Efficient programmable self-replicating machinery does not need nanotechnology, and it will change economics and human life in many ways that are currently inconceivable.
The first and most obvious change is that it will change the nature of factors of production in a major way. If we analyze factors of production into land, labor, capital goods, information, and energy, it will provide abundant capital goods for any conceivable production process. This will increase the value of land, labor, information, and energy, but of these, information and energy are likely to be abundant.
Because capital goods will be abundant, we will stop counting them as a significant part of our assets. A brewery does not measure its wealth by how much yeast it has on hand, or a farmer measure his wealth by the amount of seed corn he has available to plant. Similarly, we will no longer consider machinery to be a form of wealth — perhaps architecture, either. We will instantiate them when we want them, then recycle them when we no longer do.
For the last three million years or so, accumulating capital goods — whether hand axes, bottle gourds, or Ferraris — has been a defining aspect of human culture. It is part of what it has meant to be human. And now that is going away.
Recycling capacity is becoming a new factor of production. So far our conversion of material into moop has been modest, despite occasional incidents like the New York garbage barges and the EU’s e-waste directives. So far, our garbage production — ultimately the same as our production of goods — has been limited to a few tons per year per person. You could dry it out and store it in your basement if you had to. This will no longer be the case; we will have to recycle our moop into something else, something like Earthships or fresh metals.
Our concepts of quality will change. Historically, high-quality capital goods were those that were durable, pleasant to use, easy and inexpensive to repair, broadly applicable rather than overly specific, and augmented our economic productivity by a large factor. Once self-replicating machinery is in play, durability becomes a secondary concern that we can easily trade off against other kinds of merit. Traditionally, a bicycle chain breaker that breaks on the fourth use would be considered crap, but if you have it fabricated on the spot when you need to change the length of a bicycle chain and recycle it when you’re done, you don’t need much durability, and you don’t need it to fit different kinds of chains. Instead you will care about how quickly it can be fabricated and how much energy it costs to fabricate and recycle it; perhaps you will also prefer it to last dozens of uses if you are in a bicycle shop, but you will be willing to trade that off against ease of fabrication.
Many repairs will likely be replaced by recycling, as they are in biological systems like our bones, or often in computer filesystems.
Two particular figures of merit are the exponential growth rate of the machines and the amount of life-cycle cost per machine (or unit of productivity) measured in factors of production other than capital goods. The exponential growth rate is determined by the replication time — the time for one machine to construct another, replicating itself — and by the durability of the machines. If a machine’s MTBF is lower than its self-replication time, the exponential growth rate will be negative; if the MTBF is equal, the growth rate will be zero. But once the MTBF is a few times greater than the self-replication time, further improvements in reliability and durability will have little effect on the population growth rate.
I think a good and plausible target is a 24-hour self-replication time and an 8-replication MTBF.
Once self-replicating machines are capable of producing solar cells with a reasonable EROEI, energy will also cease to be a significant limitation on their growth, leaving only labor, land, and knowledge.
Many jobs that can be done by self-replicating machines can be done at very small scales, perhaps even submicron scales, even if they do not amount to molecular nanotechnology in themselves. This reduces the material resources needed for replication proportionally. For things that can be reduced in this way, land for resource extraction will be a minimal, even negligible, cost; only labor and knowledge will add significant costs.
We should strive to ensure that everyone has access to knowledge and self-replicating machines, so that they will not be reliant on owners of capital goods for jobs, as they have been since the Industrial Revolution.