We are already midway through exploring the galaxy (probably)

12.16, Thursday 29 Sep 2022

The von Neumann probe is a proposed method for the rapid, automated exploration of the universe.


An interstellar probe is designed such that it can self-replicate. When it arrives at a star system, it hoovers up the local asteroid belt, then uses a built-in 3D printer to print out 1+ copies of itself. Copies then head off to new star systems; rinse and repeat. The spread is exponential.

The lineage! As far as I can tell, this history goes a bit like this…

ONE: John von Neumann is the super genius who came up with both modern computer architecture, programmable computers (disputed), and also the nihilistically practical atomic era defence policy of Mutual Assured Destruction. In the 1940s he came up with cellular automata (e.g. Conway’s Game of Life), and from there the philosophical concept of a (software) machine that could reproduce itself: the universal constructor. Which is a wild idea: the “machine” has to contain both the schematic of the machine plus also the apparatus to recreate the schematic. In code terms it’s a wonder that it could possibly fit.

TWO: Separately, in 1960, Robert Bracewell tackled the problem of how to communicate with advanced societies elsewhere in the galaxy. In Communications From Superior Galactic Communities (Nature; here’s a PDF) Bracewell suggested that, since radio signals would be too faint over great distances, it would be a better idea to send a smart interstellar probe and then open the conversation locally.

THREE: Project Daedalus, 1978, the first full engineering study of what it would mean to actually build an interstellar craft. As previously discussed but here’s the tl;dr: it’s huge. Twice the height of a Saturn V and almost the same in diameter, it has to be huge for speed and for reliability.


A major alternative to both the Daedalus flyby and “Bracewell probe” orbiter is the concept of the self-reproducing starprobe. … In theory, each self-reproducing device dispatched by the launching society would become an independent agent, slowly scouting the Galaxy for evidence of life, intelligence and civilization. While such machines might be costlier to design and construct, given sufficient time a relatively few replicating starprobes could search the entire Milky Way.

(Actually it seems like the general idea emerged in fiction in the late 1970s, and this was the first feasibility paper? I will have to dig into this more…)


Freitas, R. (1980). A Self-Reproducing Interstellar Probe. Journal of the British Interplanetary Society, 33, 251-264.

(Yes this is in JBIS. I love JBIS.)

Freitas’ paper is brilliant and deeeeep.

REPRO (the self-reproducing probe) weighs in at 1,000x the mass of Daedalus itself, 10 billion kg.

I had trouble visualising that and fortuitously found some great comparisons on Quora: 10 billion kg is a little under two Hoover dams, or the total mass of chicken consumed in the US every year. A single probe!

The probe, REPRO, arrives at a star system and drops SEED. SEED has two jobs: to explore, and to build FACTORY. FACTORY is what constructs and launches more REPROs.

The actual reproductive apparatus consists of 13 distinct robot species, including:

  • Chemists
  • Mining (atmospheric, in the atmosphere of Jupiter-like planets): An Aerostat robot floats like a hot air balloon in the jovian air
  • Miners (misc.)
  • Metallurgists
  • Computers with abilities such as lateral thinking and intuition.
  • Fabricators
  • Assemblers
  • Warehousers
  • Crawlers (for surface hauling)
  • Tankers (for fuel transportation)
  • Wardens (manipulators and maintenance)
  • Verifiers
  • Power plants

It is anticipated it would take 500 years for SEED to build FACTORY (step 1), and a further 500 years for FACTORY to duplicate REPRO (step 2). Over 5,000 years, 10 self-reproducing interstellar probes could be constructed and launched.

Sexual reproduction:

At the end of Freitas paper, he points out that the offspring of the von Neumann probes will likely be imperfect replicas. There will be data mutations along the way.

BUT! This provides for the possibility that some mutations may be beneficial. Oh, so we get evolution. But it’s too slow.

Majestically: Freitas then alters the REPRO design to include sexual reproduction, which he calculates will accelerate evolution enough to have a significant impact during the exploration time of our galaxy. There would be, he says, enough time for machine speciation to occur.


Niche specialization is plausible and there is a remote possibility that a simple machine ecology might have time to arise, complete with predators and prey. Sexual starprobe designs may be imagined as REPRO vehicles preprogrammed for target star overlap every few generations. The usual reproduction scenario might then include two starprobes landing on opposite sides of the same jovian moon and jointly engaging in the construction of two FACTORY complexes and two REPRO offspring. Memory Caches could be compared, evaluated and edited “consciously,” offering the exciting possibility of yet faster development by means of intelligent participative evolution.

Aha hahahahahah.

Emphasis mine.

Can we just take a moment to appreciate that this is at the tail end of an engineering feasibility paper. If I were an academic reviewer, I would 100% insist that they all had mind-bending and poetic speculative conclusions like this.

In my head, before today, a von Neumann probe was something like a starwisp – a probe the size of a smartphone strapped to a giant solar sail. Lightweight, tiny, quick. Using, a don’t know, a drill bit and a special-purpose printer, it would be able to land on any asteroid and build a general-purpose printer which it would then use to reproduce itself.

What I love about Freitas’ paper is that it shows how hard a real von Neumann probe would be. And heavy! And slow!

Is the Freitas design for a von Neumann probe too big?

If anything I think REPRO is not big enough.

I mean, I don’t know, but in the latest JBIS there’s a paper by Stephen Ashworth with a line that made me shift my perspective:

Any self-replicating probe must carry a complete embryonic industrial base in its payload - a seed economy - capable of recreating the infrastructure which built it in the first place. It should be clear that the total mass of the mature economy needed to build an interstellar probe will be greater than the payload of that probe, thus growth will be an inherent feature of the seed economy.

(The paper is Self-Replicating Interstellar Probes and Runaway Growth Reconsidered and Ashworth maintains an archive of his papers, although this one isn’t online yet.)

FACTORY is an industrial base!

FACTORY is a whole economy!

I’d been thinking about this in terms of robots – but 5,000 years is pretty much the whole duration of Western society! Some robots.

We can’t even run computers from a few decades ago without human maintenance – and that maintenance is provided by the tip of a pyramid of university education and surplus labour that allows for a cadre of specialistic computer historians!

How much would you actually need to boot up and run an industrial economy, from scratch, to copy and launch an interstellar probe (the launching of which would take a meaningful fraction of global GDP and energy budget) – given your economy also needs to maintain and grow itself?

And don’t forget that the industrial base templated in REPRO needs to be robust enough to function in whatever star system it finds itself in… ones with scarce heavy metals, ones with violent radioactive stellar storms, ones with dim and dying binary stars, ones dominated by a single Jupiter-type and all the good stuff is stuck at the bottom of a gravity well, ones that are still an accretion disc, etc.

What if the minimum size of FACTORY is effectively the same as human history and human civilisation?

Von Neumann probes might have to be way bigger than starwisps.

Oh! So perhaps we are FACTORY.

Perhaps all of human life is basically the larval stage of a self-replicating von Neumann probe.

We shouldn’t be looking out with our radio telescopes for the galactic spread of self-replicating explorers, we are the galactic spread.

I feel like I’ve finally gotten the joke which is decades old. The point of the von Neumann probe thought experiment is that we are the von Neumann probes, or at least a single iteration of them.

Well I got the gag eventually I guess.


Perhaps we’re the first iteration of the von Neumann probe. Perhaps our civilisation is the first FACTORY, and we’ll eventually make a probe (or whatever) that copies human civilisation to new star systems, and so on and so on.

If our actions and values today result in a successful REPRO construction, then the very same actions and values will be replayed countless times in countless star systems across the Milky Way.

In which case I’m reminded of Alasdair Gray’s line engraved on the Scottish parliament building, Work as if you live in the early days of a better nation.

(Itself a mutation of a line from Canadian poet Dennis Lee.)

Work as if you live in the first iteration of exponential galactic exploration?

Hello, fellow FACTORY worker.



There’s Nick Bostrom’s argument that we live in a simulation, as discussed: if there would be vastly more simulated minds than base reality minds then, by weight of numbers, you and I are more likely than not to be in the simulation.

By analogy:

If human civilisation is a single iteration of a von Neumann probe exploring the galaxy, and given that there will be hugely more iterations than beginnings, then by weight of numbers it is highly unlikely that our civilisation is iteration 1. Call us iteration N.

Therefore we can learn about aliens by introspecting ourselves: who would design a self-replicating machine that unfolds into an industrial base that looks like human civilisation? Like us?

How would they think?

Could we trace the path back and find them? (I mean, surely yes? Because the probes will be sending messages of their discoveries back to base, somehow.)

Hey: look at whatever device you’re reading this on. Look out of the window; look up and down the street. Listen to the traffic. This is the sound of the universal constructor in action. Look at your hands. Witness FACTORY.

More posts tagged:
Follow-up posts:

If you enjoyed this post, please consider sharing it by email or on social media. Here’s the link. Thanks, —Matt.