This is such an interesting explanation from the Mutualist Blog, at that I’m quoting it in full-length.
Kevin Carson:
In the comments on “The Two Economies,” Matthew Claxton of Little Iguanodon referred me in the comment thread to material on desktop manufacturing. As he says,

Manufacturing is already, in small ways, breaking out into an open-source model.

“Desktop manufacturing,” as a catchall term, includes a wide spectrum of different innovations.

At the most science-fictiony end of the specturm, the Center for Bits and Atoms at MIT operates on the philosophy that “reality is information”; and at the point where bits intersect with atoms, via all kinds of nifty stuff like nanotech and von Neumann replicators, you get something like the transporters and replicators on Star Trek. While that may be on the way, it’s too far out for me to get my mind around.

At an intermediate stage, with one foot still in the old-fashioned world of meatsphere-style production, he links to a Salon article on desktop manufacturing via 3-D printers. The idea is a printer that, over a period of hours, lays down layer after infinitesimal layer of glue and conductive material, gradually fabricating an electronic device or appliance with circuitry embedded in its frame; or, alternatively, “printing” circuitry onto a sheet. You transmit the specs for an electronic device, and it gets “printed” in 3-D at the destination.

But at the other end of the spectrum, most relevant to what I’ve written in the past on decentralized production, he links to some sites like Squidlabs, eMachineShop, and Big Blue Saw that will custom machine individual parts to your specs by online order, and ship them to you. As eMachineShop says,

Welcome to eMachineShop – where you can design, price, and order your custom parts online!

eMachineShop is the remarkable new way to get the custom parts you need – the first true online machine shop. Download our free software, draw your part, and click to order – it’s easy! Your part will be machined and delivered – at low cost.

With Big Blue Saw, likewise,

you can upload a part and have it shipped to your door in 14-21 days. We use state of the art computer controlled robotic machinery to convert your ideas into a reality.

Of all the material Matthew referred me to, this is the most interesting to me, because it ties in with a lot of other ideas I’ve been toying with lately. In “P2P,” I wrote:

In a decentralized economy, individual stages of production that are currently carried on at a single large site by a vertically integrated corporation may be more economically done in small machine shops (cf. Jane Jacobs’ account of the origins of the Japanese bicycle industry, in The Economy of Cities), or in Kirkpatrick Sale’s neighborhood repair-recycling centers. In some stages of production, the substitution of lower-tech, partially human powered operations (cf Mumford’s discussion of “polytechnic” in The Pentagon of Power) or Sale’s and Bookchin’s general-purpose production technologies, will be economical when savings on bureaucratic and distribution costs, and overhead, are taken into account.

I dealt with these ideas at much greater length in “On the Superior Efficiency of Small-Scale Organization.”

In The Visible Hand, Alfred Chandler idealized the kind of high-tech, capital- and energy-intensive production in which “economies of speed” or “throughput” are used to “transform high fixed costs into low unit costs.” But the success of such a model depends on artificially large market areas with a swift and uninterrupted flow of goods through a mass-distribution and -marketing pipeline, reinforced by an advertising-mediated culture of mass-consumption–all of which depends on state-subsidized infrastructure. When those artificial assumptions are removed, and production is in limited runs for small local markets, Mumford’s and Sale’s industrial model is likely to be more efficient than Chandler’s.

The custom machining of parts fits right in with these ideas. What really interests me is the potential for using the kind of thing eMachineShop and Big Blue Saw are doing–but in the context of small machine shops integrated into a local economy.

F.M. Scherer, a specialist on economy of scale, wrote of the false economies involved in higher-tech, more product-specific forms of production than the size of the market would support:

Ball bearing manufacturing provides a good illustration of several product-specific economies. If only a few bearings are to be custom-made, the ring machining will be done on general-purpose lathes by a skilled operator who hand-positions the stock and tools and makes measurements for each cut. With this method, machining a single ring requires from five minutes to more than an hour, depending on the part’s size and complexity and the operator’s skill. If a sizable batch is to be produced, a more specialized automatic screw machine will be used instead. Once it is loaded with a steel tube, it automatically feeds the tube, sets the tools and adjusts its speed to make the necessary cuts, and spits out machined parts into a hopper at a rate of from eighty to one hundred forty parts per hour. A substantial saving of machine running and operator attendance time per unit is achieved, but setting up the screw machine to perform these operations takes about eight hours. If only one hundred bearing rings are to be made, setup time greatly exceeds total running time, and it may be cheaper to do the job on an ordinary lathe. [Industrial Market Structure and Economic Performance, p. 97]

Now, if production runs sufficient to make the product-specific machinery profitable require government intervention to make distribution artificially cheap and to aggregate artificially large market areas, then it’s not really profitable at all when all the costs are internalized. It costs more than it comes to. In this case, again, Sale’s model of general-purpose production technology is more efficient. Treating transportation subsidy as a “public good,” as Chandler does, because mass distribution and marketing enable these dubious “efficiencies” in manufacturing, is ass-backward. As Peter Drucker said,

There is nothing so useless as doing efficiently that which should not be done at all.

In a related matter, Oliver Williamson argued in Market and Hierarchy and The Economic Institutions of Capitalism that internalizing separate production stages in a large firm, rather than tying them together contractually on the open market, was a governance structure made necessary by the moral hazard problems involved in “asset specificity.” In the absence of asset specificity, he said, firms would reach the point at which internal bureaucratic inefficiencies outweighed the transaction costs of market contracting at a much smaller size. But as we’ve seen, asset specificity is itself a dependent variable depending on market size–not, as Chandler seemed to believe, a good in its own right. Government policies that promote large market size and artificially increase the division of labor also lead to artificially high asset specificity, and thus make large firm size artificially efficient.