By Andrew Kornuta — 31 Dec 2025

Industrialist Paper No. 2

The Network Is the Factory

A nation that cannot make essential goods is not sovereign. It is exposed.

Domestic manufacturing, free from adversarial meddling, is a prerequisite for national security and for a healthy middle class. We can see this in the hard lessons of world wars, and in the modern reality of Ukraine, where the limiting factor is not courage, it is industrial throughput. We can see it in hollowed-out towns where the making left but the buying remained, and in a job economy that over-weights service roles that can be offshored, automated, or economically fragilized.

Many people believe offshoring was inevitable, because the first world cannot compete with cheap labor abroad. That belief is false, particularly in the large and resource-rich United States.

The United States still has enormous industrial capacity, deep process knowledge, and world-class machines. Labor constraints are real, but they are not destiny. The tipping point between being a net importer of essential goods and being meaningfully self-sufficient is closer than most people realize.This series argues one practical claim: US manufacturing can be better, cheaper, and faster than offshore when we fix the coordination layer, using modern software to structure demand, translate messy inputs, and route work through verified suppliers based on evidence, not vibes.

The gap we are covering in this paper is coordination. Within a single enterprise, people think of this as “logistics,” but we use “coordination” because it’s a network of related but unaligned nodes. In the modern world, a single factory cannot stand alone. Even the most vertically integrated operations have dependencies. The manufacturing industry has been reluctant to acknowledge this because it feels inevitable, and therefore untouchable.

We will not copy authoritarian coordination. We will build voluntary protocols that reduce friction for the buyer, protect autonomy for the shop, and compound trust through consequences.

The good news is that we already have the tools to fix this problem. To coordinate at national scale, we must embrace software fully, but build it with respect for the machines, the process knowledge, and the incentives of the shop floor. Not software as ideology. Software as infrastructure.

A software lesson that manufacturing has not learned

Sun Microsystems used to say, “the network is the computer.” It sounded like marketing. In fact, it was marketing… but it was also remarkably farsighted for a computer company that most people don’t even remember.

Software stopped being a program running inside a box. It became a living system spread across services, APIs, and shared protocols. The machine mattered, but it was no longer the unit of progress. Connectivity was.

The same shift happened in office work. Not long ago, people still argued about the “paperless office,” as if it were a dream. Today, printing a form, walking it to another department, and retyping it into a second system feels insane. We moved the work into shared networks. We stopped moving paper.

Manufacturing never finished that transition.

We computerized islands: CAD on one workstation, CAM on another, ERP in the office, inspection reports on a shared drive, emails everywhere. But the work itself still moves between companies like paper did. A PDF drawing gets attached to an email. A STEP file gets re-exported because the last one fails. A note block gets paraphrased over the phone. A material cert arrives as a scan. A critical tolerance gets pointed out in a screenshot.

In software we used to have a term for an inefficient network that involved printing out paper and carrying it from workstation to workstation; we derisively called it a “sneaker-net”.

The factory is already a network. In the US, it is still a sneaker-net.

The factory is larger than a building

Most people picture “manufacturing” as a place: a shop, a floor, a row of machines.

That picture is outdated.

A single part can touch more than one company even when it is “domestic.” The raw stock is purchased from a distributor. The roughing happens in one shop. Heat treat happens in another. Coating happens in another. Inspection might happen in-house, or might be outsourced for specialty measurement. Packaging, labeling, and logistics might be handled by yet another company.

Even inside one facility, the same part moves through different disciplines: estimating, programming, setup, machining, deburr, inspection, finishing, shipping, accounting. Every handoff carries risk.

So the real unit of manufacturing is not the building. It is the chain.

And chains do not run on machines alone. They run on information: what the part is, how it will be measured, what is critical, what certs are required, what timeline matters, what assumptions are allowed, and what happens if something goes wrong.

A machine can cut metal perfectly and still lose the job, because the network that routes the job is slow, noisy, and blind.

Work packages are the packets of industry

In software, the internet scaled because it learned how to move packets reliably across unknown networks. It did not require every network to be identical. It required protocols that made traffic legible.

Manufacturing has its own packets. We just do not treat them as first-class objects.

A work package is not one file. It is a bundle:

A PDF drawing with symbols and tolerances
A model file with geometry and coordinate assumptions
A material spec and cert requirements
A finish requirement and any special processes
An inspection expectation, sometimes explicit, often implied
A quantity, a delivery target, and a definition of “on time”
A set of acceptable assumptions, usually unstated
A definition of what happens when the drawing is silent
A way to reconcile revision control when files disagree

That package is the real product long before the first chip is made. And today, it does not move through a coherent system. It moves through inboxes. When a work package moves through inboxes, it is not routed. It is begged.

Why “capacity” looks scarce when it is not

This is where the national argument turns practical. People look at a failed sourcing attempt and conclude: “We do not have capacity.” Often what they actually experienced was routing failure.

A buyer sends an RFQ to a handful of suppliers. Two never respond. One responds with questions. One responds with a quote that lands too late. The buyer awards to the incumbent or pushes the work overseas.

To the buyer, the story feels like “nobody can make this here.” From a network perspective, the story is simpler: the work package was not legible enough, the risk was not priced confidently, the trust evidence was weak, and the routing mechanism was primitive.

In software terms, the packet got dropped. Not because the country lacks servers, but because the network does not know where to send it, and does not know which nodes will behave. That is the defining confusion in modern manufacturing. We blame the machine count when we should be looking at the routing layer.

The invisible tax: qualification resets and defensive behavior

A healthy network amortizes trust. A broken one cannot. Today, buyer teams repeatedly rebuild the same supplier knowledge from scratch:

Is this shop real?
Do they actually have the machines they claim?
Are they stable enough to deliver?
Can they hold tolerance consistently?
Will they answer fast when the program is on fire?
Will they tell the truth when something goes wrong?

Those are not abstract questions. Each one touches a fixture, a probe routine, a CMM program, and a ship date. When trust cannot travel, every buyer becomes a private detective. Every shop becomes defensive.

Shops respond rationally:

Ignore RFQs that smell ambiguous
Quote high when the note block is unclear
Quote slow because triage is unpaid engineering time
Favor known customers because the trust is already established
Decline work that lacks clear inspection expectations
Avoid buyers who seem likely to churn suppliers

Buyers respond rationally too:

Shrink the vendor list after one quality escape
Demand more documentation after one late shipment
Send RFQs to more suppliers to “increase odds,” which increases noise
Default to brokers and incumbents because they reduce perceived risk
Offshore when lead times and uncertainty tighten

None of this requires bad actors. It is mechanical behavior in a weak network.

Why manufacturing stayed unconnected

It is tempting to say, “Just connect it.” But manufacturing resisted connectivity for reasons that are rational. The data is sensitive. Drawings are IP. Supplier lists are leverage. Prices are strategic. The inputs are messy. Scans, redlines, mixed revisions, and partial notes are normal. Layer on top of this the carnage of broken attempts to fix these concerns; fragmented standards. ISO and ASME conventions coexist. Tribal defaults are everywhere.

The consequences are real. A bad assumption can scrap expensive stock or cause a field failure. The incentives are misaligned. Most tools optimize inside the four walls, not across firms. And of course, the classic “atoms not bits” argument continues to impede progress here. Because manufacturing does indeed have a very real physical component that cannot be digitized, it’s easy for participants reluctant to change to say “well this is different from software, we are making real things.”

So what happened is the predictable half-step: machines became digital, but handoffs stayed analog. CAD and CNC advanced… but stayed disconnected and just created more islands of confusion. That is why the nation feels slower than the machines.

The missing layer is protocol, not control

When outsiders hear “coordination,” they imagine central planning. They imagine a controller dictating how shops work. That is not what a good network does. The internet did not centralize every computer. It created protocols that let strangers exchange reliable packets without surrendering sovereignty.

Manufacturing needs the same philosophy.

A shop should keep its ERP, its CAM, its fixtures, its inspection routines, its local process knowledge. A buyer should keep its PLM, its procurement systems, its internal quality gates. Nobody needs to be conquered.

What must be shared is a minimal envelope that makes work packages routable:

A way to express what is being asked, with constraints
A way to represent uncertainty explicitly, not as hidden assumptions
A way to verify identity as fact, not as marketing
A way to carry performance evidence across programs and companies
A way to enforce consequences for bad behavior without coercion

This is voluntary coordination: protocols that reduce friction for honest participants and punish noise. It is also the only approach compatible with a decentralized industrial base.

A concrete example: what “routing” really means

Consider a simple aerospace bracket. Aluminum. Medium complexity. A few datum callouts. A tight positional tolerance on one hole pattern. A surface finish requirement on one face. A cert requirement. A delivery target driven by downstream assembly.

The routing problem is not “find a machine.” The routing problem is:

Who can hold the positional tolerance without heroic inspection?
Who already has the fixture patterns and probing routines for this geometry class?
Who can get the material cert chain right without weeks of back-and-forth?
Who has open schedule windows in the correct week?
Who will ask the right clarification questions quickly instead of delaying?
Who has a track record of not hiding bad news?

Those answers exist inside the country. They exist inside shops. They exist inside buyer histories. They exist inside quote logs, NCR logs, and delivery records.

They do not exist in a shared routing layer.

So the buyer routes by guessing, and the shop quotes by defending.

That is why the network underperforms.

The real promise: coordination compounds

The reason this series is optimistic is simple: coordination compounds faster than machine tools.

One new machine adds capacity in one place and requires large capital outlay. One protocol improves routing everywhere it is adopted.

One shared identity layer removes repeated detective work. One shared performance signal reduces defensive quoting. One shared way of structuring a messy RFQ package makes a thousand work packages clearer without demanding perfect inputs.

That is how you move a nation toward self-sufficiency without fantasy.

You do not need to abolish global trade to regain sovereignty. You need the ability to make essential goods reliably when the world becomes hostile, expensive, or chaotic (and it is all of those things right now).

That ability is not just a matter of machines. It is the ability to convert machines into throughput through fast alignment.

Implications

Manufacturing is already a network of firms. The “factory” is the chain, not the building.
Work packages are the packets of industry. If they cannot move reliably, throughput collapses even with world-class machines.
Most “capacity shortages” are routing failures: weak legibility, weak trust evidence, and slow handoffs.
The winning move is voluntary protocol: shared envelopes for work packages, verification, and performance signals, without central control.
Coordination compounds. Protocol improvements scale faster than capital investment because they reduce waste everywhere they spread.

Next: Paper 3, Local Optimization Creates National Failure, why rational behavior inside each company produces a nation that feels slow, fragile, and more offshore-dependent than it needs to be.