Heterogenous v. homogenous manufacturing

For some time, I've come to the conclusion that would not actually be possible for us to establish a colony on another planet — Mars, for example — which is independent of Earth with our existing systems of technology.

I don't just mean with our existing technology; obviously, we would need many technological advances beyond what we currently have to make such a thing happen. Rather, I believe that our entire system of technology is fundamentally unsuitable for such a task. This isn't something that can be rectified simply by creating more contemporary manmade technologies.

The reason for this is simple: it seems apparent that it is simply not possible to solve the spare parts problem with any technology we possess. This is not necessarily a problem with a colony if we accept it being dependent on Earth for a supply of spare parts. Even then, the logistics of delivering such parts in a timely fashion poses serious issues. But to form a colony which is entirely independent of Earth, it poses a seemingly intractable problem.

The spare parts problem is as follows: any system for the colonisation of a planet such as Mars obviously relies on immense amounts of artificial, manmade technology, even for basic things such as life support. This technology is highly complex and the sum total of technology found in such a colony will comprise an almost innumerable array of constituent parts. Any of these parts could potentially fail at any time, therefore, there needs to be a way to replace any one of them.

At this point we hit a problem, which relates to how complex technologies are manufactured on Earth. Fundamentally, all of our existing systems of technology are predicated, at their core, on the notion of the factory. The problem with factory-made technologies is that the entire model of factories does not transfer viably to a space colony.

The reason for this is simple: a factory, fundamentally, is intensive in terms of the land area consumed, in terms of its costs of construction and maintenance, and in terms of its total weight. Yet a factory, once created, can only really create one kind of thing. While it is not necessary for every part produced by a factory to be literally identical, a factory designed for producing screws can't be used to produce silicon chips. In fact, the segmentation of factory capabilities is far more narrow; silicon fabs, in particular, are highly specialised to a particular market, and any given silicon design can in general only be fabricated by the specific silicon fab it was designed for.

The only reason the factory model of production is viable is because the immense capital and operational costs of building and running a factory can be amortised over global or regional demand for the products produced. What is clearly infeasible is to construct an entire factory for a specific kind of part on Mars just in case a replacement part of that kind is needed. The annual demand for the parts of such a factory on Mars, solely for replacements, could be less than 1 unit, even.

In other words, the factory model of mass production to amortise high infrastructure costs fundamentally can't work in a space colony. It would surely be considered impractical to construct even just one silicon fab on Mars. This problem is then compounded even moreso when considers that the sum total different kinds of parts found in the totality of a Mars colony could be ten thousand, a hundred thousand, a million, originating from probably at least a thousand different factories across the world specialised to make the particular parts they produce. It is beyond economical and practical viability to reconstruct even one contemporary factory on Mars, let alone a thousand, all of which will go 99.9% unused on the rare chance that a part fails and needs to be replaced. Also, none of this even begins to consider the labour issues involved.

Of course, the comment might be raised that factories are large in part because they are designed to facilitate high output. However, the problem here is that we actually don't know how to scale down production. Silicon chips are the canonical example here; even during the R&D of a silicon chip, we don't actually have any way to make silicon chips cheaper than the mass production process. We don't actually know how to make a compact silicon fab with lower output at the state of the art node; oddly enough, the mass production model for silicon chips is the only model we have.

Essentially, what it comes down to is this: pretty much all physical technology we manufacture is made in a factory based around the assumption that we can specialise a large quantity of land, materials and labour for a very specific manufacturing task because we can amortise that cost over a large global demand. And many of those manufacturing techniques aren't actually trivially scaled down, in terms of landmass, mass or trained, specialised labour consumed just because you want lower output. The manufacturing system for practically all manmade technology, of the entirety of global human civilization, is based, in other words, around a process of heterogenous manufacturing: the processes, equipment and materials needed to manufacturer some type of part A, are completely and utterly unlike those processes, equipment and materials needed to manufacture some other type of part B.

The property of process dependence confounds this issue. By process dependence, I refer to the fact that, ultimately, any manufactured item ever produced contains some sign in it of the process that created it. In other words, even if you have two manufacturing processes, process A and process B, which can produce the same, apparently interchangeable part, part X, in actuality, these different processes are not producing a literally identical arrangement of atoms. In practically all cases, some distinction between process A and process B will fingerprint itself into the output, and produce a detectable difference in the part produced, even if that difference might be inconsequential in terms of the part's function.

In other words, pretty much every manufacturing process on Earth has no truly interchangeable substitute. For every given part manufactured on Earth, there is approximately only one manufacturing process that can produce it.

Of course, where the functional requirements of a part are not too tight, total equivalence of a manufacturing process may not be needed. A CNC machining facility on Mars might get you a long way in terms of spare parts, even if you need to machine parts which usually wouldn't be machined; and so on. But there are many kinds of manufactured part where process equivalency is in large part a pipe dream; again, silicon fabs pose an obvious example.

To reiterate: manufacturing on Earth is based on a model in which we have a different factory for more or less every different type of part. We can refer to this as a heterogenous manufacturing system. But at the same time, it is impossible for us to construct, or even hypothesise of a replacement system of production whereby, for example, one machine can construct any part, because of the problem of process dependence. 3D printers may exist — but you will never make a state of the art silicon chip with one.

In other words, we're trapped, in a sense, in that we've found a sort of local maxima in our technological evolution, that based on factories. We put practically all of our investment in research into new physical technologies into this kind of manufacturing, but not only can this technology never become viable for independent space colonies, there is no obvious way for it ever to be made so viable. It is, in terms of the specific challenges of space colonisation, a technological dead end. The state-of-the-art silicon chip, a pinnacle of human achievement — yet a technological dead end, in the face of the extraordinary challenges of space colonisation.

It seems to me that the only viable method of solving the spare parts problem, and thereby facilitating viable, independent space colonies which are not dependent on Earth, is to construct all aspects of such a colony using a homogenous manufacturing system.

The most obvious example of a homogenous manufacturing system, of course, is biology. I like to say that biology is the ultimate “field system”: animals, for example, are essentially designed to operate “in the field” in all ways, without any dependence on any kind of centralised infrastructure whatsoever. Compare this with basically all modern technology and human society, which is wholly dependent on massive amounts of centralised infrastructure, which can only be maintained using very large amounts of land, mass, energy and labour.

By comparison, the non-dependence of biology on infrastructure is extraordinary. Animals are literally designed to work “in the field”, and humans, for example, can eat almost anything. Not only that, but the forces of evolution have seen fit to force you to carry your reproductive equipment around with you wherever you go. Even if you're just nipping to the shops for a bottle of milk and you'll only be out of the house for five minutes, you're forced to take your reproductive equipment with you. This seems needless most of the time, but may make an immeasurable difference if a meteor ends up falling on your house during that five minutes. You can escape from death by a millisecond and still possess the means of reproduction. So no matter where you find yourself “in the field”, without any centralised infrastructure to support you at all, you can potentially eat almost anything and reproduce without limit, expanding the size of human existence until all available resources are harnessed.

Of course, biology is even more amazing as a homogenous manufacturing system. Fundamentally, biology is a homogenous manufacturing system in the sense that everything is made of cells, and organisms can produce more cells of whatever type needed to meet requirements. There is essentially only one process for producing more of your body, and any existing part in your body — a cell — can carry it out. This is the critical capability for a homogenous manufacturing system. The size both of an organism, and a colony of organisms, can be scaled up and down flexibly without limitation. Not only that, animal biology possesses an extraordinary capacity to repair itself when damaged; yet it can do this while sustaining itself from seemingly almost any food source. Given nothing but protein and minerals, energy and water, all life can be maintained, repaired, thrive, and reproduce. Compare this with the inflexibility of a typical manmade technology, which can only consume necessary resources, whether energy or materials, in one very specific preconceived form; manmade technologies are in this way infinitely more inflexible and unadaptable than even the simplest of animals.

As I said: biology is the ultimate field system. And a field system is exactly what we would need to create an independent colony on Mars. For this reason, therefore, I believe that all manmade technologies to date, which are based around manufacturing, are irrecoverably unsuited to the challenge of creating an independent space colony; I believe that in order for a space colony to be fully independent of Earth, every aspect of its physical construction would need to have its basis in a homogenous manufacturing system.

Of course, the concept of “bioships” — alien ships made of animal-like flesh — is a common theme in science fiction. However, when I say that a homogenous manufacturing system would be needed, I don't necessarily mean actual biology as it exists on Earth, but simply something which is, at its core, biology-like, in that its method of manufacturing more things is homogenous, and thus does not require the specialisation of large amounts of mass and space to making specific kinds of things. Moreover, biology itself is not constrained to manufacturing soft tissues; for example, it is capable of producing things like elephant ivory. (Presumably, people would rather occupy a building made of ivory than a building made of animal flesh.) The possibilities of homogenous manufacturing systems seem potentially endless, and the capacities for self-repair, and the flexible use of whatever resources are available to maintain (a space station's equivalent of) “homeostasis” extraordinary. A modern jet engine is an extraordinarily reliable device; yet it cannot hold a candle to the human heart, which can run for a hundred years without fault, and for that matter without downtime or maintenance. There is no technology in possession of humanity which would allow us to produce such a thing. Jet engines require constant maintenance, and even then last for far less than a hundred years.

Of course, the problem is that we have barely the slightest of ideas about how we might go about constructing such a thing. We are essentially talking about the construction of artificial life. Of course, science fiction is also abound with the common theme of humans successfully figuring out how to do this — and immediately losing control of it, in the form of the “grey goo” scenario. It's hard to deny the significant likelihood of something like this happening if we did figure out how to create such a thing.

Yet at the same time, we still feel strangely powerless when it comes to the biological domain. The human study of biology, is the greatest reverse engineering project ever undertaken by humanity. It has been running for many centuries now, and yet it feels like we've barely got anywhere. The complexity of biological lifeforms is incomprehensible, and challenges our ability to understand it. Nobody has enough confidence in our contemporary understanding of the human body, for example, to approve a new drug purely on the basis of theoretical understanding; we don't have enough confidence in our own understanding and models to forego testing. (Compare with civil engineers designing bridges, or the formal verification of avionics software for a space mission that will only ever be run once.) Our powers over the biological form remain crude: surgery, which can only perform the most macroscopic of interventions (consider the scale of a surgeon's hand to that of a single cell), and drugs, which are indiscriminate, and most of which largely amount to putting a thumb on the scales of one of the body's trillions of PID controllers because it has been given the wrong setpoint, and we don't know why, and we don't know how to fix the underlying cause.

Consider the fact that I have (thankfully now mild) asthma; while I probably owe my life to asthma medication, which I am thus very grateful for, a cure remains elusive. Yet ultimately the state of having asthma exists only as information somewhere in my body, a product of an adversarial respiratory environment when I was a baby (that is, living in a polluted city). Theoretically, this information could be changed; of course, we have no idea how to do this. Instead, therefore, we are forced to crudely override the consequence of that information, via bronchodilatory medications, because we don't know how to address the root cause.

Indeed, other organisms with their own peculiar specialised capabilities can have greater capabilities than all of human technology in this regard. There is, for example, a parasite which cures asthma for as long as it remains resident in the body. I recall the story of one person with asthma who deliberately went to Cameroon to expose himself to this parasite for the purpose of curing his asthma, after being unable to find any western doctor willing to deliberately infect him with it.

In short, biology remains, in the large part, hopelessly beyond our control and only partially within our understanding. Rectifying this would be an extraordinary step change and probably constitute, or immediately lead to, a Singularity, so of course, one should potentially be careful what one wishes for. Though at the same time, it doesn't seem that any breakthroughs here are on the horizon. The construction of a homogenous manufacturing system that would render an independent Mars colony viable, remains elusive.