What would it be like to work on Mars?

“[People and robots] wanted for hazardous journey. Low wages, bitter cold, long hours of complete darkness. Safe return doubtful. Honour and recognition in event of success.”

Part of my series on common misconceptions in space journalism.

In both science fiction and fact-based journalism there are a bewildering variety of future visions for human existence in space. Indeed, a primary function of science fiction is to examine the human condition by juxtaposing our evolutionary legacy with hypothetical future technology. For example, Kim Stanley Robinson used Mars as a rich palette to explore alternative sociopolitical structures in his epic Mars trilogy.

My purpose here isn’t to knit pick or grade authors, but to add some perspective to contemporary media narratives. Because no-one, including me, has ever lived on Mars, no-one really knows what it might be like. Some narratives hold that humans on Mars will be challenged only by the magnitude of their riches and entertainment as they partake in fully automated luxury space communism. Others hold that humans will have to lurk in underground caverns, tending to potato plants and trying to remember what warmth feels like. Still others foresee a future of indentured servitude, 50 years of hard labor under the lash of Elon’s socialist workers’ paradise…

While my previous sentence was facetious, there are solid arguments both for and against space cities being inherently prone to human freedom. On one hand, a hostile environment and dependence on external supplies seems to vest incredible power in the centralized authority that controls the life support system. On the other, the available labor pool is likely to be overwhelmingly highly motivated, highly educated technical people upon whose continuing good will the success of the enterprise will rest.

I personally believe that the level of productivity required could not be obtained under duress, as ideology is a much more effective motivator than fear of punishment. Nevertheless, it is clear that to serve SpaceX’s vision of a self-sustaining civilization on Mars, there will have to be an extraordinary amount of labor both human and robotic. This blog is my attempt to understand how this system will form, allocate work, scale, and mature.

While I have previously explored aspects of this question in speeches, blogs, and books, here I will bring the reader up to date with my latest thoughts while remaining focused on the human experience.

One common misconception to clear up first are the dangers of reasoning by analogy. The most accessible metaphor for self-sufficiency is the rugged individual pioneer pushing the frontier with marginal agriculture, 40 acres, and a donkey. Numerous popular works on Mars settlement, including by Robert Zubrin, Buzz Aldrin, and Andy Weir, play to this metaphor with varying levels of seriousness. They appeal to our resourceful instincts and feel very concrete.

Unfortunately, living on Mars will be nothing like camping or building an old fashioned ranch, as Andy Weir ultimately makes clear in his novel “The Martian”. The character Mark Watney is every bit a rugged resourceful type, and yet just surviving an extra year on the surface is a difficult and realistic challenge. If one takes this situation to its logical conclusion – an attempt at indefinite survival – the result is inescapable. Death by systems failure.

So if pioneering agriculture is a bad analogy, are there better ones? I think there are. The key difference is that growing plants somewhere that plants already grow is not that difficult. Mars’ surface is a frozen poisonous irradiated cratered hellscape. It is as hostile to human life as any of the places on Earth not yet overrun by humans.

What are those places? Mountains about 5500 m (18,000 feet). Deeper than 10 m in the ocean. The open ocean. Antarctica. The atmosphere above 40,000 feet. Volcanic areas. Very steep cliffs. What do all these places have in common? They are places that humans rapidly die without both advanced technology and technical skills, usually embodied by the following of checklists. Like the astronaut, the SCUBA diver and mountain climber both share an appreciation for procedural risk reduction and reliance on specialist gear to keep them alive. Importantly, this specialist gear is highly non-trivial to make from scratch.

Let us develop these analogies further. Let’s buy a retired aircraft carrier and cruise to Antarctica, where we will tie it to the ice a respectful distance from any penguin rookeries. Putting on our logistics hats, the challenge is to select a crew and any relevant machinery, strand them on this carrier, and keep them alive for as long as possible.

Let’s employ some physics-based reasoning to understand how this may play out. Assuming we work out how to keep the living areas heated, to grow crops on the deck during summer under greenhouses, and to keep all the mechanisms running with copious spare parts and a nice machine shop, what will limit the duration of this experiment? How can the system break down in a way that no carrier crew, no matter how motivated and skilled, can prevent? Ultimately the hull will rust and deteriorate and the carrier will sink. The hull can’t be repaired without a supply chain for steel, and there is no source of iron on the Antarctic ice.

This is a bit of a downer, but ultimately any Mars city will either need total ongoing replacement of its constituent parts, or the ability to make them locally, and in abundance. Abundance is the key point. It’s not difficult to imagine a small Mars base with a few hundred skilled engineers and technicians and a bunch of machines, essentially able to make anything given enough time. While impressive, that’s not so different to the aircraft carrier example. What is needed is not only a prototyping ability, but a full manufacturing capacity. The ability to build new things much faster than the old ones break down and wear out.

When we think of people doing jobs on Mars, it is less the careful, time consuming repair of some fancy widget, than the incremental improvement in manufacturing rate for hundreds of new widgets. It is simply an inefficient waste of labor to perform meticulous bespoke surgery on machinery. There is too much to do, and human time is too valuable.

How valuable? It is important to remember that while transport increases the cost of having things on Mars, this increase is different for humans and robots. Let’s say that SpaceX can launch cargo to Mars for $5000/kg. A human and their stuff could get a ride for a million dollars. A 200 kg robot would cost the same to deliver, and in almost all cases, the transport cost would exceed the sticker cost of the device. If the robot in question is a generic industrial CNC machine, then the cost of delivering it to Mars increases its cost by about a factor of 10. But the cost of keeping a human alive on Mars also includes all the cargo they need to be delivered over time, the life support and living space that needs to be provided, and the cost of the supporting operations team on Earth. To pick a round number, let’s say the cost of labor on Mars is 100 times higher than on Earth, or 10 times higher relatively than a robotic CNC machine. A highly qualified engineer’s salary + overhead in Los Angeles, 2020 might cost $250,000/year. On Mars, relocation and logistics costs would inflate that to $25m/year, per person.

This is why, I think, concerns over indentured servitude are overwrought. Salary is a tiny fraction of this cost, so immediately we can see that, to the extent that salary is motivating, it’s unlikely to cost the managers of the project much in the grand scheme of things. More generally, the overhead cost is a way of accounting how many other salaries are needed to support the labor of the person on the (Martian) ground.

How does this change the work environment? Consider the work environment of humans with similar levels of overhead. This could range from oil rig specialists to heads of state. The aim is to extract the maximum value of labor over a long period of time. This means optimizing efficiency, ergonomics, and scheduling. For technicians, it means having the right tools and parts where they’re needed, as well as LEGO-style parts that integrate with a minimum of fuss. For software engineers, it means working remotely back on Earth, where the air is free. Overall, it means responsive management and emphasized productivity incentives. It also means the ability to switch between different tasks as skill demand and interest shifts.

A perfectly executed workers paradise is a necessary, but not sufficient, condition for self-sufficiency. Even if the Mars city was full of healthy, well-trained technicians all swinging wrenches at optimal efficiency, the city would still be unable to avoid systems failure. Mars is simply too hostile an environment for a city of any size to survive while relying only on human labor. There is way too much to do.

The industrial revolution is a big deal because, for the first time, the mechanical output of a civilization was no longer limited by how much food the human and domestic animal population could digest and convert, rather inefficiently, to motion. Instead, machines powered by burning coal, oil, and more recently solar electricity, can perform mechanical labor. This is a big deal – just compare the process of agriculture before and after tractors were invented. The total mechanical output of developed nations has grown by about a factor of 50 since the industrial revolution. That is, of the total mechanical power produced, only 2% of it is derived from human muscles, so our industry performs 50 times more work than humans could alone.

We may have to improve this force multiplier by another factor of 50 to achieve self sufficiency on Mars with fewer than a million people. What could this look like? Humans operating tractors and bulldozers than are 50 times bigger? To an extent, yes, but not all production requires a really big digger. Instead, it needs additional layers of abstraction between human labor and production.

The best historical example of this that I’m aware of occurs in software engineering. Since the invention of computers, programming languages have become more powerful through the use of abstraction. Coders using Python don’t (usually) need to worry about hardware-specific CPU instructions, memory management, or specifics of matrix inversion. Instead, they have compilers and interpreters that, combined with hundreds of thousands of user-generated libraries containing condensed human thought ready to be recycled to solve the next problem, allow the coder to quickly get to the point. This is what the robot revolution looks like. A single programmer can solve a problem in exponentially less time than ever before.

Hundreds of skilled technicians pushing parts on lathes and mills is the assembly code of manufacturing. It’s better than filing chunks of rock by hand, just as assembly was better than buildings full of human computers with slide rules. And to some extent, the first humans on Mars will be performing construction and assembly by hand. To stay on the trajectory of population vs self-sufficiency, however, human hands will need to rapidly retreat up the process chain. What does the C of manufacturing look like? The C++? The Python? What do factories look like when labor is really expensive? To some extent we can answer that by comparing factories making identical competing products in low wage and high wage countries. Higher labor cost means fewer humans, higher productivity, newer machinery, and a lot of industrial automation. When we build factories on Mars, we are building factories of the future.

Let’s jump back to Mars. There’s a big city, SpaceX Starships are landing thousands of tons of cargo and new people every two years. The city is expanding rapidly using the latest technology and is climbing the ladder of self sufficiency. How does this work on a day to day basis? More people doesn’t automatically lead to better specialization and process efficiency.

Let’s say that every two years, a new arrival of cargo and people doubles the population of the city. In order to scale productivity at the same rate, essentially all the new arrivals need to be employed building a new industrial capability, while the existing Martians need to double, during the preceding two years, the productivity of their industries to meet demand from the new arrivals. Achieving this scaling in lockstep over three or four orders of magnitude is a preposterously understated challenge.

At first, workers might directly assemble machinery. Then build an automated assembly line. Then double the production rate on that line. Then automate production of lines. Then automate procurement of parts and disposition of products. By the end state, a handful of old-timers will be perched atop a pyramid of abstracted labor, having automated their own labor all the way up from the ground level. Robot-constructed robots will autonomously mine ore, produce billets of thousands of grades of industrial metals, and oversee their use to produce the millions of different products that fill out the McMaster-Carr and Amazon catalogs.

At first, a small Mars city will be preoccupied with production of living spaces and mass-heavy bulk materials, such as fuel, oxygen, water, concrete, steel. As the city grows, the make/buy trade will favor the local production (vs import) of steadily more complex, relatively low weight products, including plastics, food, textiles, electrical machinery, chemicals, pharmaceuticals, and eventually integrated circuits and baby humans, both of which are extremely labor intensive and relatively easy to import.

By the time the city has reached its winning condition, so much time and technology will have passed that it is even more difficult for anyone to predict what human labor conditions might be like by then. Extrapolation suggests that transport costs to Mars will fall, even as local production improves. Economic forcing will also cause travel times to fall. Where once high costs might justify only specialist migration and one-way trips, automated production of Starships and carbon-neutral fuel production may open the solar system to anyone who can spare a week’s salary.

As for Mars, a century of terraforming efforts combined with exponentially increasing industrial capacity will bring forth the initial promise of the industrial revolution, a post-scarcity cornucopia on a greening world with limitless frontiers.

11 thoughts on “What would it be like to work on Mars?

  1. Not bad. Makes lots of good points.
    I would like to see a more quantitative critical path version based on a reasonable evolutionary software simulation. I would also like to see a ladder of self sufficiency in detail. What are the required tools to make tools? How much redundancy should there be in a kilowatt modular nuclear fission reactor system (assuming Lockheed is busy using their compact fusion reactors to take E.T. home) and how many units for spare parts? Once propellant and water production were online, could methane be used as an emergency source of heating and electrical energy? I don’t expect a detailed ladder of self sufficiency to be any more accurate than the proposed schedules in Energy, Earth and Everyone and Ho-Ping by Medard Gabel from the mid to late 1970’s and early 1980’s, but it would be something external that could be worked with by others. Are there any scenarios in current MMORPG’s that address these issues?
    The professional writer, L. Ron Hubbard, A former Naval Officer once pointed out that life at sea required a higher level of anticipatory attention to detail than on land and a more cooperative teamwork effort, all just to survive on the open sea and through infrequent storms. Buckminster Fuller used to say something similar and wrote about Navy Yard Industrialization and industrialization in general as a global cooperative enterprise and an external metabolic system. Mr. Hubbard formulated the idea that survival of an individual was accomplished through several dynamic thrusts at the same time and not just the individual urge to survive. So, I could envision a team-spirited community culture that would outlaw monopoly control of the life support system by an individual and enforce that internally within their group.
    There is probably something like this among mountain climbers and deep sea diver welders.
    The tragedies on Mt. Everest might be useful case histories.


  2. the premise for Martian settlement always seems to be on the model of a city, not of villages, hamlets and ranches. It brings up all the problems of concentration we see in terrestrial cities.

    In Earthly village life, we have a few one-off investments such as a wood boiler or a water purification plant, then repeated and more massive inputs such as wood and water.

    IMO, the same principle should apply on the Moon or Mars. Having brought in the solar panels and the water purification system, then the inputs are repeated and far more massive.

    At the start, habitats are imported but as time goes on, it becomes possible to manufacture building blocks and glass. Similarly, fertilizers and soil substitutes would be needed. Later, it becomes possible to clean local regolith, clays and whatever, to produce the necessary support for greenhouse food production. Similaly, imported animal proteins should later be replaced by fish and insect farming with their related transformation processes.

    Heavy equipment needs to be designed such that the wear elements are as light as possible to make their replacement cheap in terms of mass. The first step towards building machines such as quarry vehicles should be chasis elements. Bearings and hydraulic parts would still be imported.

    The annual imported mass diminishes over time and after many decades the required inputs are only things like microcircuits and medicines. Each new step corresponds to a higher entry barrier or technological threshold level.

    During this time, technical progress on Earth should cause the threshold level for each product to fall.

    I’m quite critical of the author’s aircraft carrier analogy:

    Let’s buy a retired aircraft carrier and cruise to Antarctica, where we will tie it to the ice a respectful distance from any penguin rookeries. Putting on our logistics hats, the challenge is to select a crew and any relevant machinery, strand them on this carrier, and keep them alive for as long as possible.

    Surely, we need to drive the ship into the beach so it no longer needs to float. We’d then be building out onto land and replacing the ship’s structure with stone dwellings.

    Also, the question is not how long the experiment could be run, but how long it takes to reduce the annual inputs to a reasonable level.

    The analogy is remenicent of the errors of Biosphere 2. The thing not to do is to set up the experiment and shut the doors. Not only do inputs need to be progressively scaled down, but mistakes need to be corrected whilst still working on a small scale. The whole system needs to iterate, much like the Starship vehicle design.

    Sociological aspects will become apparent and require adaptations to the habitats and surrounding infrastructure. The behavioral side of a young generation (not selected for the trip!) will affect things too. There will be unexpected physiological changes. For example, people may be taller. Design is easier to adapt if its not rigid at the outset. These adaptations are probably easier to effecutate on a decentralized village structure than the Artemis novel (Andy Weir) that portrays a monolithic moonbase. The same should apply on Mars.

    When expecting the unexpected better be a little scattered. What happens in case of fire, plague and several other horsemen of the Apocalypse? A decentralized system is likely more resilient in all cases.

    City vs villages. I rest my case!

    BTW. This is a copy of a comment I made here:



  3. Nice post! Much food for thought.

    I think your argument makes sense that the wages of workers on Mars would be marginal, thus making indentured servitude unlikely, however, what happens over the long-run?

    Once self-sufficiency is achieved and once Martian growth is based mostly on local reproduction, as opposed to migration, the overhead of transport would no longer be a substantial factor in the total cost of labour, and wages would no longer be marginal.

    Given that this would occur after self-sufficiency and the major initial migration, again indentured servitude (as in KSR’s Mars Trilogy and as occurred during European migrations to the Americas) again would be unlikely, however, life on Mars over the long-run could still be inherently prone to tyranny.

    For a deeper dive the subject I suggest Charles S. Cockell’s “Human Governance Beyond Earth” and “The Meaning of Liberty Beyond Earth”.

    His basic argument is it’s hard to resist tyranny if the Tyrant controls the life-support systems.

    To summarize, it would seem tyranny is unlikely pre-self sufficiency, but possible post-self sufficiency, so the Martian settlers will have a finite window to establish the physical infrastructure and governance institutions necessary to preserve their freedoms over the long-run.


  4. I’m not convinced that terraforming is the way to go.

    Even if we do, it’s one small planet. To really have limitless frontiers, we need to live free in space, in self-sufficient artificial worlds that can drift off into interstellar space on trajectories that will have them arrive at new systems before their resources are exhausted.


  5. Regarding city vs. village model. There is a reason that cities were created in the first place: because they are far more efficient. It takes less pluming, wiring, paving, wall material, transportation costs, etc. The only reason that villages existed before cities was because a village is organized around inefficient agricultural methods that have very high labor inputs. Note how small towns in the US are dwindling and blowing away? Because it is too expensive to maintain compared to larger, more dense cities. With mechanized farming, less than 4% of the available labor force is used today compared to ~90% back when villages dominated. The same would be true on Mars… and much of that would be automated beyond what we do on Earth today.

    No, when building out a Mars colony, the city model is the one to follow.

    –Seaby Brown, Author of “All The Stars Are Suns”

    Liked by 1 person

  6. I’m still inclined to guess that a lot of the work on Mars, in the early stages, will be done by semi-autonomous remotely operated machines overseen by humans on Earth, with the processes designed around the need to tolerate a half-hour delay when something goes wrong. Where it isn’t feasible to have a process tolerate a half-hour of light-speed lag plus human response time, I still think a lot of the work will be done by semi-autonomous remotely operated machines, and the humans on Mars will spend most of their time overseeing those. Pressurizing a workspace to human habitability is costly, and my guess is that it will be unnecessary for most stuff.

    I think your point in other posts, that there’s a big difference between being able to do anything and being able to do everything, is spot-on. But I don’t think the solution is to send extremely large numbers of humans very early in the development of a Martian city. I haven’t tried to quantitatively check out what it takes to support a human in space, but the ISS is insanely expensive and houses a tiny number of people. Of course it will be substantially less expensive per person to support people on Mars, both because of economies of scale and because of better design. But my guess is that even the marginal cost won’t be all that many orders of magnitude less, because it takes a lot to keep a human alive. So I’m guessing that there will be more that needs to be done than will be affordable to have humans do locally.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s