Kinematic Self-Replicating Machines

© 2004 Robert A. Freitas Jr. and Ralph C. Merkle. All Rights Reserved.

Robert A. Freitas Jr., Ralph C. Merkle, Kinematic Self-Replicating Machines, Landes Bioscience, Georgetown, TX, 2004.


3.11 Freitas Interstellar Probe Replicator (1979-1980)

In 1963, Bracewell [1003, 1004] first suggested using material space probes for interstellar exploration and communication instead of the radio signal approach of traditional SETI (Search for Extraterrestrial Intelligence). In 1974, Arbib [1005] became the first researcher to suggest in the scientific literature* using self-replicating probes** for this purpose: “Developments in artificial intelligence raise the question of the extent to which our eventual interstellar communication will be directly with living creatures, and the extent to which it will be with man-machine symbioses or even with a purely machine intelligence alone. In any case, much of the discussion of interstellar communication posits radio communication as the basic medium. However, we might well imagine trying to design a self-reproducing machine that carries out its own synthesis starting from the interstellar gas. These machines could then reproduce every time they travel a constant distance out from the home planet to yield a sphere moving out from the home planet with a constant density of these self-reproducing machines....What if they start to mutate?...”

* It is unknown whether Arbib was inspired by even earlier speculations on spacegoing machine replicating systems by science fiction writers, e.g., the 1967 Berserker novel by Fred Saberhagen [671] or Van Vogt’s 1950 novelette trilogy Voyage of the Space Beagle [662] which features self-replicating manufacturing plants that eventually fill a galaxy. Physicist Michio Kaku [1045] claims that Paul Davies “raised the possibility of a ‘von Neumann probe’ resting on our own moon, left over from a previous visitation in our system aeons ago. If this sounds a bit familiar, that’s because it was the basis of the [1968 science fiction] film, 2001 [2894]. Originally, Stanley Kubrick began the film with a series of scientists explaining how probes like these would be the most efficient method of exploring outer space. Unfortunately, at the last minute, Kubrick cut the opening segment from his film, and these monoliths became almost mystical entities.” However, a careful inspection of the text of the screenplay-derived novel 2001 [2894] reveals no mention of self-replicating machines, nor does the movie hint at this capability. The first of Clarke’s sequels to 2001, titled 2010 [693], written in 1982, makes it clear for the first time that the black monoliths are now self-replicating entities, but Clarke’s original 1951 short story from which 2001 was derived, called “The Sentinel” [1006], explicitly describes the found artificial lunar artifact (originally a pyramid, not a rectangular prism) as a lone non-replicating object having been left by extraterrestrial visitors in Earth’s distant past, and not as an interstellar-traveling self-replicating machine.

** These probes are sometimes called “von Neumann probes” [1043-1045] or “von Neumann machines” [1007-1009], which generates confusion because the terms “von Neumann machine” and “von Neumann architecture” have, since the mid-1940s, properly referred to the serial-processed stored-program digital computer [1010, 3069], a device which von Neumann personally described [1011] in detail and later helped build (Section 2.1), and thus is (arguably) properly attributable to his name. However, there is no evidence that von Neumann ever described a self-replicating machine in the context of interstellar exploration or interstellar communication and so it seems technically incorrect to name this concept after him. As already noted, Arbib [1005] apparently was the first scientist to propose self-replicating interstellar probes in the technical literature; perhaps “Arbib machines” could be suggested as a more appropriate nontechnical term for this concept, though it is unknown whether Arbib himself was inspired by earlier discussions of the same concept in the science fiction literature [671]. However, the authors (Freitas and Merkle) prefer the simpler technical term “self-replicating probes” since these words are most directly descriptive of the technology involved, do not generate confusion with a previously-coined term currently actively employed in a closely allied technical field, and do not improperly attribute a concept to a person who neither originated nor proposed it. Similarly, personalizing all concepts of nanorobots as “Feynman machines” [1012, 1013] or all self-replicating machines of all possible designs as “von Neumann machines” [1013] seems only to opacify and confuse, not enlighten, the discussion.

Following similar informal speculations by Calder [1041] in 1978 and by Boyce [1043] and Davies* in 1979, in 1979-1980 Freitas [1014] performed the first quantitative technical engineering analysis of a complete self-replicating interstellar probe, with special attention to materials, structural, and functional closure issues. The possibility of nanotechnology was ignored. The idea of self-replicating interstellar probes was later adopted and widely promoted in 1981 by Tipler [1015, 1044] (a physicist often erroneously credited with originating the idea)** who inadvertently did not mention the earlier work, an omission which he recognized and corrected 13 years later (see Tipler [2885] at pp. 44 (note 36), 350, and 380). Other writers [1016-1018] continue to discuss this idea.

* Writing in 1999 [1019], physicist Paul Davies notes: “Twenty years ago I suggested [1020] in a flight of fancy that an advanced alien civilization might be able to manufacture hybrid machine-organisms that could grow from seeds sent to a suitable host planet. They would develop ‘eyes’, ‘ears’ and other sensors, and even sprout a radio antenna to send back the data, using local resources and energy supplies.”

** Tipler’s most useful contribution to this discussion was his argument [1044] that if advanced extraterrestrial civilizations exist, then they must be capable of building self-replicating probes, and the speed of such replication on geological timescales would make these devices ubiquitous throughout the galaxy in our current epoch; our failure to observe these probes thus serves as evidence that extraterrestrial civilizations do not exist. Sagan and Newman [1021] subsequently published a rebuttal to this argument.

In Freitas’ paper, a nonreplicating Project Daedalus [1022] interstellar flyby space probe (Figure 3.33), with its engineering plans modified to include all subsystems necessary for self-replication, was inventoried for its mass requirements, including 86 distinct device structures. This inventory was then converted to specific mass requirements of 84 chemical elements, conservatively assuming a machine composition distribution similar to the materials consumption of the entire U.S. manufacturing system during 1972-1976, with special adjustments made for certain elements used in industrial processes not relevant to an interstellar probe (e.g., carbon black in tires and dyes, gypsum in construction, silver in coinage), and for other specialty materials preferentially employed in aerospace, electronics, or optical applications. The required element-by-element inventory was then compared to the estimated elemental abundances in a jovian planetary atmosphere or jovian moon materials resource in order to scale the onboard chemical extraction system. The performance characteristics of the extraction system were extrapolated from the known parameters of contemporary ore-processing technologies (and additionally assuming a factor of ten improvement in these parameters) – thus providing the first published worked example of replicating systems materials closure engineering (Section 5.6).

In addition to the fuel and propulsion systems employed in the original Daedalus design study, Freitas’ proposed “seed” payload unit (carried aboard each self-replicating interstellar probe) required more than a dozen specified major subsystems to allow self-replication of the entire probe. These subsystems included aerostats for jovian atmospheric mining, surface mining robots, chemical processors, metallurgical processors, a central computer with six redundant data caches (for reliably storing the self-description), fabricator robots, assembly robots, warehouse robots, factory floor crawlers, fuel tankers, repair wardens, inspection robots called verifiers, and power plants. The replicating seed unit was to have a mass of 443 tons, an initial power consumption of 412 MW, and a doubling time of 53.7 years. The final replicated probe had a dry mass of 107,000 tons (roughly the mass of a modern seagoing petroleum supertanker) and was to carry ~10 megatons of He3/D fusion fuel. The basic replicative strategy [1089] was to (a) use the probe to deliver the seed to a distant site, (b) use the seed to build more of itself, growing itself into a large factory, (c) build more probes (each containing a single seed) using the large factory, and finally (d) launch the new probes to the next distant sites.


Last updated on 1 August 2005