA is “working alongside and enabling stakeholders... to develop a balanced set of Mars implementation requirements” and states “We don't have all the answers just yet”; work is “initiating” to establish an agency risk posture for forward and backward contamination; a “first order biological contamination model based on a well documented architecture and operational scenarios” is in development, meaning the microbial release from a crewed vehicle has not been quantified; and NASA Interim Directive 8715.129 expired on 30 September 2025 with permanent procedural requirements still in conversion. Frontier This matters more than it reads. A crewed vehicle vents, carries kilograms of human-associated microbial biomass, and wants to land where subsurface ice makes ISRU feedstock attractive — which is exactly where Special Regions are. The resource and the contamination hazard are in the same places. Established And the calibration datum for Mars operations generally: the Mars Sample Return Independent Review Board found in September 2023 a lifecycle cost of $8–11 billion, “a near zero probability” that the main elements would launch in 2027–28, and that “MSR was established with unrealistic budget and schedule expectations from the beginning”. A robotic mission to return a few hundred grams under containment is costing that much and has slipped indefinitely.

Frontier Finally, the feedstock, which is at the same epistemic stage as lunar polar ice and for the same reason. The SWIM project — Morgan, Putzig, Perry and colleagues — combined thermal analysis, subsurface radar and geomorphic assessment into ice-consistency maps and found that Arcadia Planitia and Deuteronilus Mensae match “the greatest number of remote-sensing criteria for accessible ice-rich, subsurface material”. Established What that establishes is candidate regions from orbit. What it does not establish is depth to ice, purity, overburden hardness, or extraction rate at any specific site. The resource is mapped and not measured, exactly as at the lunar poles, and the architectural consequence is identical: every Mars plan that needs hydrogen from local water rests on remote sensing. Zubrin's architecture sidesteps this by importing eight tonnes of hydrogen; every architecture that does not import hydrogen inherits the unverified premise. Lunar Industry states the same distinction on the other body, and the parallel is the strongest structural connection in this cluster.

Speculative Position one is the framing itself: Mars is a plausible second home for humanity. Held by Zubrin and the Mars Society, by SpaceX, and by most of the settlement literature. For it: a demonstrated atmospheric ISRU device, mapped subsurface ice, a transport vehicle in flight test, and a surface radiation number that is survivable with shielding. Against it: 1.01 Sv against a recommended 600 mSv career limit, a twenty-to-forty-fold landing-mass gap, no closed ecology ever run at human scale for the mission duration, and no human ever exposed to 0.38 g. The honest split is speculative on “second home” and frontier on “crewed missions”, and collapsing the two is the single commonest error in the subject.

Frontier Position two: propellant ISRU is essentially solved by MOXIE. A common popular reading, and its own authors refute it: full scale is “hundreds of times larger” at 2–3 kg/hour against 6–8 g/hour, with long-term durability, continuous operation, autonomous control and seasonal response all “remaining to be demonstrated”. established that oxygen ISRU works off Earth; frontier that propellant ISRU is close. Frontier Position three: radiation is manageable. Several early readings of the RAD results used exactly that word, and there is something to it — 0.64 mSv/day on the surface is not acutely dangerous, and several hundred g/cm² of habitat shielding brings dose to Earth levels. Against it: 1.7 career limits per mission, the National Academies' waiver statement, and NASA's own finding of a dose minimum at 20–30 g/cm² of aluminium beyond which dose rises. The tension between “manageable with enough mass” and “there is an optimum thickness” should be stated rather than resolved.

Frontier Position four: regolith shielding solves the surface radiation problem cheaply. Held implicitly by essentially every surface habitat concept, and it has real support — Mars has unlimited regolith and the cheapest known way to move it into a shielding configuration is bagging, which Moon-Based Manufacturing reports as the top-ranked lunar construction technique for exactly this reason. Against it: the required areal density is “several hundred g/cm²”, and the free-space equivalent is 10–11 t/m² of regolith. A brief that says “cover it with dirt” without giving the areal density has not said anything.

Speculative Position five: closed-loop life support will be ready when it is needed. Implied by every settlement timeline and contradicted by every number: six months is the longest human closure, 95.4% the best, no nation has closed waste, Lunar Palace 1 closed food to 60% over a year, MELiSSA has never run closed with humans, and four to eight years of operational experience is estimated as necessary for full readiness with no facility currently accumulating it. Frontier Position six: the perchlorate problem is soluble. Held by the agricultural-ISRU literature; 0.4–0.6 wt% is a well-characterised concentration and perchlorate is both biologically and chemically reducible. Against it: no demonstration at habitat scale, and no human dose–response data at all.

Frontier Position seven: large-mass Mars entry, descent and landing is solved in principle. Held implicitly by the Starship architecture, and supersonic retropropulsion being routine on Earth-return boosters is genuine support. Against it: nothing above about a tonne has landed on Mars, the requirement is 20–40 t, and the agency's own study calls Mars SRP low-TRL with poorly characterised plume interaction and unstudied stability effects. Frontier Position eight: Mars gravity is sufficient for long-term health. This brief does not adjudicate it. No human has been exposed to 0.38 g; the only partial-gravity data in orbit are in mice and show protection that is system-specific rather than proportional; NASA's own evidence report calls the human response to continuous Mars gravity unknown; and the one quantitative modelling claim puts a health floor at 0.4 g, above Mars gravity. All of that is sourced and argued in Artificial Gravity and this brief cites rather than re-derives it.

Frontier What this brief does own on the physiology question is the compounding, and no source in this pack studies it. A settler faces 0.38 g and 0.64 mSv/day and 3-µm perchlorate-bearing dust and a life-support system that has never run closed at human scale for the duration, simultaneously, with no evacuation option for up to five hundred days because of orbital mechanics. Every published risk assessment treats these separately and the mission does not. Naming that as an unaddressed gap is more honest than assuming the risks are additive, which is itself an assumption nobody has tested.

Frontier Position nine: Mars settlement will be much cheaper than agencies estimate. Held by Zubrin at $42 billion, by Hunt and van Pelt at $53 billion, and by the general commercial-cost-curve argument. Against it: two independent re-derivations converging on $275–460 billion from ISS cost outturn, by a NASA analyst, with the explicit observation that the low estimates describe austere and minimal missions and that “we can hope but not plan” for better. Present the tenfold spread as the finding. Speculative Position ten: stated timelines mean something. The Mars target moved 2018, then 2022, then 2026; on 9 February 2026 SpaceX announced it would prioritise a “self-growing city” on the Moon instead, on the reasoning that “the overriding priority is securing the future of civilization and the Moon is faster”, with Mars work resuming “in the next five to seven years” — the first announcement in the series to state a reduction in priority rather than a new date. Frontier The rule must be applied evenhandedly or it is not a rule. Artemis II moved from 2025 to 2026 and flew; Artemis III moved from 2026 to 2027 to NET 2028. The honest generalisation is that nobody's stated dates in this domain have held, which is a more useful statement than singling out one company, and the slip record is evidence about the reliability of stated timelines rather than about whether Mars settlement is possible.

Frontier Position eleven: planetary protection will be resolved before it binds. Implied by every crewed-Mars plan, and trending negative: the governing interim directive expired on 30 September 2025, the contamination model is in development, the Planetary Protection Officer's own words are “we don't have all the answers just yet”, and a robotic containment mission is costing $8–11 billion with near-zero probability of its stated launch dates. Speculative Position twelve: terraforming makes settlement easier. A persistent conflation in popular treatment. Nothing in this brief depends on it, the two are separated by orders of magnitude in energy and time, and Terraforming owns every claim about changing the planet.

Established The first bottleneck is a dose that exceeds a recommended standard, with a shielding lever that reverses. 1.84 mSv/day in transit, 0.64 mSv/day on the surface, about 1.01 Sv for the mission, against a recommended 600 mSv career limit — and aluminium dose-equivalent bottoms out at 20–30 g/cm² and then rises from secondary production. This is not a bottleneck that more mass removes. Hydrogen-rich shielding and habitat burial move the number in the right direction; the areal densities involved are hundreds of grams per square centimetre, which is tonnes per square metre of roof.

Established The second is closure, and it is a duration problem more than a percentage problem. Six months is the longest human closure ever run; 95.4% is the best closure fraction; the best food closure is 60% over a year, with waste uncontained; no nation has demonstrated a completely closed system including waste processing; and four to eight years of operational experience is the estimated path to full readiness, which nobody is currently accumulating. The mission is 900 days.

Established The third is a landing capability short by a factor of twenty to forty. One tonne demonstrated, 20–40 t required, with the enabling technology called low-TRL by the agency's own study and “not yet planned in future Mars missions”. Established The fourth is a transport whose gating milestone is unattempted: ship-to-ship cryogenic propellant transfer, never flown by anybody, scheduled for 2026, and required for roughly ten tanker flights per mission.

Frontier The fifth is the scale factor on the only ISRU that has ever worked off Earth. 250 to 500 times, by the device's own authors, from 6–8 g/hour intermittent to 2–3 kg/hour continuous for months, autonomously, through dust storms — and MOXIE made only oxygen, with methane, liquefaction and cryogenic storage all unflown. Frontier The sixth is that the water is mapped and not measured. SWIM identifies Arcadia Planitia and Deuteronilus Mensae from orbit; no depth, purity, overburden hardness or extraction rate has been measured in place anywhere on Mars.

Frontier The seventh is a soil that is 0.4–0.6% perchlorate in dust that may average 3 µm and contains beryllium and arsenic, with no human dose–response data and a proposed countermeasure set — iodine supplementation, filtration — that has never been tested against the actual exposure. Frontier And the eighth is money, which is the bottleneck most often assumed away. Two independent re-derivations from ISS cost outturn give $275–460 billion, against advocacy figures a factor of ten lower, in a budget environment where the same agency's auditor calls the lunar campaign unsustainable at $4.2 billion per launch.

Established The adjudication routes this brief through Lunar Industry, and the reason is operational rather than chemical. The Moon is where the delivered-cost record, the surface-operations failure rate, the dust experience and the fission surface power hardware are being generated, and NASA's own architecture sequences Mars behind the lunar segments. Frontier But the limit of the dependency has to be stated in the same breath, because it is routinely overstated: lunar and Martian ISRU share almost no chemistry. Lunar work is high-temperature reduction of silicate rock; Martian ISRU as demonstrated is solid-oxide electrolysis of atmospheric CO₂. What transfers is operations, autonomy, dust handling, thermal design and cost discipline — not process. Lunar ice, lunar oxygen routes and CLPS economics belong to FR-II-01 and are not restated here.

Frontier The second edge, to Terraforming, should be read as a definitional dependency rather than a technical one. This brief defines the settlement problem that terraforming proposes to dissolve: it owns living in hardware on Mars — shielding, closure, dust, perchlorate, entry and descent — and FR-II-17 owns any claim about changing the planet itself: atmospheric pressure, magnetic field, warming, ecopoiesis. Nothing in this brief assumes or requires terraforming, and if terraforming were achievable the constraints here would change category rather than degree, which is exactly what makes the edge worth typing.

Frontier What this brief supplies downstream is a constraint set rather than a capability. Multi-Planetary Civilization owns the civilizational argument — redundancy against existential risk, population thresholds for self-sufficiency, governance of a settlement — and what it inherits from here is the 900-day mission, the 1.01 Sv dose, the closure gap, the 20–40 t landing requirement and the $275–460 billion parametric cost. This brief makes no claim about what a settlement would mean.

Established And two handoffs that are not typed edges but are load-bearing. Orbital propellant depots, cryogenic transfer architecture and in-space assembly belong to Deep Space Infrastructure and Space-Based Manufacturing; this brief states only that ship-to-ship transfer is unattempted and gates the mission. And all rotation and partial-gravity physiology belongs to Artificial Gravity.

Established The single most informative experiment is one nobody is currently funding: a closed ecology run with humans for the mission duration, including waste. The literature is unusually precise about what is missing — regenerative operation, feedstock and nutrient recycling, and human waste processing, none of which any nation has demonstrated in an integrated closed system — and about the cost of finding out, which is four to eight years of continuous operational experience. Frontier That is a facility programme with a schedule, not a research question, and it is the one item in this brief where the required experiment is completely specified and completely unfunded.

Frontier Second: a Martian water extraction demonstration. NASA's 2005 roadmap set one for 2013 and rated the capability “just initiated”. Nothing has flown. What it would establish is exactly what orbital mapping cannot: depth to ice, purity, overburden hardness and achievable extraction rate at a specific site. Without it, every non-hydrogen-importing architecture rests on remote sensing.

Established Third: a MOXIE successor at one to two orders of magnitude larger, running continuously. The authors' own gap list is the experiment specification — thousands of hours of continuous operation rather than intermittent cycles, autonomous control of a scaled stack, and characterised response to daily and seasonal atmospheric variation. Frontier And the rest of the chain has never been assembled anywhere: Sabatier methane synthesis from Martian CO₂ and local water, then liquefaction and long-duration cryogenic storage on the surface.

Frontier Fourth: supersonic retropropulsion in the Martian atmosphere, and a large decelerator. Terrestrial booster entries have retired part of the risk and none of the Mars-specific part — plume interaction with a thin CO₂ atmosphere at Mach numbers no vehicle has flown there, and stability effects the NASA study calls unstudied. Nothing is manifested.

Frontier Fifth: ship-to-ship cryogenic propellant transfer in orbit, which gates both this brief and the lunar programme, has never been attempted, and was scheduled for 2026. Speculative And sixth, the experiment that would resolve the biggest single unknown and cannot currently be run at all: sustained human exposure to 0.38 g. There is no facility, no vehicle and no funded programme; Artificial Gravity owns what would be needed and why the animal data do not substitute.

Established Take the requirements in the order the mission imposes them, and the first is mass through the atmosphere. Twenty to forty tonnes landed against a demonstrated one tonne, using either supersonic retropropulsion at low technology readiness, hypersonic inflatable decelerators about 20% lighter than rigid aeroshells, or a heavier mid-L/D rigid aeroshell. The aeroshell diameter, the deceleration profile and the terminal guidance for a 40-tonne vehicle have no flight heritage of any kind.

Established The second is shielding mass, and the specification is unusual because it is non-monotonic. Bringing surface dose to Earth-equivalent levels requires several hundred g/cm² of habitat shielding; aluminium reaches a dose-equivalent minimum at 20–30 g/cm² and then gets worse. The engineering consequence is that a habitat wants a thin structural shell and a thick hydrogen-rich or regolith overburden, not a thick metal hull — and the overburden is a mass-per-unit-area requirement, which means a bulldozer rather than a factory.

Established The third is the ISRU plant, scaled. MOXIE's stack was ten cells at 800 °C with scandia-stabilised zirconia, drawing atmosphere through a HEPA filter and a 3,500 rpm scroll compressor with no true pressure regulation, and degrading measurably in area-specific resistance each cycle. A full-scale plant is 250 to 500 times larger, must run for months rather than an hour at a time, must survive dust storms, and must be followed by a Sabatier reactor, a water source, a liquefier and cryogenic storage — none of which has flown. Frontier Zubrin's architecture avoids the water problem by importing eight tonnes of hydrogen and accepting a 14:1 leverage, which is a real engineering answer and one whose principal cost is that the hydrogen must be stored cryogenically for the outbound cruise.

Established The fourth is the life-support system, and its requirement is a duration rather than an efficiency. 98% water recovery is achieved and sufficient. Food closure on any flown system is zero, the best ever achieved with humans is 60% over a year with waste uncontained, and the mission needs about 900 days with no resupply and no return option for up to 500 of them.

Frontier And the fifth is the requirement nobody writes down: everything above must work at once, in the same vehicle, with the same crew, for the same 900 days. The published risk assessments treat radiation, closure, partial gravity and dust separately. The one engineering statement this brief can make about the combination is that it has never been analysed jointly, and that assuming the risks add rather than interact is an untested assumption rather than a conservative one.

Established Artificial Gravity owns the physiology and this brief hands off rather than re-deriving. Bone and muscle loss with countermeasures, the AGBRESA randomised trial and its missed endpoint, the partial-gravity animal data, rotation-rate and Coriolis tolerance, spaceflight-associated neuro-ocular syndrome, and the dose–response question of how much gravity for how long — all of it is FR-I-02's. What this brief keeps is one sentence of fact and the compounding argument above.

Established Historical Space Colonization Concepts owns Biosphere 2 in full — the two-year eight-person closure, the oxygen decline, and what the episode did and did not prove — along with the 1970s Mars-settlement rhetoric. This brief takes only the Test Module datum from the closure ladder and owns the modern advocacy record from Zubrin in 1991 onward. In the other direction, FR-X-08 should not carry the ISS's 98% water figure, the 2025 “no nation has demonstrated” statement, or the four-to-eight-year readiness estimate; those are current rather than historical.

Frontier Lunar Industry is the operational rehearsal and the parallel resource story. Both bodies are at the mapped, not measured stage on their ice, and both briefs state that distinction in compatible words. What differs is the chemistry, which barely overlaps at all. Moon-Based Manufacturing supplies the regolith-bagging result that a Mars surface habitat would inherit for shielding, and O'Neill Cylinders supplies the free-space shielding areal densities against which the Mars numbers should be read.

Frontier And two forward neighbours. Terraforming owns the planet-changing claims that this brief neither assumes nor requires. Multi-Planetary Civilization owns the argument about what a settlement would mean — redundancy, self-sufficiency thresholds, governance — and inherits this brief's constraints as its inputs.

Established The most striking institutional fact is that the strongest evidence against every optimistic claim on this page comes from inside the agency that would fly the mission. NASA's thick-shielding project reports that its own primary lever reverses above 20–30 g/cm². NASA Langley's own study says human-scale Mars entry technologies “have not yet been demonstrated, and are not yet planned”. A NASA Ames analyst re-derives the programme cost at $275–460 billion, an order of magnitude above the advocacy figure, and publishes it. And NASA's Planetary Protection Officer told an international body “we don't have all the answers just yet”. Interest running against the finding in four separate places, which is why those four sources carry more weight here than any external critique.

Established The governing planetary protection policy for a crewed Mars mission does not currently exist. NASA Interim Directive 8715.129 expired on 30 September 2025 and is being converted into permanent procedural requirements; the agency risk posture for forward and backward contamination is “initiating”; and the first-order biological contamination model — the thing that would say how much microbial material a crewed vehicle actually releases — is in development. Frontier The institutional problem is a sequencing one: the requirement is being written after the architectures that must satisfy it, and the sites that make ISRU attractive are the sites where terrestrial life could plausibly propagate.

Established Mars Sample Return is the calibration datum for how this institution actually performs on Mars problems. An independent review board found an $8–11 billion lifecycle cost, near-zero probability of the stated 2027–28 launch, and a programme “established with unrealistic budget and schedule expectations from the beginning”, with an undefined Orbiting Sample design constraining many systems including planetary protection. That is a robotic mission returning a few hundred grams under containment. Any institutional forecast for a crewed mission should start from that outturn rather than from an architecture study.

Frontier The cost literature is itself an institutional exhibit and should be read as one. Estimates for the same mission span $42 billion to $600 billion across three decades. The low end is architecture advocacy explicitly designed to be austere and minimal; the high end is parametric re-derivation from the only large orbital assembly programme anyone has actually completed. The spread is not a disagreement about Mars; it is a disagreement about whether a programme costs what previous programmes cost.

Speculative And the timeline record, applied evenhandedly. An interested party's Mars dates moved 2018, 2022, 2026, then to a deprioritisation in February 2026 in favour of a lunar city on the reasoning that “the Moon is faster”. A government agency's Artemis dates moved from 2025 to 2026 for the crewed lunar flyby, which flew, and from 2026 to 2027 to NET 2028 for the landing. The useful institutional generalisation is that no stated date in this domain has held, from anyone, and a brief that applies the observation to one party and not the other is doing advocacy rather than assessment.

Established The sharpest ethical fact on this page is that the flight requires a waiver. The National Academies recommended a universal career limit of about 600 mSv and stated that Mars crews are likely to receive doses significantly exceeding it, so that “NASA would need to seek waivers to the radiation health standard to pursue these missions”. That is not a technical detail; it is a decision by an institution to exceed a limit it set for the protection of the people it is asking to fly. Frontier The informed-consent question that follows is genuinely hard, because the residual risk is quantified — roughly a 5% increase in fatal cancer risk per sievert — and quantified risk is precisely the kind a competent adult can consent to. The counter-argument is that a waiver granted by the employer to itself is a weaker instrument than a standard.

Frontier Second, and specific to Mars: there is no abort. Orbital mechanics fix the surface stay at up to five hundred days, and the transit at roughly six months each way. A medical emergency, a life-support failure, or a crew-compatibility failure has no evacuation option. Every terrestrial analogue of extreme isolation — Antarctic winter-over, submarine patrol — retains an evacuation path in principle, and the removal of that path is the qualitative change rather than the duration.

Frontier Third, contamination in both directions, with no policy in force. Forward contamination risks destroying the scientific value of the only accessible body where extant life is a live question, and the model that would quantify microbial release from a crewed vehicle is still in development. Backward contamination is the reason a robotic sample return costs $8–11 billion. Speculative A crewed mission internalises the return leg without the containment architecture, and no framework currently allocates that risk.

Frontier Fourth, occupational exposure to a soil whose toxicology is inferred rather than measured. 0.4–0.6 wt% perchlorate is a hard number; grain sizes possibly averaging 3 µm with beryllium and arsenic present are hard numbers; predicted silicosis, thyroid disease and severe anaemia are predictions from terrestrial analogues. The honest statement is that a crew would be the exposure study, and the proposed countermeasures — iodine supplementation and habitat filtration — have never been tested against the actual material.

Speculative And fifth, the framing's own ethical claim, which belongs elsewhere but should be named here. “Second home” carries an implicit argument that Mars settlement is insurance against terrestrial catastrophe. The constraints in this brief — a 900-day mission, a dose above the career limit, no closed ecology, a landing capability short by a factor of thirty — are the inputs to that argument rather than the argument itself, and Multi-Planetary Civilization owns it.

Speculative The civilizational claim attached to Mars is redundancy, and this brief's contribution to it is a constraint list rather than a verdict. A settlement that provides redundancy must be independent of resupply, which means closed at a level nobody has demonstrated at any scale, for a duration nobody has approached, while producing its own propellant from a resource that has been mapped and not measured. Every one of those conditions is currently unmet by a measurable margin, and the margins are now known rather than guessed. Multi-Planetary Civilization owns whether redundancy is the right frame at all.

Established What is genuinely civilizational, and already banked, is the epistemic change. Ten years ago the radiation dose for a Mars mission was a model, the soil chemistry was a spectrum, ISRU was a viewgraph and the cost was a debate. Now the dose is measured in transit and on the surface by an instrument that flew, the perchlorate concentration comes from a wet-chemistry laboratory that landed, the ISRU has a gram count and a degradation curve, and the cost has two convergent parametric re-derivations from a completed programme. Mars is the only body other than Earth where a human-relevant resource-conversion device has actually been operated.

Frontier And the consequence of that change is counter-intuitive and worth stating plainly: the measurements have made the settlement claim harder, not easier. A field that says “we do not know the dose” can hope. A field that says “the dose is 1.01 Sv and the recommended career limit is 600 mSv” has to argue. The same inversion applies to closure, where the honest number is six months against nine hundred days, and to landing, where it is one tonne against twenty to forty. Progress in this subject has consisted largely of replacing optimistic uncertainty with unfavourable precision, and that is what progress looks like when it is real.

Speculative The terminal statement is a conditional rather than a date. If closure reaches the mission duration with waste closed, if a landing capability of twenty to forty tonnes is demonstrated, if propellant ISRU scales by two to three orders of magnitude and runs autonomously through dust storms, if the radiation standard is met or lawfully waived, and if the human response to 0.38 g turns out to be tolerable — then Mars is a place people can live in hardware. None of those five is close, all five are now quantified, and not one of them requires new physics.

These horizons track measured capability gaps rather than announced dates, because the announced-date record in this domain is uniformly bad from every party:

• 10 yr: Frontier Expect ship-to-ship cryogenic propellant transfer to be attempted; it was scheduled for 2026 and gates both this programme and the lunar one. Frontier Expect a MOXIE successor at larger scale and possibly a methane loop, but not a full propellant chain with liquefaction and storage on the surface. Speculative Expect no crewed Mars mission on this horizon: the landing capability is short by a factor of twenty to forty, the enabling entry technologies are unmanifested, and the closure programme that would take four to eight years of operational experience has no facility running. Frontier Expect the planetary protection requirements to be issued, since the interim directive has already expired and the conversion is in progress.
• 25 yr: Frontier This is the earliest horizon on which a crewed landing is defensible, and only if two things happen that currently are not: a demonstrated large-mass entry, descent and landing capability, and a closed life-support system with an operating record measured in years rather than months. Speculative A first crewed mission is not a settlement. On the numbers here it is a 900-day flight delivering 1.01 Sv to a crew under a waiver, with a landed mass that has to carry its own return propellant or make it. Speculative The cost question will not have been settled by argument; it will have been settled by whichever architecture actually flies, and the two independent re-derivations from ISS outturn are the prior worth holding until then.
• 50 yr: Speculative A permanent inhabited station is coherent at this horizon, in the sense that nothing in the physics forbids it and every barrier is engineering, mass or money. A self-sufficient settlement is a different claim, and it requires closing the food and waste loops at scale, which nobody has done anywhere for any duration. Speculative The partial-gravity question would by then have been answered by exposure rather than by experiment, which is an uncomfortable way to answer it and the one the current programme is on course for. Route to Artificial Gravity.
• 100 / 250+ yr: Handwave Beyond useful forecasting, and the defensible statement is about which constraints are permanent. Radiation and gravity are properties of the planet and do not improve; closure and landing mass are engineering and can. Speculative Anything at this horizon that resolves the radiation and gravity constraints is either massive shielding, a rotating facility, or terraforming — and the last belongs to Terraforming, which this brief neither assumes nor requires.

• Depends on Two edges, both real. Lunar Industry because the Moon is where the surface-operations record, the delivered-cost discipline, the dust experience and the fission surface power hardware are being generated, and because NASA's own architecture sequences Mars behind the lunar segments — with the limit stated in the same breath: lunar and Martian ISRU share almost no chemistry, one being high-temperature reduction of silicate rock and the other solid-oxide electrolysis of atmospheric CO₂, so what transfers is operations, autonomy, dust handling and cost discipline rather than process. And Terraforming as a definitional dependency: this brief owns living in hardware on Mars and FR-II-17 owns any claim about changing the planet, and the boundary decides which set of constraints applies. Nothing in this brief assumes or requires terraforming.
• Requires (not on this map) The first is not a technology: NASA Interim Directive 8715.129 expired on 30 September 2025, the agency risk posture is “initiating”, the biological contamination model for a crewed vehicle is in development, and the Planetary Protection Officer's own words to COSPAR were “we don't have all the answers just yet”. The second is a facility rather than a discovery: no nation has demonstrated a completely closed bioregenerative system including waste processing, the longest human closure on record is six months, and the published estimate of what full readiness requires is four to eight years of operational experience that nobody is currently accumulating.
• Enables Multi-Planetary Civilization, and the edge is a constraint transfer rather than a capability claim. What passes forward is the measured input set: a 900-day conjunction-class mission, 1.01 Sv against a recommended 600 mSv career limit with waivers required, a closure record of six months at 95.4% and zero closed waste loops, a landing capability short by a factor of twenty to forty, an unattempted orbital propellant transfer, and a parametric programme cost of $275–460 billion re-derived two ways from ISS outturn. FR-II-25 owns redundancy, self-sufficiency thresholds and governance; this brief makes no claim about what a settlement would mean.
• Adjacent Artificial Gravity owns all partial-gravity and rotation physiology and this brief cites rather than re-derives it. Historical Space Colonization Concepts owns Biosphere 2 and the pre-1990 settlement literature. O'Neill Cylinders and Space Habitats supply the shielding areal densities and the operational habitat record against which the Mars numbers should be read; Moon-Based Manufacturing supplies the regolith-bagging result a surface habitat would inherit; and Deep Space Infrastructure and Space-Based Manufacturing own depots, cryogenic transfer architecture and in-space assembly.

Established “MOXIE proved we can make rocket fuel on Mars.” MOXIE made 122 grams of oxygen across 16 runs at a peak of 12 g/hour, and its own authors state that a full-scale plant must be “hundreds of times larger” at 2–3 kg/hour — a factor of 250 to 500. Established It made no methane, did not liquefy, did not store, and did not run continuously; its authors list long-term durability, thousands of hours of continuous operation, autonomous control and seasonal response as remaining to be demonstrated. It is the only in-situ resource utilisation ever demonstrated off Earth by anybody, and it is a technology demonstration rather than a solved capability.

Established “Mars radiation is manageable.” The measured mission total is about 1.01 Sv — 1.84 mSv/day in transit and 0.64 mSv/day on the surface — against a recommended 600 mSv career limit, which is 1.7 career limits in one flight. Established The National Academies' own words are that Mars crews would receive doses “significantly exceed[ing] the proposed new standard” and that NASA “would need to seek waivers”. Nothing about that is unmanageable in the sense of lethal; it is unmanageable in the sense of exceeding the standard the field wrote for itself.

Established “Just add more shielding.” NASA's own thick-shielding project finds that a dose-equivalent minimum may exist at 20–30 g/cm² of aluminium, beyond which added mass raises dose through secondary neutrons and light particles — stated explicitly against the conventional assumption that more shielding is safer. Frontier Hydrogen-rich materials and regolith behave better, and the requirement to reach Earth-equivalent surface dose is several hundred g/cm², which is tonnes per square metre of roof. The answer is a thin shell under a thick pile of dirt, not a thick metal hull, and that is a different engineering problem.

Frontier “There is water on Mars, so there is Mars propellant.” The SWIM project maps ice-consistency from orbit and identifies Arcadia Planitia and Deuteronilus Mensae as best matching the remote-sensing criteria for accessible ice-rich subsurface material. Established No depth, purity, overburden hardness or extraction rate has been measured in place anywhere on Mars. NASA's own 2005 roadmap set a Mars water extraction demonstration for 2013 and rated the capability “just initiated”. This is the identical present-versus-minable distinction that governs Lunar Industry, on a different body, at the same stage.

Established “Starship makes it a solved transport problem.” Twelve flights, seven successes, five failures or partials, Block 3 flying — a real and improving record. Established And the single capability the architecture depends on most has never been attempted by anyone: ship-to-ship cryogenic propellant transfer. Flight 3 moved propellant between two tanks on the same vehicle, which is a different thing; the ship-to-ship demonstration was expected in 2026; and roughly ten tanker launches are needed to fill a depot for one mission.

Speculative “A Mars programme would cost about $50 billion.” Zubrin's 1991 figure is $42 billion and Mars Semi-Direct was costed at $55 billion over ten years. Established Those numbers sit in a table compiled by a NASA Ames analyst alongside $517 billion (SEI, 1989), $300–600 billion (NRC, 2014) and $500 billion (Garver and Sommerer), and beside two independent re-derivations from ISS cost outturn: $320–460 billion mass-based and $275–398 billion life-cycle. Frontier The low figures describe deliberately austere and minimal architectures and the compiler's own caution is that “we can hope but not plan” for better. Print the spread, label the advocacy, and do not average.

Established “Closed-loop life support is basically working on the ISS.” Water recovery is at 98% and that is genuinely sufficient. Food closure on the ISS is zero, the best food closure ever achieved with humans is 60% over a year with waste uncontained, and the longest human closure of any kind is six months at 95.4%. Established The 2025 statement in npj Microgravity is unambiguous: no nation has demonstrated a completely closed system including waste processing, and full readiness is estimated to require four to eight years of operational experience that nobody is accumulating. The mission is 900 days.

Frontier “The perchlorate is a solvable nuisance.” The concentration is measured at 0.4–0.6% by mass and perchlorate is chemically and biologically reducible, so the chemistry is genuinely tractable. Frontier What is not established is the human side: no dose–response data exist, the dust may average 3 µm and also carries beryllium and arsenic, the predicted effects — silicosis, thyroid disease, severe anaemia — are inferred from terrestrial mineral dusts, and the paper that measured the chemistry explicitly does not address human use of the soil.

Frontier “Mars gravity is enough, it's a third of Earth's.” No human has ever been exposed to 0.38 g. The orbital partial-gravity data are in mice and show protection that is system-specific rather than proportional; NASA's own evidence report calls the human response unknown; and the one quantitative modelled health floor sits at 0.4 g, above Mars. Speculative This brief does not adjudicate it and neither should any other — route to Artificial Gravity — but the compounding is this brief's: 0.38 g and 0.64 mSv/day and perchlorate dust and an unclosed loop, at once, for 900 days, with no evacuation, is a combination no published assessment analyses jointly.

Speculative “Terraforming will make Mars habitable, so the settlement problem is temporary.” Nothing in this brief assumes or requires terraforming, and the two subjects are separated by orders of magnitude in energy and time. Terraforming owns every claim about changing the planet, and a settlement argument that leans on it has substituted a harder problem for a hard one.

Established And the framing itself. “Mars is a plausible second home” fails on home and survives on plausible destination. Established The weak point is not any single obstacle but that the obstacles are all measured and they are simultaneous: a dose above the recommended standard, a life-support system never closed at human scale for the duration, a landing capability twenty to forty times short, an unattempted orbital refuelling, a gravity level no human has experienced, a soil with an untested toxicology, and a parametric price of $275–460 billion. Established The honest version is narrower and still remarkable: Mars is the only body other than Earth where a human-relevant resource-conversion device has been operated, and the barriers to a crewed mission are now quantified rather than guessed. Every number above is a measurement, and that is exactly what makes the settlement claim harder.