Multi-Planetary Civilization

This brief asks one question: does becoming multi-planetary secure the species? It is a question about an argument, not about a habitat, a launch vehicle, a reactor or a treaty. The framing under test is that a second world is insurance.

The argument has a careful form and a popular form, and almost the entire difference between them is a conditional clause. In its careful form, stated by the philosopher who has done most to formalise existential risk, the claim is that once we have created many self-sustaining space colonies, any disaster confined to a single planet cannot eliminate all of humanity. That is logically sound. The load-bearing words are “self-sustaining” and “many”, and neither is decorative. In its popular form the conditional is dropped, and what remains is an assertion.

So this brief prices the condition. The pricing is done with the closure record — what has actually been achieved in sustaining humans in a sealed loop — with the minimum-viable-population literature, and with the list of things about human biology off Earth that have never been measured. Every number in it is imported from another domain for the purpose of testing the claim, and the engineering and institutions belong to the six briefs this one depends on.

The result is not that settlement is worthless. It is narrower and harder: an outpost that imports its food is not a backup. It is an additional dependency — a population whose survival is contingent on the very civilisation it is supposed to insure. The strongest single sentence in the critical literature comes from two authors who set out to write an enthusiastic book about settling space and reported changing their minds: even a hypothetically devastated Earth would be more habitable than anywhere else in the Solar System.

Frontier Start with the argument in its most careful form, because the popular version is a corruption of it. Bostrom's Existential Risk Prevention as Global Priority (Global Policy 4(1), 2013) defines an existential risk as one “that threatens the premature extinction of Earth-originating intelligent life or the permanent and drastic destruction of its potential for desirable future development,” and sets out a scope-severity taxonomy with four classes of catastrophe: human extinction; permanent stagnation, subdivided into unrecovered collapse, plateauing and recurrent collapse; flawed realisation; and subsequent ruination.

Frontier The priority argument is expected-value and its magnitude is stated explicitly: “the expected loss of an existential catastrophe is greater than the value of 10¹⁶ human lives,” motivating the maxipok rule — “Maximise the probability of an 'OK outcome', where an OK outcome is any outcome that avoids existential catastrophe.” None of that is in dispute here and none of it is an empirical claim; it is a decision rule.

Established And here is the settlement claim, verbatim, with its conditional intact: “once we have created many self-sustaining space colonies, any disaster confined to a single planet cannot eliminate all of humanity.” The argument as stated by its most rigorous advocate is conditional on a state of affairs that has never been demonstrated at any scale, for any duration, anywhere. The rest of this section is what that state of affairs would cost.

Frontier The argument's history, named rather than gestured at. Louis J. Halle Jr. appears to be the first to make it explicitly, in 1980, holding that colonisation of space would keep humanity safe should global nuclear war occur. Carl Sagan and Stephen Hawking are its popular carriers — Hawking's “Colonies in space may be only hope.” And the same figures rejected a different and routinely conflated claim: Sagan and Arthur C. Clarke both denied that space settlement addresses overpopulation, Clarke stating that “the population battle must be fought or won here on Earth.”

Frontier The strongest critique available is the Weinersmiths' A City on Mars (2023), and its weight comes from the direction its authors moved. They report changing their own minds while writing: the challenges were “more daunting than they expected” and “ignored by popular depictions.” Their conditional matches Bostrom's exactly and then prices it: settlement improves survival odds only on achieving economic self-sufficiency, which requires “millions of settlers” or “incredibly advanced hypothetical robotics.” Interest ran toward the framing and the finding ran against it, which is the strongest evidential configuration available in a literature this thin.

Frontier Their specific obstacles are worth listing because each is checkable. Lunar regolith lacks carbon and phosphorus; food would need alternative sources such as insects; pregnancy and child-rearing in low gravity are unstudied and the artificial-gravity workaround — a banked centrifuge for gestation — is impractical; existing law offers precedents for survival homicide and no framework for property; the Outer Space Treaty needs updating. And the sentence this brief treats as the argument's terminus: “even a hypothetically devastated Earth would be more habitable than other options in the Solar System.” The book was not obtained directly for this research; these claims are reported through a tertiary summary and labelled.

Speculative Two further critics, named without being quoted beyond what was obtained. Daniel Deudney's Dark Skies: Space Expansionism, Planetary Geopolitics, and the Ends of Humanity (Oxford University Press, 2020) is the standing scholarly counter-position, arguing that space expansion increases rather than reduces existential risk. The book's argument was not obtained — publisher and encyclopaedia sources both refused during this research — so it is named and not quoted, and no claim is attributed to it beyond its existence and its author's position. Joon Yun describes space expansion as “imperialist reasoning... only relocating collective shortcomings and not solving them.”

Established Now the closure record, which is where the conditional is priced. The field-level statement first: “no artificial closed ecological systems involving humans have been able to successfully support life for long periods of time.” Historical Space Colonization Concepts owns Biosphere 2's history, personalities and narrative; this brief carries only the numbers the argument turns on and links.

Established Biosphere 2, Mission 1 (26 September 1991 to 26 September 1993, eight people, two years): oxygen fell from 20.9% to 14.5% over sixteen months — “equivalent to the oxygen availability at an elevation of 4,080 metres” — requiring two oxygen injections, in January and August 1993, after hypoxic symptoms including sleep apnoea and fatigue. The agriculture produced 83% of the diet; the crew lost an average of 16% of pre-entry body weight and reported “continual hunger.” Carbon dioxide swung 600 ppm daily and ran 4,000–4,500 ppm in winter. Most introduced vertebrates and virtually all pollinators died; cockroaches and invasive ants flourished.

Established Biosphere 2, Mission 2 (6 March to 6 September 1994, seven people): planned for ten months and ended at six. It did achieve complete food self-sufficiency and needed no oxygen injection. That is the best closure result on record with humans — and it is half a Mars transit, at under half the crew of even the smallest proposed settlement, on Earth, at one gravity, with rescue available.

Established BIOS-3 (Krasnoyarsk; built 1965–1972; 315 cubic metres; up to three people): the longest run was 180 days in 1972–73. Plants supplied roughly 25% of air filtering; water recycling reached 85% by 1968. Dried meat was imported, and “urine and feces were generally dried and stored, rather than being recycled.” A system that imports protein and stores its waste is not closed; it is a long-duration isolation study, which is a valuable but different thing.

Established The International Space Station (continuously crewed since 2000) is the only operational long-duration case and its figures are the sharpest in the record. Water recovery reached 98% in June 2023 with the Brine Processor Assembly, up from 93–94%. NASA's own framing of why this matters: “The inability of resupply during exploration means we need to be able to reclaim all the resources the crew needs on these missions.” And food closure on the ISS is zero. Every calorie is launched from Earth.

Frontier One programme that would bear on this could not be assessed. ESA's MELiSSA is a multi-compartment closed-loop development programme; its current status was not obtained during this research and no closure percentage for it is printed here. Recording the gap is more useful than filling it with an approximation.

Established The arithmetic, assembled. A conjunction-class crewed Mars mission is a multi-year commitment. The best human closure ever demonstrated is two years with two oxygen injections and 16% body-weight loss, or six months fully food-closed at a crew of seven, or 180 days with imported meat and stored waste. None of these is a closed system, and all of them were on Earth, at one gravity, at one atmosphere, with rescue available.

Speculative The minimum-viable-population literature spans more than two orders of magnitude, and the spread is more informative than any of its members. The anthropologist John Moore gives 80 — couples boarding in their early twenties and delaying reproduction to their mid-to-late thirties, without genetic testing and without accounting for advanced-maternal-age risk — and 160 for a 200-year journey with “normal family life”, about ten potential partners each. Marin and Beluffi's HERITAGE Monte Carlo (JBIS, arXiv:1806.03856, 2018) gives 98 for a 6,300-year voyage to Proxima Centauri b, with annual population evaluation, offspring restrictions and breeding constraints — and reports that 25 breeding pairs went extinct in 50 ± 15% of runs when inbreeding was completely forbidden. Cameron Smith gives tens of thousands, including accident and disease risk and at least one severe population catastrophe over a 150-year journey.

Frontier The observation that matters is not any one figure. It is what every figure is about. Every estimate in that range is a genetics and demography calculation. Not one of them prices an industrial base. A population of ninety-eight that cannot manufacture a replacement air-handling membrane is not a viable population; it is a countdown. And Marin and Beluffi's own extinction rate at twenty-five breeding pairs shows how brittle the low end is even on its own narrow terms. A further estimate around a hundred and ten exists in the literature; its source was not obtained and no figure from it is printed here.

Frontier Then the things that are simply unknown, and the word is exact. Human reproduction off Earth has never been studied, at any gravity, for any duration. There is no counter-evidence to cite because no experiment has been run. The gravity dose-response curve is unmeasured: human physiology has data at one gravity and at microgravity and nothing between, so every claim about long-term health at 0.38 g is an interpolation between two points. These are not contested questions. They are unasked ones.

Established And the destination's actual conditions, from the agency. Mars runs from +20 to −153 degrees Celsius, with a thin atmosphere “made up mostly of carbon dioxide, nitrogen, and argon” and a 24.6-hour rotation. Frontier The radiation dose for a Mars mission — the cruise and surface measurements and the agency's permissible exposure limits — was not obtained during this research, and no dose figure is printed here. The qualitative statement that survives without it: deep-space and Mars-surface dose rates exceed anything in the human Earth-orbit experience, and that is flagged pending the number.

Frontier Assemble the pieces without adding anything, and the argument fails on its own terms. Bostrom's claim is conditional on many self-sustaining colonies. The Weinersmiths price self-sustaining at millions of settlers or robotics that do not exist. Every population estimate in the literature is a genetics number spanning 80 to tens of thousands, and none models an industrial base. The best human closure ever demonstrated is two years with oxygen injections and 16% weight loss, on Earth, with rescue available. Food closure in operational spaceflight is zero. And a devastated Earth would still be more habitable than anywhere else in the Solar System.

Frontier Position one: existential-risk reduction is a global priority. Bostrom, 2013, argued from expected value with the maxipok rule. It is a decision-theoretic position rather than an empirical claim, and this brief does not contest it. Position two, and it is the one that matters: many self-sustaining colonies would make a single-planet catastrophe non-fatal. Logically sound as stated, and conditional on a state of affairs never demonstrated anywhere. Flagging this frontier rather than speculative is deliberate: the argument is valid and its premise is unmet.

Speculative Position three: space settlement is humanity's insurance. Halle, Hawking, and the popular framing. It drops Bostrom's conditional, and that is precisely where it fails — not in its logic but in its premise, which the closure record shows is unmet by several orders of magnitude on every axis that has been measured.

Frontier Position four: a devastated Earth beats anywhere else in the Solar System. The Weinersmiths, from comparative habitability, having moved toward this position while writing a book they began as enthusiasts. Position five: self-sufficiency requires millions of settlers or robotics that do not exist. An estimate rather than a calculation — and no counter-estimate prices industry at all, which is the more important fact about it.

Speculative Positions six through eight are the population estimates, and they should be read as a spread rather than as competing answers. About 80 to 160 from genetics alone with no accident or disease modelling; 98 from a Monte Carlo that reports 50 ± 15% extinction at 25 breeding pairs under strict no-inbreeding rules; tens of thousands once accidents, disease and one severe catastrophe per 150 years are included. The disagreement is not about arithmetic. It is about which risks are in the model, and industry is in none of them.

Frontier Position nine: closed-loop life support is achievable. Held implicitly by every life-support programme. The evidence: 98% water closure in orbit, zero food closure in orbit, and a best-ever human closure of six months on Earth. Position ten: humans can reproduce and develop off Earth. Assumed by every settlement plan and supported by zero human data at any gravity. That is not a contested question but an unasked one, and the distinction matters for how the flag is set.

Speculative Position eleven: space expansionism increases existential risk. Deudney's standing scholarly position, named here without a quotation because the argument was not obtained. Position twelve: settlement solves overpopulation. A popular framing rejected by Sagan and Clarke, two of the people most often invoked in support of the general case — which is why the two arguments should never be run together.

Frontier And the question that would actually settle the subject: what would a demonstration look like? Not a habitat, not a launch cadence, not a treaty. A human crew closed for a Mars-mission duration with no resupply, closing air, water, food and waste, and manufacturing at least one of its own critical components. Nothing in the record approaches it, nobody has proposed a programme to attempt it, and the absence of that programme is the most informative fact in this brief.

Established The first bottleneck is atmospheric closure, and the failure mode is documented in detail. Biosphere 2's oxygen fell from 20.9% to 14.5% in sixteen months — the equivalent of living at four kilometres' altitude — and required two injections. Carbon dioxide swung 600 ppm daily and reached 4,000–4,500 ppm in winter. The system was a two-year experiment on Earth with an atmosphere available outside the glass, and it still could not hold its air.

Established The second is food, and it is the sharpest number in the brief. Biosphere 2's first mission produced 83% of its diet and its crew lost 16% of body weight with continual hunger; its second achieved full food closure for six months with seven people. In operational spaceflight, food closure is zero — every calorie on the ISS is launched. A settlement that imports calories is a logistics node, not a backup.

Established The third is waste and water, where the numbers look best and still do not close. The ISS recovers 98% of its water, up from 93–94%, which is a genuine engineering achievement. Two percent over years without resupply is not a rounding error, and BIOS-3 dried and stored its solid waste rather than recycling it. Nothing human has closed a waste loop.

Speculative The fourth is the one nobody has attempted: manufacturing closure. No experiment at any scale has demonstrated a human population making the components its life support depends on. Every minimum-viable-population estimate is silent on this, which means the entire population literature answers a question about genes while the binding constraint is about factories. The engineering belongs to Moon-Based Manufacturing and Lunar Industry; the observation that no estimate prices it belongs here.

Frontier The fifth is biological and it is an absence of data rather than a difficulty. Human reproduction off Earth: no studies, at any gravity. Long-term health at Mars or lunar gravity: no data, only an interpolation between one gravity and microgravity. A settlement plan that assumes a multi-generational population is assuming the resolution of two questions on which the evidence base is empty.

Frontier The sixth is radiation, stated qualitatively because the numbers were not obtained. Deep-space and Mars-surface dose rates exceed anything in the agency's Earth-orbit experience. No dose figure and no exposure limit is printed in this brief because the primary sources were unreachable during this research, and a brief that invents one to complete a table is doing the reader a disservice. Mars Colonization and Artificial Magnetospheres own the shielding question.

Speculative And a seventh that is institutional: no governance framework exists for an isolated population. The Weinersmiths note that existing law offers precedents for survival homicide and no framework for property, and that the Outer Space Treaty predates the question entirely. Space Law and Governance owns the regime; the point here is that “self-sustaining” has a legal component nobody has specified.

Established Six adjudicated dependencies, and the list is not a topic cluster — it is the enumeration of what “self-sustaining” would require. Lunar Industry and Moon-Based Manufacturing own whether an off-world population could make anything; Mars Colonization and Space Habitats own where it would live; Space Resource Economies owns whether it could pay for itself; Space Law and Governance owns whether it could govern itself. Bostrom's conditional is satisfied only if all six produce affirmative results, and none of them has.

Established The seam is identical in every case and is worth stating once: they own the engineering and the institutions, and this brief owns the argument. Every number in sections 2 and 4 is imported for the purpose of pricing a claim. This brief does not re-specify a habitat, a reactor, a mining operation or a treaty, and where it needs one it links.

Established Biosphere 2 is a cross-reference and not a subject here. Historical Space Colonization Concepts owns the history, the personalities and the narrative. This brief carries only the figures the argument turns on — oxygen from 20.9% to 14.5%, two injections, 83% of the diet, 16% of body weight, and Mission 2's six months at full food closure — because without them the conditional cannot be priced at all.

Frontier Two further relationships that are boundaries rather than dependencies. Exoplanet Colonization owns the destination question beyond this system, and the interstellar arm of the backup argument inherits its findings directly: the nearest terrestrial candidates have no measured atmosphere and sit 75,000 years away at demonstrated cruise speed. Interstellar Civilization Models owns where expansion leads; this brief asks only whether it insures.

Speculative And one methodological connection worth naming because the same people are involved. Sandberg and Drexler appear both in the existential-risk literature and in the Fermi literature, where their contribution was to show what happens when a deeply uncertain multiplicative claim is propagated honestly rather than as point estimates. “Multi-planetary secures the species” is exactly such a claim, and it has never been given that treatment.

Frontier One dependency is unowned: a closed-loop life-support demonstration at Mars-mission duration. It is not a habitat, not a vehicle and not a treaty, and no brief on this map claims it. It is the single experiment that would move this subject, and nobody is running it.

Established Experiment one, and everything else is secondary to it: close a human crew for a conjunction-class Mars duration with no resupply. Air, water, food and waste, at a crew size in double figures, for the length of a real mission. The best result on record is six months at full food closure with seven people, on Earth, at one gravity, with rescue available — and that mission was planned for ten months and ended at six. The experiment is expensive, terrestrial, and requires no new physics.

Frontier Experiment two: manufacturing closure at any scale. Take one critical life-support component — a membrane, a pump, a seal — and demonstrate its manufacture from local feedstock within a closed system. No experiment of this kind has been attempted, and it is the missing term in every minimum-viable-population estimate ever published.

Frontier Experiment three: mammalian reproduction across a full life cycle at partial gravity. There is no human data and the animal record does not answer the question at Mars or lunar gravity over generations. A partial-gravity facility capable of running a multi-generation study is the only route to converting an assumption held by every settlement plan into evidence, and Artificial Gravity owns the hardware case for it.

Established Experiment four: the gravity dose-response curve. Human physiology is characterised at one gravity and at microgravity and nowhere between. Every claim about decades at 0.38 g is an interpolation between two points, and a curve with three points would be a qualitative improvement on the entire literature.

Speculative Experiment five: a minimum-viable-population model that prices industry. Take the demography, add a manufacturing sector with real component lifetimes, spare-part inventories and skill-transmission requirements, and report the population at which the system survives rather than the population at which the genome does. It is a modelling exercise, it costs one team a year, and its absence is the most striking gap in the published record.

Frontier And experiment six, cheap and clarifying: apply distributional uncertainty propagation to the backup argument itself. The same rigour that dissolved a famous multiplicative estimate elsewhere — propagating ranges rather than point values — has never been applied to “settlement reduces existential risk.” Given how many of its terms are unmeasured, the honest output would be a distribution, and nobody has computed it.

Established The engineering requirement can be written as a checklist, and none of its rows is met. Air closed without resupply: best demonstration is Biosphere 2, which failed at sixteen months and required injection. Water closed: the ISS at 98%, with the residual two percent over years being the unsolved part. Food closed: Biosphere 2 Mission 2 at 100% for six months, seven people, on Earth. Waste closed: BIOS-3 dried and stored rather than recycling. Manufacturing closed: never attempted at any scale.

Frontier The biological rows are worse, because they are empty rather than partial. Population genetically viable: modelled from 80 to tens of thousands, never tested. Human reproduction off Earth: zero data. Health at 0.38 g over decades: zero data. Governance for an isolated population: none, and the governing treaty predates the question.

Established The requirement the checklist implies is unglamorous and specific: the binding engineering problem is not transport. Launch capability, habitat structures and power systems are the subjects of six other briefs and each has real progress to report. Closure has essentially none, and closure is the term Bostrom's conditional actually requires.

Speculative A second requirement follows from the manufacturing row and is rarely stated. A self-sustaining settlement needs not only the ability to make components but the ability to make the tools that make them, and to transmit the skills to do so across generations in a population of a few hundred. The Weinersmiths' estimate of millions of settlers is an estimate about this, not about genetics, and it is the only published attempt to price it.

Frontier And a third: the environment is worse than the analogies suggest. Mars runs from +20 to −153 degrees Celsius under a thin carbon-dioxide atmosphere; lunar regolith lacks carbon and phosphorus, which are not optional inputs to a food system. The nearest analogue site in the Solar System is more hostile than any terrestrial location that has ever been used to test closure, and every closure result quoted in this brief was obtained somewhere far easier.

Established The six dependencies are the adjacent briefs that matter, and the division is consistent: they own engineering and institutions, this brief owns the argument. Lunar Industry, Mars Colonization, Space Habitats, Space Resource Economies, Moon-Based Manufacturing and Space Law and Governance between them cover everything “self-sustaining” would require.

Established Second, Historical Space Colonization Concepts, which owns Biosphere 2 in full. This brief cross-references rather than restates, carrying five figures and no narrative, no history and no personalities. Third, O'Neill Cylinders, which owns the rotating-habitat lineage that the artificial-gravity workaround for gestation would depend on.

Frontier Fourth, Exoplanet Colonization, which the interstellar arm of this argument depends on entirely. Its finding is that no terrestrial habitable-zone candidate is confirmed to hold an atmosphere and the nearest is 75,000 years away at demonstrated cruise speed. A backup beyond this system is therefore a backup at a destination with no measured air.

Speculative Fifth, Artificial Gravity and Artificial Magnetospheres, which own the two mitigations — for the gravity dose problem and the radiation problem — that this brief names and does not assess. Sixth, Interstellar Civilization Models, which owns where expansion leads.

Established And one boundary drawn to prevent a specific error: Civilization Timelines owns the reading of multi-planetarity as a developmental stage a civilisation passes through, flags that reading speculative, and records that the periodisation scheme behind it has zero observed instances above the base type. This brief owns the risk-reduction argument and not the periodisation one, and the two are routinely conflated in exactly the places where the conditional gets dropped.

Established The first institutional fact is that nobody owns the experiment that would settle the question. Closed-loop life support at Mars-mission duration is not a launch programme, not a habitat programme and not a science mission; it is a terrestrial experiment costing far less than any of them, and no agency has it as a deliverable. The most consequential unrun experiment in this subject is also the cheapest one in it.

Frontier Second, the programmes that exist are hard to assess from outside. ESA's multi-compartment closed-loop programme is long-running and its current closure status was not obtainable during this research, so no percentage for it appears here. A field whose flagship programme's performance cannot be established from public sources is a field where the popular account and the evidence base can drift apart indefinitely.

Established Third, the legal position, which is stated by the critics rather than by the advocates. Existing law offers precedents for survival homicide and no framework for property; the Outer Space Treaty predates every question a settlement would raise. Space Law and Governance owns the detail; the argument-level point is that “self-sustaining” includes governing itself, and no one has specified what that would mean for a few hundred people who cannot leave.

Speculative Fourth, an incentive problem that shapes the literature. Settlement advocacy is well funded and well publicised; closure research is neither. The result is a large body of work on how to get people to Mars and a very small body on how to keep them alive there without resupply, and the imbalance is visible in the sourcing of this brief, where the best evidence about closure comes from a 1990s experiment and a 1970s Soviet facility.

Established And fifth, an interest-disclosure norm this brief applies to itself. Settlement advocates are interested parties and are marked as such in the reading list; so is the one national decadal assessment that could not be obtained. The strongest critical source here is strong precisely because its authors began as enthusiasts and report being moved by the evidence, and that direction of travel is worth more than any number of confident assertions from either side.

Frontier The first ethical question is about consent, and it is not hypothetical. A settlement dependent on resupply exposes its inhabitants to a risk profile nobody on Earth accepts, and a multi-generational settlement creates people who did not choose it, in an environment where reproduction and childhood development have never been studied at any gravity. The absence of data is itself the ethical problem: a plan that requires children to be born in an unstudied environment is proposing an experiment on them.

Speculative Second, the legal vacuum has a specific and unpleasant shape. The most careful critics note that the nearest legal precedents for a closed population facing resource failure concern survival homicide, and that no property framework exists at all. These are not exotic edge cases; they are the standard hazards of a small isolated population with no exit, and the governing treaty was written before anyone considered them.

Established Third, the rhetoric of the backup, which is where this brief's finding bites hardest. Presenting a resupplied outpost as insurance against civilisational catastrophe is a factual error with policy consequences: it invites treating an additional dependency as a hedge. An outpost that imports its food would be among the first casualties of a terrestrial catastrophe, not the survivor of it.

Frontier Fourth, and this is the point at which the two arguments must be separated: settlement is not an answer to overpopulation, and its most famous advocates said so. Sagan and Clarke both rejected that claim explicitly. Running the overpopulation argument and the existential-risk argument together borrows the credibility of the second for the first, and the people usually cited in support rejected precisely that move.

Speculative And fifth, the counter-position that this brief names and cannot assess: that space expansion increases existential risk rather than reducing it. Deudney's book is the standing scholarly statement of it and was not obtained here, so nothing is attributed to it beyond its existence. Recording an unassessed serious counter-argument as unassessed is the correct handling; summarising it from memory would not be.

Established The civilizational finding is a conditional and its price. The argument's most careful proponent does not say settlements secure the species. He says that once we have created many self-sustaining space colonies, a single-planet disaster cannot eliminate humanity. Strip the conditional and the claim is unearned. Keep it, and what it costs is everything in section 2.

Established The record, in one paragraph, because it is the whole argument. The best closure ever achieved with humans is Biosphere 2's two years, with oxygen falling from 20.9% to 14.5% in sixteen months, two emergency injections and an average 16% loss of body weight. The best complete food closure is six months, seven people, on Earth, at one gravity, with rescue available. BIOS-3's longest run was 180 days with imported meat and stored, unrecycled waste. The ISS closes 98% of its water and none of its food.

Frontier The population literature adds the second failure, and it is a failure of scope rather than of arithmetic. Estimates span 80 to tens of thousands and every one of them is about genetics. None prices the industrial base that would let those people make the thing that keeps them breathing. A crew of ninety-eight with a viable gene pool and no factory is a countdown with a demography paper attached.

Frontier And the comparison that ends it comes from authors whose interest ran the other way. Two writers who set out to make an enthusiastic case for settling space, and who report that the problems were more daunting than they expected, conclude that even a hypothetically devastated Earth would be more habitable than other options in the Solar System. That is a comparative habitability claim, and no counter-claim in the retrieved literature engages it directly.

Speculative What follows is narrower than either camp usually states. A settlement that cannot close its loops is not a hedge against civilisational collapse; it is a population whose survival is contingent on the civilisation it is meant to insure — the first casualty of the catastrophe rather than the survivor of it. The insurance policy, as currently conceivable, is underwritten by the asset it insures.

Frontier And the reframing this brief recommends, which is not a rejection. Settlement may be worth doing for reasons that have nothing to do with insurance: science, capability, industry, the value of the attempt itself. The risk-reduction argument is the one that fails, and it fails on a condition its own author stated. Civilization Timelines owns the separate and more speculative reading in which multi-planetarity is a stage a civilisation passes through.

These horizons track the closure and biological evidence the argument depends on — not settlement schedules, which the six dependency briefs own.

• 10 yr: Established Water closure in orbit improves incrementally from 98%; food closure in operational spaceflight remains at or near zero, because no flight programme has a food-closure deliverable. Frontier Whether any agency funds a multi-year closed-loop human demonstration is decidable in this window, and nothing currently indicates one will. Speculative The gravity dose-response curve gains a third point only if a partial-gravity facility flies; no human data at Mars or lunar gravity exists today.
• 25 yr: Frontier If crewed Mars missions are attempted in this window they will be resupplied, which tests transport and habitats rather than the condition this brief is about. Speculative A manufacturing-closure demonstration at any scale would be the first evidence bearing on the term every population estimate omits; none is proposed. Frontier Reproduction and development off Earth may acquire animal data at partial gravity; human data at multi-generation scale will not exist on this horizon under any plan currently described.
• 50 yr: Speculative A settlement that closes air, water, food and waste without resupply is conceivable on this horizon and would be the first object the backup argument could actually refer to. Speculative Even then, Bostrom's condition requires many such settlements, and the plural is doing as much work as the adjective. Handwave Any claim that the species is secured at this horizon is an assertion about a state of affairs that has no demonstrated ancestor at any scale.
• 100 / 250+ yr: Handwave Beyond forecasting, and this brief declines rather than extrapolating. Established The structural statement that survives any horizon is the comparison: a devastated Earth remains more habitable than anywhere else in the Solar System, because the floor of terrestrial catastrophe is above the ceiling of every other environment we have measured. Speculative The interstellar arm inherits a harder problem still — no candidate world confirmed to hold an atmosphere, and 75,000 years to the nearest at demonstrated cruise speed.

• Depends on Six edges, and the list is not a topic cluster but an enumeration: it is exactly the set of things that would have to be true for the phrase self-sustaining to be satisfied, and none of them is. Lunar Industry and Moon-Based Manufacturing own whether an off-world population could manufacture anything, which is the term every minimum-viable-population estimate omits. Mars Colonization and Space Habitats own where such a population would live and what would keep it breathing. Space Resource Economies owns whether it could pay for itself, which is the sense in which the strongest critics use “self-sufficiency”. Space Law and Governance owns whether it could govern itself, and the treaty in force predates every question it would face. They own the engineering and the institutions; this brief owns the argument, and imports every number in it to price a claim.
• Requires (not on this map) Three constraints that are not briefs on this map, and together they are the whole of the conditional. The first is an experiment nobody has run: a human crew closed for a conjunction-class Mars duration with no resupply, against a best-ever record of six months at full food closure with seven people, on Earth, at one gravity, with rescue available. The second is an evidence base that is empty rather than thin: there is no human data on reproduction or development at any gravity, and none on long-term health between one gravity and microgravity, so every claim about a multi-generational settlement interpolates between two points. The third is industrial and is the term the entire population literature omits: no demonstration at any scale of a closed population manufacturing the components its life support depends on, which is why estimates spanning 80 to tens of thousands all answer a question about genes rather than about factories.
• Enables No typed enabling edge is claimed. This brief produces an assessment rather than a capability, and its assessment is negative: the condition on which the risk-reduction argument rests is unmet on every measured axis. An argument that fails its own premise enables nothing, and recording an edge outward would misrepresent what a finding of this kind is.
• Adjacent Historical Space Colonization Concepts owns Biosphere 2 in full, and this brief carries five of its figures and links rather than restating the history. O'Neill Cylinders owns the rotating-habitat lineage. Exoplanet Colonization owns the destination question beyond this system, and the interstellar arm of the backup argument inherits its finding that no candidate world is confirmed to hold an atmosphere and the nearest is 75,000 years away at demonstrated cruise speed. Interstellar Civilization Models owns where expansion leads. Artificial Gravity and Artificial Magnetospheres own the two mitigations this brief names for the gravity and radiation gaps. Civilization Timelines owns the reading of multi-planetarity as a developmental stage and flags it speculative.

Established “A Mars base would be a backup of humanity.” Not unless it is closed, and nothing human has ever been closed for a Mars duration. Established The best air closure on record failed at sixteen months and required two oxygen injections; the best complete food closure lasted six months with seven people on Earth; food closure in operational spaceflight is zero. An outpost that imports its calories is an additional dependency, not a hedge.

Frontier “The existential-risk case supports settling Mars now.” Its most careful formulation is conditional: “once we have created many self-sustaining space colonies.” Established The condition is the argument. Speculative The most detailed attempt to price it puts self-sufficiency at millions of settlers or at robotics that do not exist, and no published estimate of a viable population models an industrial base at all. Dropping the conditional converts a valid argument into an assertion.

Speculative “About a hundred people is enough to found a settlement.” That figure comes from genetics. Speculative The published range runs from 80 to tens of thousands, and the disagreement is about which risks are in the model — accidents, disease, one severe catastrophe per 150 years — not about the arithmetic. Frontier One Monte Carlo reports extinction in 50 ± 15% of runs at twenty-five breeding pairs under strict no-inbreeding rules, and not one estimate in the literature prices manufacturing. A gene pool is not a supply chain.

Established “Biosphere 2 showed that closure is achievable.” It showed the opposite at the two-year scale. Established Oxygen fell from 20.9% to 14.5% over sixteen months — the equivalent of 4,080 metres of altitude — two injections were required, the agriculture produced 83% of the diet, and the crew lost an average of 16% of body weight reporting continual hunger. Frontier The genuine success was Mission 2: full food closure with no oxygen injection — for six months, at a crew of seven, planned for ten. Historical Space Colonization Concepts owns the full account.

Established “The ISS is nearly a closed system.” It recovers 98% of its water, up from 93–94%, which is a real achievement and is the strongest closure number in operational spaceflight. Established It closes none of its food: every calorie is launched from Earth. Frontier And two percent of water lost over years without resupply is not a rounding error — the agency's own framing is that exploration means reclaiming all the resources a crew needs.

Frontier “We know humans can live and raise families in low gravity.” There is no human data on reproduction off Earth at any gravity, and none on health between one gravity and microgravity. Established Every statement about decades at 0.38 g is an interpolation between two points. Speculative This is not a contested question with evidence on both sides. It is an unasked one, and a settlement plan that assumes it resolved is proposing the experiment rather than citing it.

Handwave “Space settlement solves overpopulation.” Rejected by two of the people most often invoked in favour of settlement generally: Sagan, and Clarke, whose formulation was that “the population battle must be fought or won here on Earth.” Frontier The overpopulation argument and the existential-risk argument are routinely run together, and only one of them has ever been endorsed by these figures.

Speculative “Anywhere is more survivable than a ruined Earth.” The comparison has been made explicitly and it goes the other way. Frontier Even a hypothetically devastated Earth would be more habitable than other options in the Solar System — a claim from authors who began as enthusiasts, report changing their minds, and set it against Mars at +20 to −153 degrees Celsius under a thin carbon-dioxide atmosphere and lunar regolith that lacks carbon and phosphorus. No retrieved counter-argument engages the comparison directly.

Speculative “Becoming multi-planetary is a stage civilisations pass through.” That is a periodisation claim and it belongs to Civilization Timelines, which flags it speculative and records that the underlying scheme has zero observed instances above the base type. Frontier This brief owns the risk-reduction argument, which is a different claim with a different failure mode: it fails on a condition rather than on a taxonomy.

Handwave “The critics have been answered.” Two named critical positions could not be assessed in this research at all: a book-length scholarly argument that space expansionism increases existential risk, and one further minimum-viable-population estimate. Established Neither is quoted here and no figure from either is printed, and the radiation dose figures for a Mars mission are absent for the same reason. An argument that has not been engaged is not an argument that has been defeated, and marking the difference is the whole point of flagging.

Established And the framing itself. “Becoming multi-planetary secures the species” is not false so much as unearned in its popular form. Frontier Its careful form is conditional on many self-sustaining colonies; self-sufficiency has never been demonstrated for humans anywhere, at any scale, for any duration approaching a Mars mission; and the population literature answers a question about genetics while the binding constraint is industrial. Speculative A settlement that cannot close its loops is not insurance. It is a population contingent on the civilisation it is meant to insure — underwritten by the asset it covers.