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.


5.5 Fallacy of the Substrate

To generate macroscopic quantities of useful products using nanoscale manufacturing systems, huge numbers of such systems must join in the work. Building a huge number of microscopic systems one by one is time-prohibitive, hence the concept of massively parallel assembly or self-replication is key to all known proposals for rapid molecular manufacturing (Section 5.7). As conceived by von Neumann (Section 2.1.2) and subsequent researchers, the most general theoretical conception of physical replication views replication as a manufacturing process. In this process, a stream of inputs enters the manufacturing device. A different stream of outputs exits the manufacturing device. When the stream of outputs is specified to be identical to the physical structure of the manufacturing device, the manufacturing device is said to be “self-replicating.”

Note that there are no restrictions of any kind placed upon the nature of the material inputs. On the one hand, these inputs could consist of a hot plasma containing equal numbers of atoms of all 92 natural elements – which is by some measures a “perfectly random” input stream. On the other hand, the material input stream could consist of cubic-centimeter blocks of pure elements. Or the stream could consist of prerolled bars, sheets, and wires, or more ordered inputs such as pre-formed gears, ratchets, levers and clips. Or the inputs could consist of more highly refined components* such as pre-fabricated whole motors, switches, gearboxes, and computer chips. A manufacturing device that accepts any of these input streams, and subsequently outputs precise physical copies of itself, clearly is “self-replicating.” Insisting that only assembly of a copy starting from highly disordered substrates like a hot random-atom plasma or even a somewhat more ordered natural free-range environment counts as “true” replication [988] is one manifestation of the Fallacy of the Substrate. In fact, construction of self-structure from any stream of inputs, however ordered or disordered, properly may be termed self-replication. Stated more succinctly, the Fallacy of the Substrate is the assertion that replication, or replicators, can be described or defined without reference to their input substrate. To ignore the substrate upon which replication will take place is to perpetrate the Fallacy.

* The nutritional requirements of the self-replicating machines described in Sladek’s 1968 novel The Reproductive System [672] clearly violate the safety recommendations of the Foresight Guidelines [271] -- guidelines which we endorse (Section 5.11). Sladek writes:

“During the week...the boxes had devoured over a ton of scrap metal, as well as a dozen oscilloscopes with attached signal generators, thirty-odd test sets, desk calculators both mechanical and electronic, a pair of scissors, an uncountable number of bottle caps, paper clips, coffee spoons and staples (for the lab and office staff liked feeding their new pet), dozens of surplus walky-talky storage batteries and a small gasoline-driven generator.

“The cells had multiplied – better than double their original number – and had grown to various sizes, ranging from shoeboxes and attaché cases to steamer trunk proportions. They now reproduced constantly but slowly, in various fashions. One steamer trunk emitted, every five or ten minutes, a pair of tiny boxes the size of 3x5 card files. Another box, of extraordinary length, seemed to be slowly sawing itself in half.

“General Grawk remained unimpressed. ‘What does it do for an encore?’ he growled....”

The 1980 NASA study on replicating systems [2] offered an amusing illustration of the Fallacy of the Substrate with its example of a self-replicating PUMA robot. This robot was conceptualized as a complete mechanical device, plus a fuse that must be inserted into the robot to make it functional. Here the input substrate consists of two distinct streams: (1) a stream of 99.99%-complete robots arriving on one conveyor belt, to the left, and (2) a stream of fuses arriving on a second conveyor belt, to the right. The active replicator robot combines these two materials input streams, and the result of this manufacturing process is a single output stream consisting of physical duplicates of itself. Undeniably, the robot has “replicated.” Similar two-component replication schemes are commonplace in simple self-replicating chemical systems [1372] of the form A + B → T, where A and B are substrates that bind to a complementary template T and become joined to form a product molecule that is identical to the template [1373-1376], i.e., AB = T.

In principle, different replicator motifs might metabolize any of an infinite number of input substrates. Depending upon its design, a particular replicator device may be restricted to replication from only a very limited range of input substrates. Another replicator device may have sufficient generality to be able to replicate itself from a very broad range of input substrates. In some sense this generality is a measure of the device’s survivability in diverse environments, and may contribute to its self-replicability (Section 5.1.9 (C11)) and evolvability (an undesirable trait in a manufacturing system; Section 5.1.9 (L)). But it is clearly fallacious to insist that “replication” can occur only when duplication of the original manufacturing device takes place from some highly-disordered, ill-defined or arbitrary substrate.

Of course, as a practical matter the economic value of replication is best realized when the substrate is inexpensive. The input stream of 99.99%-complete robots might prove expensive to make in the absence of some prior manufacturing step able to operate on a less expensive substrate.

Replicating systems may be very simple. Von Neumann concluded that kinematic machine replication was possible and that perhaps twelve different kinds of subunits of unknown complexity might be required as building materials. Falling prey to the Fallacy of the Substrate, Haldane [2680] subsequently inferred that as many as ~105 parts might be needed to make a replicator. This inference was refuted just three years later with the arrival of the first in a series of ingenious designs for mechanical self-replicating machines that were built and operated or designed in the late 1950s (Sections 3.3-3.5), composed of no more than about a dozen parts that were made readily available in the external environment and were assembled into a complete (and fertile) daughter machine by the parent machine. Replication is fundamentally so simple a task that artificial machines capable of displaying this behavior in primitive form pre-date most of the modern electronic computer era. Analogizing from results in cellular automata replication studies, James Reggia [402] notes: “Self-replication is not an inherently complex phenomenon but rather an emergent property arising from local interactions in systems that can be much simpler than is generally believed.”

Perhaps the most important message of the Fallacy of the Substrate is that the replicative capacity of a replicator cannot be defined by specifying the replicator in isolation from its surroundings. Replicative capacity can only be defined by simultaneously specifying both the replicator and the input substrate upon which the replicator will be required to operate. This requirement is well-known in biology, where DNA, while representing the “recipe” of an organism and thus information about the replicator’s own structure, more importantly represents a plan for how to build an organism that can best survive in its native environment and pass on that information to its progeny [504]. Deutsch [2681] refers to this view as: “genes embody knowledge about their niches.” In cell biology the replicative environment can be extremely complex, consisting of the ribosomes which translate mRNA-based genetic messages, an abundance of nutrients inside and outside of the cell, the environment of the organism proper (e.g., ambient temperature and oxygen in the air), and so forth. As Adami [504] explains: “An organism’s DNA is not only a ‘book’ about the organism, but is also a book about the environment it lives in, including the species it co-evolves with. Accordingly, Mycoplasma mycoides (which causes pneumonia-like respiratory illnesses) has a complexity of somewhat less than one million base pairs in our nasal passages, but close to zero complexity most everywhere else, because it cannot survive in any other environment – meaning its genome does not correspond to anything there.”

A dramatic demonstration of this now-obvious truth was provided by Sol Spiegelman, an American microbiologist who in the mid-1960s posed the deceptively simple question: What is the smallest molecule capable of replicating itself? In Spiegelman’s classic experiment [2682], a primitive phage Qb virus consisting of a single 4500-nucleotide RNA molecule was supplied with an abundance of replicase enzyme (the viral enzyme that duplicates RNA) and free nucleotides that the virus needed for replication and survival, in a test tube using a flowthrough system that continually added the nutrients. With the provision of these materials, the virus was no longer dependent upon a cell to continue its life cycle and began to compete against itself, improving the efficiency of replication through the evolution and survival of viruses that did not produce the supplied materials. Within seventy generations, rapidly reproducing mutant strands having only 220 nucleotides had replaced all other variants. The successful mutants had neither viral coats nor the ability to produce the replicase enzyme, and represented a specialized new creature adapted to its whole environment, including the biotic and abiotic components. The replicating RNA had shrunk to the minimum size of the supplied replicase enzyme’s recognition site. (Had the RNA shrunk any smaller, it would not have been able to use the free replicase, and thus to achieve closure it would have to have either made its own replicase or invented a jig that would permit use of the existing replicase. This would be very difficult because losing nonessential components is far easier than evolving new structures through mutation, hence the size of the replicase site acts as a closure attractor.) Although the mutant molecule could reproduce itself at a fantastic rate in the protected test tube environment, it could not survive in the unprotected natural world where the input substrate was no longer ideal. Its replicative capacity had become permanently diminished in, and uniquely dependent upon, the new richer substrate.

Concludes Luksha and Plekhanov [2683]: “It is evident that any information may only exist relative to (or in relation with) a complex system that processes it (self-replicators in particular); this may be generalized as a principle of information relativity.”


Last updated on 1 August 2005