Lyle Burkhead wrote:
> Billy Brown writes,
> > A general-purpose nanotech fabricator, capable of
> > making most kinds of physical goods in a sealed environment,
> > is similar in complexity to an automated factory.
> > It still requires human supervision, but you don't need
> > a whole ecology of different factories to make different
> > kinds of goods.
>
> Where does the sealed environment come from? It's made in
> another factory, right? What about all the machines (including computers)
> in the factory that makes sealed environments? Where do those machines
come
> from?
This is a bit silly. Of course fabricators come in more than one size, and the biggest ones will have to be assembled manually out of components small enough to be built by fabricators. But I'm not claiming that nanotech gives us 100% automation - merely that it greatly reduces the complexity of our industrial ecology.
> Where does the building come from? -- i.e. the building that
> houses the sealed environment. It's built by a contractor, right? The
same
> contractor that builds all the other buildings in Silicon Valley. And
what
> about all the machines that the contractor uses? And so forth. You can't
> get away from the whole ecology of factories.
That doesn't follow from your questions. Of course we still need building contractors, distribution systems, and all the other parts of the economy. What we don't need is a hundred thousand different models of machinery designed to perform different manufacturing steps on bulk matter. The fabricator that makes small electric motors is at least as automated as a conventional factory that does the same thing, but it can just as easily make can openers, plumbing supplies, light bulbs or sheet metal.
> The matter-compiler could be simulated. Instead of filling a volume of
> space with a regular pattern of atoms, suppose you have a program that
> fills a simulated space with simulated atoms, and puts the result on the
> screen. At the lowest level, you have simulated atomic positioners
putting
> simulated atoms into a pattern, and then there are higher levels where
> larger systems coordinate the activities of the atomic positioners. Then
> you have the compiler that takes a high-level design and converts it into
> low-level instructions. Apparently you're a programmer, so you tell me --
> how complex would this simulation be? What would it be comparable to?
That's a big programming job, but its feasible. You wouldn't ever actually simulate the whole thing at once - the task of making steel bar (for instance) naturally decomposes into the task of filling a tiny volume of space with a given pattern of atoms, and then repeating as necessary to get the desired object.
Similar programs actually exist now, for converting CAD files into instructions for teams of factory robots. They are of limited usefulness because you need different programs for different kinds of manufacturing equipment, and its hard for the program to figure out whether it is actually possible to make a given item with the equipment it "knows" about.
I'd guess that the whole process conversion process could be made completely automated for the limited problem of making uniform bulk materials with a nanotech fabricator. Making biological materials, or objects with nano-scale components, would be a much more difficult task.
Billy Brown, MCSE+I
bbrown@conemsco.com