Molecular manufacturing does not require self-replicating machines

The development of molecular manufacturing poses the problem of starting with very small devices and using them to build larger ones, eventually building desktop-scale nanofactories. The initial idea for how to do this involved imitating biology: We know that a single self-replicating cell can grow into a redwood tree, which shows that self-replication provides one way to scale up from small systems to large. A genuine self-replicating machine, however, has needless complexity. It would have to contain, in a single package, all the tools and information needed to build something like itself — and more, if it is to build something else that is useful.

As one might expect, practical proposals for developing nanofactories have avoided that complexity. Even approaches that researchers have described in terms of “self replication” typically use devices that are merely tools under external control. Crucial parts of the system are outside, and not replicated. Since these devices lack an internal description of how to build things — whether like themselves, or different — they are quite unlike anything alive (and by nature, they cannot be autonomous).

In the macroscopic world, we use machines to prepare materials, to make parts, and to assemble parts. Overall, we use machines to build a wide range of products, including more machines useful for production — yet none is self-replicating. Nanomachines can be used in the same way, making a wide range of products, including parts that can be snapped together to make more nanofactories.