von Neumann Self-Replicating Systems

Neumann at Dactyl
Image from Steve Bowers
Neumann vecs replicating near Dactyl. Many von Neumann self-replicating systems reproduce by creating miniature copies of themselves which then grow larger over time.

Most people watched them through the feeds of course, tapping in every now and then from the comfort of their virches. But I've had a strong fascination with Neumanns, for me the feeds were constant. Over the years I did little else but watch them work from orbit, my augmented senses drinking in the data flooding in. We'd brought about a hundred of them with us, each one no bigger than a caleb. They flew from our craft in drop pods barely bigger than their own bodies. All but two survived re-entry which was well within statistical parameters. We dropped them in small packs and singletons at landing sites distributed across the barren world; far enough apart to give each other breeding room but close enough to cooperate once they'd set up.

Those first few years were slow and steady. They grazed and dug and roamed, spawning little ones that would do the same. For a while it was just the generalists but specialisation came, bringing diversity. Small bots that burrowed, smaller bots that buzzed through the air, and larger models that would grow big enough to scrape quarries in megaseconds. I watched as they went from disparate packs to clusters of early civilisation. Little mechanosystems of solar trees, transport tubes and fab-houses. They grew like insect hives made up of hundreds of species. The ground was pocketed with living tunnels, cillia undulating on their walls to transport raw materials to the surface. Giant beetle-like Neumans sat next to storage towers, their probiscis' sucking down raw material to process in their specialist organs. The refined mattercache they produced was distributed among the ever growing, ever gravid workforce.

After the first pockets of industry were established the population began to explode. In the same time hundreds became thousands, thousands became millions. From our orbit we could have seen the progress with our biological eyes if we had them; it was like silver fungus growing on an apple. Each night the cities of industry would grow further, dropping spores way outside their borders that would blossom into mechanical life before connecting to the main. It wasn't long before resources were increasingly reallocated from growth to the terraforming proper. There were hundreds of billions of them now and the surface of the planet was encrusted with manufactories and infrastructure. Thousands of fusion generators pumped energy into the ground to melt the permafrost. Channels, dug by billions of bots, directed the water to excavation sites. There the work of trillions more geosculpted seas to our aesthetic preferences (the models there were amphibious and kept working even as the first floods came). Bioforges synthesised soil, biota, seeds and basic life that countless machines laid across continents. From my lofty view it was like watching a plague of locust in reverse, swarming over a dead land and leaving it alive.

After mere decades the planet was prepared enough that we could move into the cities (so simple compared to the rest of the efforts they were thrown up in the last megasecond, as though an afterthought). There were tens of billion continuing the work of nursing the world from its nascent biosphere to one as rich as Old Earth, but the vast majority of the mechanosystem had been dismantled. No longer necessary it was recycled. My fellow colonists barely gave it a moments thought. But in all the centuries I lived there I could never forget the majesty and the awesome power of those Neumann swarms.

Excerpt from Tyrol's Travel Notes: Terraformers

Self replicating systems have many variant types, some of which have been developed by design technology, and others have evolved naturally. The so-called von Neumann replicating system concept was first analysed in detail by John von Neumann during Earth's Atomic Age.

A classification system of von Neumann replicators

This classification system is commonly used to distinguish basic, relatively low-tech replicating systems from sophisticated self-repping universal assemblers.
  1. Self-assembling systems can assemble working copies of themselves from finished, manufactured components delivered to the self-assembler, or selected by the assembler from a storage location. Note; John von Neumann's original though experiment about a self-replicating system concerned a device of this sort.
  2. Self-reproducing systems can manufacture all the parts necessary to reproduce themselves, but the raw materials need to be delivered to the reproducer in prepared form. Such raw materials (feedstock) may include pure liquids, plastic, mineral, carbon or metal powders, metal bar, wire, coil, sheets or ingots. Atmospheric gases and other abundant fluids may be collected from the environment, but other fluids need to be supplied.
  3. Self-replicating, autotrophic replicator systems can search for, refine and process their own raw materials. They must be competent enough to collect enough material from the environment to maintain their own systems in working order, and also to reproduce working copies of themselves.
  4. Self-replicating universal assemblers can gather the raw materials to manufacture copies of themselves, and also to manufacture other goods and infrastructure.
  5. Self-replicating, evolving universal assemblers can respond to the environment by evolving, either by selection of the fittest or by smart or sentient self-redesign.
  6. Natural self-replicating systems have emerged spontaneously via abiogenesis and evolution, and include terragen lifeforms, alien lifeforms and a range of prebiotic or non-biotic life-like processes.
A von Neumann replicating system of type 3 or above can to build a working copy of itself using materials in its environment. Use of von Neumann systems of type 4 or above allows any construction project to proceed at an accelerating rate, since the number of devices increases exponentially (within certain limits imposed by the environment and thermodynamics). Exponentiating von Neumann systems can be extremely dangerous especially if they have the capability to evolve and act in unpredictable ways, but they are still often used locally or under heavy control to quickly build enormous infrastructure projects or weaponry.

The first type 3 von Neumann systems were developed in 215 AT using the Genii-2 matter compiler in tandem with a prospector/disassembler system in a joint venture between Neotech Labs in Clarke Orbital, the Xerox Nanoscale Collective in Pasadena, Earth, and the Centre for Self-Replicating Technologies, a Eurasian laboratory on Copernicus Base, Luna. This development created widespread anxiety regarding the possibility of a grey goo outbreak, even though theGenii-2 assemblers were still quite limited in function.

The development of type 4 universal assemblers occurred shortly before the Technocalypse, and contributed significantly to the serious nature of that outbreak. Since that time the trend in self-replicating technology has been to include sophisticated safeguards and fail-safes to contain and control self-repping populations, and also to reduce the size of each system (most, but not all, von Neumann systems in the present day are self-contained units capable of reproducing like a biological organism).

In the Current Era many sophont moravec robots are capable of self-replication, which means they can be considered to be 'Neumanns'. There are, however, certain populations of vecs which can only be constructed inside specialised fabricators, a characteristic which allows their population numbers to be controlled by the entities or factions who own or control those fabs.

Because self-contained, self-replicating vec populations are quite often found in space environments, on asteroids, airless planets or in ring systems, the use of the term 'Neumann' is generally confined to space-borne populations of self-repping vecs.

In certain cases the self-replicating devices in a swarm are not individually sentient, but become sentient or sophont when they reach a certain critical population by the formation of a hive mind. This process of toposophic growth may continue until the swarm becomes transapient. Most transapient entities are capable of self-replication, although the largest archailects are too large and distributed too widely to replicate readily.
Related Articles
Appears in Topics
Development Notes
Text by M. Alan Kazlev
Additional material by Steve Bowers
Initially published on 09 December 2001.

2018 Snapshot by Rynn
Additional Information