On the Gulf coast of Louisiana, a launch site is being built for thousands of launches a year. Around a hundred billion dollars, a former industrial property, an operator who spent seven years looking for the right location. The figure is impressive, but it is not the interesting part of the announcement. What is interesting is how the site is described: its own propellant production, its own power generation, a deepwater port, its own vehicle processing, probably its own airfield. A place that needs nothing it does not produce itself.
I read that as a diagnosis rather than an announcement. A site like this gets built because something at the far end of the chain has to be supplied indefinitely. Anyone planning thousands of launches a year is not planning a departure. They are planning a schedule that never ends.
Which means the question this brief is about is already on the table, just in the wrong place. We are building self-sufficiency where it is easy to obtain, on a stretch of coast with roads, gas, power and harbour access. At the other end of the chain, where it decides whether people live, we leave it open. That distribution is not carelessness and not the failure of individual engineers. It shows precisely what we can do today and what we cannot.
Space is almost always discussed in quantities that point upward. Payload in tonnes, thrust in meganewtons, launches per year, cost per kilogram. All of these describe how much we send. None of them describes what happens at the destination when nothing arrives. That quantity is missing, and it is the only one that ultimately counts.
The Number That Appears in No Brochure
German engineering has a word for it, Standzeit, and I am going to keep it. In manufacturing it names the running time of a tool before it fails, and it fits here without alteration: the span an inhabited facility keeps working once deliveries stop. In English I will call it unsupported duration, because that is what it measures. The word matters less than what it refuses to count. Standzeit is not set by what was stored. It is set by the first part that fails and cannot be replaced on site. A station with five years of food and a pump seal that turns brittle after fourteen months has an unsupported duration of fourteen months.
Follow that number and the record becomes uncomfortably short. On the International Space Station, water processing recovers up to ninety eight percent of water from urine, and roughly ninety percent across the system as a whole. That is a remarkable engineering achievement. At the same time, solid waste is simply bagged and burned during a cargo vehicle's reentry, and food comes entirely from below. The loop for carbon and nutrients is not even attempted.
On the ground the picture is barely better. The European Space Agency has worked since 1989 on a closed loop built from five biological compartments. The programme has produced solid research and has never flown as a complete cycle. The best known attempt of all, a sealed facility in the Arizona desert with eight people inside, needed oxygen injected from outside after about a year and a half, because the concrete of the structure absorbed carbon dioxide and the atmosphere shifted. Soviet experiments in Krasnoyarsk reached runs of several months at roughly eighty five percent closure, with stored food as a buffer.
That is the entire record. No human system has ever run fully closed, and the longest runs are measured in months, each with intervention from outside. We are building launch sites for thousands of flights a year while never having reached the basic quantity that makes a single place independent.
What Local Production Closes and What It Does Not
There is a standard objection at this point, and it is a fair one. Producing raw materials on site changes the arithmetic considerably. Oxygen can be extracted from regolith, water from the ice in permanently shadowed craters, propellant from both. On Mars the atmosphere supplies carbon dioxide from which methane and oxygen can be synthesised. This is not speculation, it is process engineering, and it removes the heaviest items from the freight bill.
But it closes the loop exactly where the loop is easiest to close. Local production covers the heavy and simple: oxygen, water, propellant, building mass, later metals and glass. What it does not cover is the light and complex. Seed stock, nutrient solutions, enzymes, catalysts, filter membranes, seals, bearings, lubricants, power electronics, sensors. Together these weigh almost nothing, and they still determine when a facility comes to a halt.
The reason is not lack of effort but depth of manufacture. A membrane that separates water from salt sits at the end of a chain running from base chemistry through polymer synthesis, cleanroom production and test equipment. A bearing requires steel of a specific alloy, heat treatment, grinding machines accurate to micrometres, and someone who has mastered them. A sensor requires semiconductor fabrication. Each of these chains has thousands of links on Earth, and none of them can be rebuilt inside a habitat for a hundred people.
A settlement that makes its own oxygen but needs a membrane from a single factory on Earth every eighteen months has an unsupported duration of eighteen months. Self-sufficiency in the heavy things conceals dependence in the light ones, and because the heavy is measured in tonnes and the light in grams, every freight calculation looks better than the situation is.
Unsupported Duration Against Resupply Interval
From this follows a rule that applies everywhere and favours no particular body: the unsupported duration of a place must exceed its resupply interval. If it is shorter, the place is not a settlement. It is an outpost at the end of a supply line whose interruption is fatal.
In low Earth orbit the interval is days to weeks. The rule is satisfied there, though not by the station itself but by the fact that rockets stand ready below at all times. On the Moon the interval is days. There too the rule can be met through high launch cadence, which is exactly why lunar plans feel so reassuringly plausible.
At Mars the number jumps into a different order of magnitude. Launch windows open roughly every twenty six months, the crossing takes months, and in between there is no opportunity to correct anything. Whoever lands there must hold out for two years without any delivery, and if one window passes unused, two becomes four. On top of that the link to Earth breaks off entirely for a period when the Sun stands between the planets. The required unsupported duration is therefore four to eight times anything ever achieved.
Beyond that the resupply interval disappears altogether. A habitat in free space, a ship between stars, a facility in the outer solar system: for them no delivery exists that could be waited for. Their unsupported duration must be unlimited, which is another way of saying they must be able to manufacture every part they need.
The rule sorts every plan by seriousness without anyone having to argue about visions. It asks only how long it runs when nobody comes.
The Scale That Only Knows Watts
There is already a scale for measuring the maturity of a civilisation, and it dates from 1964. Nikolai Kardashev wanted to know how brightly an alien civilisation would appear in a radio telescope, and proposed three levels: the energy arriving at a planet, the energy of an entire star, the energy of a galaxy. This later became a continuous formula that places humanity at around 0.7.
The scale was never meant as a development plan. It was a search criterion for observers. It measures exactly one quantity, power. For the question of whether someone can be detected from outside, that was the right choice. For the question of whether a civilisation endures, it is the wrong one.
Power says nothing about endurance. A civilisation can multiply its consumption a hundredfold and still be unable to run a single habitat for two years without resupply. It can build solar collectors spanning a planet while depending on a supply chain that a single conflict interrupts. Counting watts alone means mistaking growth for maturity, and that confusion runs deep in the way we talk about progress.
The scale we need instead measures time rather than power. The class of a civilisation is the longest span one of its systems has demonstrably sustained without input from outside. On that scale humanity stands at months. That is the figure that belongs next to the 0.7, and it is the more uncomfortable of the two, because it cannot be forecast, only measured.
Knowledge Nobody Can Execute
Behind unsupported duration sits a second problem that is named less often. Knowing how something is built is not enough. You have to actually be able to build it, and those are not the same thing.
The complete design documentation for the engine that lifted the Saturn V off the pad survives. Nobody could build a new one from it. The tooling had been scrapped, the suppliers had vanished, and the decisive adjustments were not in the drawings but in the hands of people who had reworked individual components during production. When a team of engineers measured a surviving unit decades later, that was not a look in the archive. It was archaeology on their own past.
The pattern repeats wherever a capability is not continuously exercised. Knowledge can be stored, capability cannot. Capability lives in machines that are maintained, in firms that have orders, and in people who apply and pass it on daily. Break the chain and what remains is a description nobody can act on.
Artificial intelligence shifts this point less than it appears. It can hold, order and explain the descriptive knowledge completely, and it will help enormously with design and diagnosis. It does not replace the machine tool, the feedstock and the practised process. A habitat that needs a seal needs a seal, not a text about one.
For a settlement meant to last, something uncomfortable follows. It must carry not only supplies but manufacturing, and it must use that manufacturing even when ordering the part would be cheaper. A capability held in reserve is gone within a generation.
The Purpose That Changes Generation
When a task outlasts a life, it is completed by people who did not witness its beginning. That holds for a habitat operated across eighty years just as it holds for any undertaking spanning centuries.
The difficulty is not transmission. We decipher scripts today whose speakers died thousands of years ago, and a technological civilisation will read its own records reliably. The difficulty is motive. The second generation knows the reason for the effort only as a story, the third only as a rule. Once the effort looks like an imposition and the purpose like somebody else's decision, the chain breaks where it is weakest, at effort without visible benefit.
The very manufacturing that secures unsupported duration is exactly that kind of effort. It produces parts that could also be delivered, it ties up people and energy, and its value shows only at the moment when nobody delivers. A settlement that economises on this feels no disadvantage for twenty years. The disadvantage is felt by a generation that no longer has a choice.
Thinking in stages therefore means handing over more than facilities. It means handing over the willingness to run something whose worth only appears in an emergency. That is not a technical task, and there is no assembly for it.
What Stellar Timescales Demand
At the top of this ladder sit tasks whose scale exceeds every human institution. The Sun grows roughly ten percent brighter per billion years. Calculated proposals exist for pushing Earth's orbit slowly outward through repeated flybys of large bodies, offsetting that increase with one encounter every few thousand years, across billions of years. The physics is not exotic. It is the same mechanics we have used for decades to send probes to the outer planets.
What is missing is nothing physical. What is missing is an entity that performs a task correctly every few thousand years without losing the reason, losing the capability, or economising away the facility. Put differently, the same quantity stands at the top of the ladder as at the bottom. Unsupported duration is not a question about habitats. It is the question of how long something runs when nobody intervenes from outside, and it scales from a filter membrane to the orbit of a planet.
That is why I consider the Standzeit of a place the most important number we can take about ourselves. It cannot be improved by announcements or accelerated with capital. It rises only when someone builds a facility, seals it, lets it run, and writes down what failed first. This work is unspectacular, it produces no images of launch pads, and it decides whether outposts ever become places.
We have learned in recent years to move a great deal of mass upward very often. We have not learned to let any of it run alone. As long as that holds, every settlement beyond Earth is an extension of terrestrial supply chains and not a second address for humanity.
Which facility in your own area of responsibility would still be running if nothing were delivered for a year, and do you know which part would be missing first?
Website: https://planet-futures.org