Self-Replicating Machines are the Next Big Thing

What About Progress?

Where” many people ask “is my flying car?

Progress is the idea that, in both the short and long term, things get better.  For much of human history, progress meant better tools for farming and hunting which allowed more people to live free from hunger.  It meant the development of strong nation-states that freed people from the horrors of civil war.  It meant medicine, urbanism, industrialization, ships, trains, cars, planes, computers, and smartphones.  It meant that people can live longer and better lives while working less and seeing more of the world.


I’m sure this is exactly what you had in mind

There is a growing feeling in the most advanced countries that progress has slowed or even stopped.  Life expectancy and income for some is lower than it was for their parents.  The growing threat of climate change makes others question the foundations of industrial civilization itself.

There is another way.

Human labor is the magic ingredient that provides us with everything we have, and progress ultimately depends on increasing people’s standards of living without necessarily increasing the amount of human labor they provide.  This improvement is frequently in terms of material goods: More living space, an air-conditioner, better food, or more clothes.  It can be a service: Healthcare, art, or a taxi ride.  It can be structural: Democracy, disease prevention, or nondiscrimination laws.

Self-replicating machines can only provide the first of these.  But they can do so in such an effective way that they will massively increase the standard of living of all people.  In doing so they will make it possible for us to fulfill even our most outlandish dreams.


We’re much better off now.

Globally, economic growth is about 2%-3%.  At this rate, the world economy will double in size every 25-35 years.  Reducing this doubling time is the most basic way in which self-replicating machines (SRMs) will help humanity:  With short reproduction times and efficient, labor-free operation, our ability to provide for ourselves will increase much faster than it ever could traditionally.

What Can Self-Replicating Machines Do?

Although it’s cool, a machine that can build a copy of itself is of no intrinsic value.  SRMs in the abstract are as economically useful as plants, animals, or bacteria in the wild: Interesting, but not of any material benefit to people.

In order to build a copy of itself, any SRM will have a wide variety of capabilities.  It will obtain resources from the surrounding environment; it will convert these resources into usable materials; it will form these materials into usable parts; and it will assemble these parts into a self-replicating machine.  It will need to generate electrical and thermal energy, dispose of waste materials, and create structures to house its machinery.  Each SRM will be massively complex and will contain a large number of different components.

In short: If you can build a machine that can build a copy of itself, you can also build nearly anything else.  An SRM will be able to shape materials in a variety of different ways to produce a large number of different components.  It will be a Universal Constructor.

A Universal Constructor is a machine that can build anything.  An SRM, on the face of it, is not this.  There will be a limit to the capabilities of the machine.  Perhaps there will be a minimum feature size it can attain.  Perhaps it will be unable to find and extract certain elements from its environment.  Perhaps there are certain kinds of assembly that it is unable to do.


Artist’s Rendition of an SRM on the Moon

The important point to recognize is that while it is possible to build a machine to do these things, an SRM will not be designed to do so.  What it could do is build a machine that could do these things, or build a machine that could build a machine that can do them, et cetera.  An SRM will necessarily be flexible enough in its abilities that such progress is possible.  All we need to do is tell it how.  This is not so different from the development of human technology:  In the beginning, we could only create things on a relatively human scale.  Now we can create things as small as a few atoms or as large as a cross-continental highway.

The difference is that, with no inputs of human labor and potentially operating on common or otherwise unused land, SRM can do so at a cost approaching zero.

Where Will This Take Us?

There are accomplishments that we can dream of, but which we don’t do because of their prohibitive cost and scale.

It’s possible to dig a tunnel from New York to London on which trains will travel at supersonic speeds.

It’s possible to build every person on Earth a house and allow them to live in it for free.

It’s possible to solve world hunger with robotic farms that create and deliver all the food humanity needs.

It’s possible to have factories in space that create all of the goods that people could want and land them right in front of you as-needed, and also in doing so to create new habitable worlds on which people live.

Self-Replication and Universal Construction are incredibly powerful technologies, especially when combined.  Given time, the only limit to our capabilities will be our imagination and our ability to describe what we want done.


The future of Mars?

Will This Be the End of the World?


SRM is most frequently discussed as part of a “Gray Goo” scenario, where self-replicating nanomachines replicate out of control and turn the whole world into replicators.  Real, near-term SRM will not be like this for the very simple reason that it won’t be made of nanomachines.  Like any large, complex system it’ll be relatively easy to stop an SRM if you want to by damaging its components.  A good design would also include simple but effective safety features such as an on/off switch.

Perhaps SRM will be designed with one small but necessary component requiring human installation.  Perhaps they will be designed to need a small amount of some rare element or compound that they can’t obtain from their environment.

No matter how well they are designed, SRM will have replication times measured in months and years, not milliseconds.  They simply won’t move fast enough to be a serious threat.

How Do We Get There?

There are Five Fundamental Stages through which matter must be transformed in order to create SRM:

  1. Resources: Matter in its state of nature, before incorporation into a machine
  2. Ores: Resources which have been extracted from nature and are ready to be transformed into useful materials
  3. Ingots: Purified materials which can be formed into different parts
  4. Parts: Single components of a machine made from ingots via some process
  5. Systems: Fully functional machines or parts of machines

The Four Fundamental Processes are the classes of technologies that allow matter to move forward through these Stages:

  1. Extraction
  2. Smelting (By analogy to Iron, meaning the processes required to create a simple substance from ore)
  3. Forming
  4. Assembly

Depending on the material being worked with and the properties of the final product, each of these four processes will be done in a number of different ways using different technologies.

I do not mean to oversimplify: What I have described above is a framework for a very challenging endeavor.  However challenging it may be, it is one of the most worthwhile things that we could strive to do.

I propose that the best way to attack this project is to begin at the middle and work outwards.  The middle is ingots of the most useful industrial material the world has ever known: Mild Carbon Steel.

Starting from this ingot, we move both backwards and forwards: How do we smelt this ingot out of the relevant ores?  By which technologies do we turn this steel into useful materials?

Carbon Steel contains just two elements: Iron and Carbon.  Iron ore is normally smelted into Iron using Coal.  However, Hydrogen works just as well, has a larger number of applications, and can be extracted from water by electrolysis.  Carbon could be obtained from charred biomass.  By mixing the two in appropriate quantities at high temperatures, you have Steel.


One way for an SRM to generate power is a Solar Power Tower, which uses arrays of mirrors to focus sunlight on a small area.

There are a number of ways in which Steel can be formed into useful parts, but two particularly useful technologies will likely be Investment Casting and Profile Cutting.  Investment Casting can take advantage of 3D Printing technology to create a complicated shape in 3 dimensions, form a mold around it, and then turn the 3D Printed part into a cast Steel part.  Profile cutting can create a wide variety of 2D shapes in Steel Plates of various thickness.

Each process will require its own mix of parts and materials.  From this it will be possible to add to the lists of Resources, Ores, Ingots, Parts, and Systems as well as to the lists of Extraction, Smelting, Forming, and Assembly processes that will be used in SRM.

Design Philosophy

There are a few ways of thinking that are vital to making SRM technology real and successful.  To my mind, some of the more important ones are as follows:

  1. Sustainability: SRM enables us to realize our ambitions on a global scale but also requires us to think of our actions globally.  Therefore it’s vital to design these machines to have no emissions of Greenhouse Gases or other pollutants.
  2. Minimum Viable Products: Perfect is the enemy of Good Enough.  The first products of any development effort will be closer to proofs-of-concept than to SRM.  Even as we approach SRM, there will be some human labor required to enable self-replication.  This will ideally take the form of cartridges of hard-to-find but important materials (Fluorine or Copper being great examples) loaded onto the machines at the beginning of the self-reproduction cycle.  Long-term it will be desirable to find substitutes for these materials such as using Aluminium in wires instead of Copper and finding ways to make Aluminium without fluorine.
  3. Modularity: SRM is similar in a lot of ways to a living organism.  Like living organisms, different configurations of SRM are better suited to different environments.  Therefore it is important to make it easy to modify the Self-Replicating system to adapt it to different environments.
  4. Commonality of Parts: It’s easier to make 100 of 1 thing than 1 each of 100 things.  The fewer different techniques are required to create a SRM, the faster its replication time will be.

We could live like this if we wanted to.

What Now?

I have laid out the bare-bones outlines of the future we can live in.  What we need to do is to make it happen.  We need to design the subsystems, develop the technologies, and implement them for our own purposes.  We can do this by direct design and by spreading awareness of the revolutionary potential of SRM.

I would encourage anyone who is interested to comment or to contact me.  Talk to people you know about the possibilities of SRM to raise their interest.  SRM is just a dream until we make it happen.

One thought on “Self-Replicating Machines are the Next Big Thing

  1. mgo says:

    It sounds like it should be so easy! Apparently it’s hard in practice. RepRap looks like the state of the art for open source SRMs. RepRap Snappy apparently can produce 72-79% of its own parts, but cannot assemble them. The RepRap community seems to be large and healthy, and it sounds like some of them are working on moving from plastic to metal.

Leave a Reply

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

You are commenting using your 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