How to Reindustrialize

The zaibatsu model is over

The United States government is an elderly drunk, staring at the pile of half-crushed Miller Lites in the corner and considering it may have at some point got sidetracked in its life plans. “I still got time, man”, he says. “Just really need to buckle down, I can pull this off. This is the year, man.” He flexes for the mirror and stares at a picture of himself from 1945. “Yeah, I still got it.”

But it doesn’t seem to, you know, take. When we try to onshore semiconductor production, we have other priorities - The Environment (making sure DNC-connected lawyers and consulting firms get paid), requirements for Community Partners (making sure DNC-connected NGOs get paid), requirements for on-site child care (need I point out whomst Child Care Workers vote for), and so on. Meanwhile the slow degeneration continues - we try to write checks, but the DTs kick in and somehow the signature comes out funky - a single $35 million grant spent thus far, roughly the budget for DoD typewritter ribbons, out of the $100 billion burning a hole in our pockets.

Bloomberg et al will tell you the easy solution is to bring over infinity real actual Taiwanese, who get the racial +2 to dexterity necessary to operate at nanometer scale in the same way that only someone with a 2035 immigration court date can understand the essence of tacos (Paracelsus considered them as a condensation of light, but they remain ultimately mysterious). Somehow, this seems unlikely.

The truth is that the US is stuck in a classic third world style development trap. When Irumba figures out how to make his farm more productive and brings it to market, the surplus is immediately swallowed up rather than reinvested. Every relative shows up to ask for a handout, and how can one refuse without becoming a stranger to them? And when the local cops realize a new factory is a moneymaker and the owner can afford a substantially higher bribe, one can’t just not pay the guys keeping the wires in the walls.

The large scale version of this consumes every large government run or sponsored program in the US, from semiconductor plants to high-speed rail, and the demands impinge upon even muh private companies that grow to a certain size and are now required to get with the program, on pain of endless harassment. So in an odd inverse of the traditional Big Iron, State Sponsored Development model, the US industrial base is forced to go covert, or at least discreet, in order to survive.

The combination of this negative return to scale with new manufacturing technologies and coordination mechanisms actually gives rise, by necessity, to an interesting new development model. If you needed to start a guerilla manufacturing powerhouse, how would you do it?

Well, you definitely do not buy a large industrial facility in an area with good logistics and access to cheap labor. We see how that works out. Ideally you are everywhere and nowhere - there are offices, sure, but unless you’re to the point of blackbagging staff, there should be nothing worth impounding. But where does the stuff get made?

There is a quiet revolution going on in design for manufacturability on new production processes. 3D printing is the most obvious example - stuff that used to need to be molded with one-off tooling costing minimally tens of thousands of dollars with an extremely long iteration cycle (which now goes to Shenzhen and back) now pops off a printer in a matter of hours, and at your door the next day. iPhone holders and Hitler busts are the consumer-facing version of this, because it’s difficult to explain to der normie the massive benefits of using it as the input to other industrial processes like lost-PLA casting, holding fixtures for CNC setup, jigs for manual mills, or even electrochemical machining.

Weirdos (Europeans and various flavor of high-T rainbow flaggers) are disproportionately prototyping a lot of these new industrial processes, mostly because they are interested in keeping things off the books and have low amounts of available capital, but the results are impressive, especially when used as the design core for hybridization with off the shelf components. Even the big guys are innovating too, albeit with different materials and capital requirements, to make things that could not be traditionally machined except by Warhammer-sized machinery, with associated cost. Building, eg, a gigantic metal stamping infrastructure from scratch in the US is too difficult, so don’t do that. Use off-the-shelf and relatively inexpensive industrial robots instead, combined with a bunch of valuable IP on how to go from design to reality. The robots themselves could be anywhere and in any number - there is no requirement for a gigafactory to achieve economies of scale, because economies of scale are not the point, and again possibly counterproductive as it paints a target on your back.

What these approaches have in common are the democratization of production and design, and a flattening of production processes. Delivering a million Macbook chassis at minimal cost is not the goal here. The idea is to deliver small production runs of semi-custom or niche products that would otherwise be unaffordable, at a price only mildly higher than what you get from huge commodity suppliers, with a limited number of effectively artisanal designers and manufacturers involved. At the high end of volume, instead of needing to find aggregate supply for ten thousand widgets, middlemen contract out bits and pieces to semi anonymous distributed production farms. At the low end, for the price of a motorcycle, a single garage can contain a number of formerly disjoint production processes and a relatively skilled designer that can crank out one-offs and iterate on design for a fraction of the cost it would take to coordinate larger shops with purely theoretical economies of scale.

This doesn’t actually give you a chip fab, or an artillery line, because unfortunately those actually are inherently Warhammer-scale production processes. But it does have implications for your defense industrial base as a whole, because big-ticket projects are more and more a perpetual string of expensive prototypes limited by available human capital and bottlenecks around key custom components. The crash in the effective price of capital encourages the distribution of human expertise, because one no longer needs a job at Ford or Boeing to justify a CNC hobby. Even the tooling itself is getting more and more capable, to the point where if the US suddenly needed a few million drone bodies in a hundred variants, it could be accomplished without doling out a ten digit defense contract for spinup.

And if the US decided to actually promote this model instead of dispatching commissars as soon as one hit fifteen employees, it might soon find it no longer has a problem with the larger stuff.