Centauri Dreams regular Keith Cooper gives us a look at self-replication and the consequences of autonomous probes for intelligent cultures spreading into the universe. Is the Fermi paradox explained by the lack of such civilizations in the galaxy, or is there a far more subtle reason? Keith has been thinking about these matters for some time as editor of both Astronomy Now and Principium, which has just published its fourth issue in its role as the newsletter of the Institute for Interstellar Studies. Intelligent robotic probes, as it turns out, may be achievable sooner than we have thought.

by Keith Cooper


There’s a folk tale that you’ll sometimes hear told around the SETI or physics communities. Back in the 1940s and 50s, at the Los Alamos National Labs, where the first nuclear weapons were built, many physicists of Hungarian extraction worked. These included such luminaries in the field as Leó Szilárd, Eugene Wigner, Edward Teller and John Von Neumann. When in 1951 their colleague, the Italian physicist Enrico Fermi, proposed his famous rhetorical paradox – if intelligent extraterrestrial life exists, why do we not see any evidence for them? – the Hungarian contingent responded by standing up and saying, “We are right here, and we call ourselves Hungarians!”

It turns out that the story is apocryphal, started by Philip Morrison, one of the fathers of modern SETI [1]. But there is a neat twist. You see, one of those Hungarians, John Von Neumann, developed the idea of self-replicating automata, which he presented in 1948. Twelve years later astronomer Ronald Bracewell proposed that advanced civilisations may send sophisticated probes carrying artificial intelligence to the stars in order to seek out life and contact it. Bracewell did not stipulate that these probes had to be self replicating – i.e able to build replicas of themselves from raw materials – but the two concepts were a happy marriage. A probe could fly to a star system, build versions of itself from the raw materials that it finds there, and then each daughter probe could continue on to another star, where more probes are built, and so on until the entire Galaxy has been visited for the cost of just one probe.

The combination of Von Neumann machines and Bracewell’s probes made Fermi’s Paradox all the more puzzling. There has been more than enough time throughout cosmic history for one or more civilisations to send out an army of self-replicating probes that could colonise the Galaxy in anywhere between three million and 300 million years [2] [3]. By all rights, if intelligent life elsewhere in the Universe does exist, then they should have colonised the Solar System long before humans arrived on the scene – the essence of Fermi’s Paradox. The conundrum it is about to be compounded further, because human civilisation will have its own Von Neumann probes within the next two to three decades, tops. And if we can do it, so can the aliens, so where are they?

To Build a Replicator

A self-replicator requires four fundamental components: a ‘factory’, a ‘duplicator’, a ‘controller’ and an instruction program. The latter is easy – digital blueprints that can be stored on computer and which direct the factory in how to manufacturer the replica. The duplicator facilitates the copying of the blueprint, while the controller is linked to both the factory and the duplicator, first initiating the duplicator with the program input, then the factory with the output, before finally copying the program and uploading it to the new daughter probe, so it too can produce offspring in the future.

‘Duplicator’, ‘controller’, ‘factory’; these are just words. What are they in real life? In biology, DNA permits replication by following these very steps. DNA’s factory is found in the form of ribosomes, where proteins are synthesised. The duplicators are RNA enzymes and polymerase, while the controllers are the repressor molecules that can control the conveyance of genetic information from the DNA to the ribosomes by ‘messenger RNA’ created by the RNA polymerase. The program itself is encoded into the RNA and DNA, which dictates the whole process.

That’s fine for biological cells; how on earth can a single space probe take the raw materials of an asteroid and turn it into another identical space probe? The factory itself would be machinery to do the mining and smelting, but beyond this something needs to do the job of constructing the daughter probe down to the finest detail. Previously, we had assumed that nanotechnology would do the duplicating, reassembling the asteroidal material into metal paneling, computer circuits and propulsion drives. However, nanotechnology is far from reaching the level of autonomy and maturity where it is able to do this.

Perhaps there is another way, a technology for which we are only now beginning to see its potential. Additive manufacturing or, as it is more popularly known, 3D printing, is being increasingly utilised in more and more areas of technology and construction. Additive manufacturing takes a digital design (the instruction program) and is able to build it up layer by layer, each 0.1mm thick. The factory, in this sense, is then the 3D printer as a whole. The duplicator is the part that lays down the layers while the controller is the computer. It’s not a pure replica in the Star Trek sense, but it can build practically anything, including moving parts, that can otherwise only be manufactured in a real factory.

Gathering Space Resources

3D printing is not the technology of tomorrow; it’s the technology of today. It’s not a suddenly disruptive technology either (well, not in the sense of how it has gradually evolved), having been around in its most basic form since the 1970s and in its current form since 1995. Rather, it is a transformative technology. The reason it is gaining traction in modern society now is because it is becoming affordable, with small 3D printers now costing under $2,000. Within a decade or so, we’ll all have one; they’ll be as ubiquitous as a VCR, cell phone or a microwave. This will have huge consequences for manufacturing, jobs and the economy, potentially destroying large swathes of the supply chains from manufacturing to the purchaser, but, whereas the factory production lines on Earth may dry up, in space new economic opportunities will open up.

As spaceflight transitions from the domain of national space agencies to a wider field of private corporations, economic opportunities in space are already being sought after, including the mineral riches of the asteroids. One company in particular, Deep Space Industries, has already patented a 3D printer that will work in the microgravity of space [4] and they intend to use additive manufacturing to construct communication and energy platforms, space habitats, rocket fuel stations and probes from material mined from asteroids and brought into Earth orbit. For now, they envisage factory facilities in orbit and the asteroids mined will be those that come close to Earth [5]. Nevertheless, it has already been mooted that astronauts on a mission to Mars will be able to take 3D printers with them and, as we utilise asteroids further afield, we’ll start to bundle in the 3D printers with automated probes, creating an industrial infrastructure in space, first across the inner Solar System and then expanding into the outer realms.

Archimedes concept

Image: A ‘fuel harvestor’ concept as developed by Deep Space Industries. Credit: DSI.

Here’s the key; these 3D printers that will sit in orbit and are designed to build habitats or communication platforms, could easily become part of a large probe and be programmed to just build more probes. All of a sudden, we’d have a population of Von Neumann probes on our hands.

Without artificial intelligence, the probes would just be programmed automatons. They’d spend their time flitting from asteroid to asteroid, following the simple programming we have given them, but one day someone is inevitably going to direct them towards the stars. This raises two vital points. One is that if we can build Von Neumann probes, then a technological alien intelligence could surely do the same and their absence is therefore troubling. And two, Von Neumann probes will soon no longer be a theoretical concept and we are going to have to start to decide what we want them to be: explorers, or scavengers.

A Future Beyond Consumption

It seems clear that self-replicating probes will first be used for resource gathering in our own Solar System. Gradually their sphere of influence will begin to edge out into the Kuiper Belt and then the Oort Cloud, halfway to the nearest stars. That may not be for some time, given the distances involved, but when we start sending them to other stars, do we really want them rampaging through another planetary system, consuming everything like a horde of locusts? How would we feel if someone else’s Von Neumann probes entered our Solar System to do the same? Once they are let loose, we need to take responsibility for their behaviour, lest we be considered bad parents for not supervising our creations. That would not be the ‘first contact’ situation we’ve been dreaming of.

On the other hand, Bracewell’s probes were designed for contact, for communication, for the storage and conveyance of information – a far more civilised task. But standards, however low, can be set early. If our Von Neumann probes are only ever used for mining, will we be wise enough to have the vision in the future to appropriate them for other means too? It seems we need to think about how we are going to operate them now, rather than later after the horse has bolted.

And perhaps there lies the answer to Fermi’s Paradox. Maybe intelligent extraterrestrials are more interested in making a good first impression than the incessant consumption of resources. Perhaps that is why the Solar System wasn’t scoured by a wave of Von Neumann probes long ago. The folly of our assumption is that we see all before us as resources to be utilised, but why should intelligent extraterrestrial life share that outlook? Maybe they are more interested in contact than consumption – a criticism that can be levelled at other ideas in SETI, such as Kardashev civilisations and Dyson spheres that have been discussed recently on Centauri Dreams. Perhaps instead there is a Bracewell probe already here, lurking in in a Lagrange point, or in the shadow of an asteroid, watching and waiting to be discovered. If that’s the case, it may be one our own Von Neumann probes that first encounters it – and we want to make sure that we make the right impression with our own probe the day that happens.


[1] H Paul Schuch’s edited collection of SETI essays, SETI: Past, Present and Future, published by Springer, 2011.

[2] Birkbeck College’s Ian Crawford has calculated that the time to colonise the Galaxy could be as little as 3.75 million years, as described in an article in the July 2000 issue of Scientific American.

[3] Frank Tipler’s estimate for the time to colonise the Galaxy was 300 million years, as written in his famous 1980 paper “Extraterrestrial Intelligent Beings Do Not Exist,” that appeared in the Royal Astronomical Society’s Quarterly Journal.

[4] Deep Space Industries 22 January 2013 press announcement.

[5] Private correspondence with Deep Space Industries’ CEO, David Gump.