The Arsenal and the Container: How Shared Infrastructure Always Wins

---
title: "The Arsenal and the Container: How Shared Infrastructure Always Wins"
section: "Jobs & Economy"
excerpt: "Every era produces the same pattern: fragmented, bespoke, expensive production transformed by shared infrastructure that democratises access to speed and scale. Venice's Arsenale, the Springfield Armory, and Malcolm McLean's shipping container all prove the same point — the breakthrough is never a better product, it is a better platform. Hardware manufacturing today is at the pre-platform stage."
share_summary: "The breakthrough is never a better product. It is a better platform. Hardware hasn't learned this yet."
sources:
 - "Behemoth: A History of the Factory"
 - "How the World Really Works"
 - "The Great Sea"
 - "Making in America"
 - "The Box: How the Shipping Container Made the World Smaller and the World Economy Bigger"
 - "The Innovator's Dilemma"
new_books:
 - title: "The Box: How the Shipping Container Made the World Smaller and the World Economy Bigger"
 author: "Marc Levinson"
 themes: ["economics", "modern"]
---

In 1574, King Henry III of France visited the Arsenale di Venezia. He arrived expecting to see a shipyard. What he saw was closer to a magic trick. Over the course of a single banquet — roughly the time it took for the wine to flow and the courses to be served — the Venetians fitted out and launched a war galley assembled from pre-made components held in inventory, armed it, and floated it. Henry was reportedly astonished. He had just watched a small city-state demonstrate a manufacturing capability that his entire kingdom could not match.

The galley was not a better ship. Venetian galleys were functional, well-made, but not dramatically superior to those built in Genoa or Constantinople. What was superior — overwhelmingly, decisively superior — was the *system* that produced it. The Arsenale was not a shipyard in any sense Henry would have recognised. It was a platform.

This is the story that repeats across six centuries of industrial history, and it is the story that hardware manufacturing has not yet learned. The breakthrough is never a better product. It is a better platform — shared infrastructure that democratises access to speed and scale, turning what was once the exclusive preserve of giants into something anyone with ambition can use. Venice proved it with ships. Springfield proved it with guns. Malcolm McLean proved it with boxes. Each time, the incumbents resisted. Each time, the economics won.

## The Arsenale: Europe's First Platform

The Venetian Arsenale was, by the standards of its age, absurd. At its peak in the early sixteenth century, it employed between 2,000 and 3,000 *arsenalotti* — rising to larger numbers still during wartime surges — a permanent, salaried workforce in an era when most manufacturing was done by independent artisans working to order. The complex covered roughly 45 to 60 acres at its maximum extent, making it one of the largest industrial sites in the pre-modern world. It had its own ropewalks, its own timber stores, its own forges, its own pitch furnaces. It was a walled city within a city, and it operated on principles that would not become commonplace in Western manufacturing for another three hundred years.

The key innovation was not any single technology. It was standardisation combined with shared infrastructure.

Venetian galleys were built to standard dimensions. The hull forms, the rib spacings, the oar lengths, the mast heights — all were specified in advance. Components were manufactured to fit those specifications and held in inventory. When a galley was needed, the assembly process drew on pre-built parts: pre-shaped hull timbers from one warehouse, pre-cut oars from another, rigging from the ropewalk, armaments from the foundry. The famous assembly process — which some historians describe as an early prototype of Henry Ford's moving production line — saw hulls towed past a series of stations, each of which added the next layer of equipment from stockpiled inventory.

The result was speed that contemporaries found incomprehensible. During emergency mobilisations, the Arsenale could fit out and launch a fully equipped war galley from pre-fabricated components in approximately one day. At the Battle of Lepanto in 1571, Venice contributed over 100 galleys to the Holy League fleet — a mobilisation that would have been impossible without the Arsenale's platform approach. As David Abulafia notes in *The Great Sea*, Venice's maritime power rested in significant part on a logistical and manufacturing system that could outproduce any rival.

What made this a platform rather than merely a large factory? Three characteristics that recur in every subsequent example.

**First, shared facilities.** The ropewalks, forges, timber stores, and dry docks were not owned by individual shipbuilders. They were shared infrastructure, maintained by the state, available to any project that needed them. An individual master shipwright did not need to build a ropewalk before he could build a ship. The ropewalk was already there.

**Second, standardised components.** Because hulls, oars, rigging, and armaments were built to known specifications, they were interchangeable across projects. A mast made in January could be fitted to a hull built in March. This decoupled production from assembly — you could manufacture components continuously and assemble on demand.

**Third, democratised access to capability.** A Venetian shipwright working within the Arsenale system could produce warships at a rate and quality that an independent shipbuilder in, say, Barcelona or Marseille simply could not match — not because the Venetian was more skilled, but because the Venetian had access to infrastructure that the Catalan did not. The platform made ordinary workers extraordinarily productive.

The Republic of Venice was a city of perhaps 150,000 to 170,000 people at its sixteenth-century peak. It should not have been able to dominate Mediterranean trade and project naval power across the eastern sea. But the Arsenale was a force multiplier so powerful that it allowed a small state to operate at the scale of an empire. As Joshua B. Freeman observes in *Behemoth*, the Arsenale was arguably the first example of industrial-scale manufacturing anywhere in the world — and its competitive advantage came not from any secret technology but from the organisational innovation of shared, standardised infrastructure.

Venice's rivals understood what they were seeing. They simply could not replicate it. The Arsenale was not a blueprint that could be copied in a season. It was decades of accumulated infrastructure, institutional knowledge, and organisational culture. By the time competitors had built their own versions, Venice had already reaped centuries of advantage.

## The Gun That Changed Nothing — And the System That Changed Everything

Two centuries later, in the young American republic, a Connecticut entrepreneur named Eli Whitney made an audacious promise to the United States government. In 1798, with war against France looking increasingly likely, Whitney contracted to deliver 10,000 muskets in two years. He had no factory, no workforce, and no experience manufacturing firearms. What he had was a concept: interchangeable parts.

The idea was simple in theory. Instead of each musket being handcrafted as a unique object — with every lock, stock, and barrel individually fitted by a skilled gunsmith — Whitney proposed manufacturing parts to such precise tolerances that any lock could fit any stock, and any trigger assembly could drop into any receiver. This would mean that broken muskets could be repaired in the field by swapping parts, that unskilled workers could perform assembly, and that production could be scaled by adding workers rather than by training artisans.

Whitney's personal contribution to this revolution was, to put it charitably, overstated. He took nearly a decade to deliver the 10,000 muskets, not two. His famous 1801 demonstration for President-elect Jefferson and senior government officials in Washington — in which he supposedly assembled a musket from randomly selected parts — was almost certainly staged, with parts pre-fitted by hand. The historian Merritt Roe Smith, among others, has argued convincingly that Whitney was a better publicist than manufacturer.

But the *idea* was correct, even if Whitney was not the man to execute it. And the place where it was actually executed — painstakingly, over decades — was the United States federal armouries at Springfield, Massachusetts, and Harpers Ferry, Virginia.

The story of interchangeable parts at Springfield is not a story of a single inventor or a dramatic breakthrough. It is a story of institutional patience. Between roughly 1815 and 1845, the Springfield Armory developed what became known as the "American system of manufactures" — a set of practices centred on precision machine tools, gauging systems, and standardised processes that gradually, incrementally achieved what Whitney had merely promised.

The key figures were not household names. John Hall at Harpers Ferry developed breech-loading rifles with genuinely interchangeable parts by the 1820s, using custom-built machinery and rigorous inspection. Thomas Blanchard invented a series of special-purpose lathes that could reproduce irregular shapes — like gunstocks — with mechanical precision. Roswell Lee, as superintendent of Springfield from 1815 to 1833, imposed systematic quality control and standardised procedures across the workforce.

What emerged was not a better gun. The Springfield Model 1842 musket was a perfectly competent weapon, but it was not dramatically superior to a hand-built musket made by a skilled London gunsmith. What was dramatically superior was the *system that produced it*. And that system had three characteristics that should, by now, sound familiar.

**Shared facilities.** The armouries' machine tools, gauging systems, and inspection equipment were shared infrastructure. An individual worker did not need to own a lathe or build a gauge. The platform provided them.

**Standardised components.** Parts were made to fixed specifications, verified by gauges, and interchangeable across units. This decoupled the skill of the individual worker from the quality of the final product.

**Democratised access.** A relatively unskilled worker operating within the Springfield system could produce musket components of a quality and consistency that previously required years of apprenticeship. The platform did not replace skill — it encoded skill into infrastructure and made it accessible.

The British, who sent a parliamentary committee to visit U.S. armouries in 1854 and subsequently purchased American machinery for the Royal Small Arms Factory at Enfield Lock, understood immediately what they were looking at. As Suzanne Berger documents in *Making in America*, the British delegation was not impressed by the quality of individual American muskets. They were impressed by the fact that the Americans could produce vast quantities of muskets of *consistent* quality using workers who were not master gunsmiths. The platform had turned manufacturing from a craft into a system — and systems scale in ways that crafts cannot.

The ramifications extended far beyond firearms. The machine tools, gauging systems, and standardised processes developed at Springfield migrated to sewing machines, bicycles, typewriters, and eventually automobiles. The [Fifteen-Minute Advantage](/articles/the-walking-distance-supply-chain) that made Birmingham's pen quarter so productive was a version of this same principle operating at the level of a city — shared, proximate infrastructure enabling small players to achieve the output of large ones. The American system of manufactures was the Springfield version: shared, standardised infrastructure enabling unskilled workers to achieve the output of craftsmen.

## The Box That Ate the World

On 26 April 1956, a converted tanker named the *Ideal X* sailed from Port Newark, New Jersey, to Houston, Texas, carrying 58 truck trailer bodies — their wheels removed — stacked on its deck. It was not an elegant operation. But it was the beginning of the most consequential logistical revolution in human history.

The man behind it was Malcolm McLean, a trucking entrepreneur from North Carolina with no background in shipping. What McLean understood — and what the shipping industry had failed to grasp — was that the problem with moving goods was not the ship, the truck, or the train. The problem was the *boundary between them*.

In the 1950s, moving a cargo of goods from a factory in the American Midwest to a buyer in Europe involved a staggering amount of friction. Goods were loaded onto trucks at the factory, driven to a port, unloaded from the trucks, sorted in a warehouse, carried by longshoremen to the dockside, loaded piece by piece into the hold of a ship, stowed, secured, and documented. At the other end, the entire process was reversed. Each transition — truck to warehouse, warehouse to dock, dock to ship — involved handling, delay, breakage, theft, and cost.

The loading and unloading of a conventional cargo ship in the 1950s typically took a week or more. Longshoremen handled cargo piece by piece — individual crates, barrels, bales, and sacks. The cost of this "break-bulk" cargo handling represented an enormous proportion of total shipping cost: by some estimates, cargo handling accounted for more than half — and in some studies as much as 60 to 75 per cent — of the total cost of shipping goods by sea. The ship itself — the expensive capital asset — spent more time sitting in port being loaded and unloaded than it spent at sea carrying cargo.

McLean's insight was that the unit of transport should not be the individual item. It should be the *container* — a standardised steel box that could be loaded at the factory, sealed, placed on a truck chassis, driven to the port, lifted by crane onto a ship, carried across the ocean, lifted onto a train or another truck, and delivered to the buyer, all without anyone touching the contents. As Marc Levinson documents in *The Box*, McLean did not invent the idea of a shipping container — others had experimented with standardised boxes before. What McLean did was build the complete system: the containers, the ships designed to carry them, the cranes to load them, the chassis to transport them, and the organisational model to make the whole thing work.

The resistance was enormous. Longshoremen's unions fought containerisation bitterly, correctly understanding that it would eliminate most of their jobs. Port authorities resisted because it required massive capital investment in new cranes and terminals. Shipping companies resisted because it meant scrapping their existing fleets. Regulators resisted because the standardisation of container sizes required international agreement. The ISO container size standards — settling on the now-dominant 20-foot and 40-foot units — were developed through a series of agreements in the late 1960s, more than a decade after the *Ideal X* first sailed.

But the economics were unanswerable. The dramatic reduction in cargo handling costs — which Levinson documents as falling by more than 90 per cent over the two decades following containerisation — was only the beginning. Port turnaround times fell from weeks to hours. Theft and breakage, which had been endemic in break-bulk shipping, virtually disappeared — the sealed container was its own security. The total cost of ocean freight as a proportion of goods' value fell so far that, for many products, it became economically rational to manufacture on one side of the world and sell on the other. Containerisation did not merely reduce shipping costs. It made globalisation economically viable.

And once again, the three characteristics of a platform were present.

**Shared infrastructure.** The container terminals, gantry cranes, and intermodal rail yards were shared facilities. A small manufacturer shipping a single container from Shenzhen used the same cranes, the same ships, and the same rail connections as a multinational shipping thousands. The infrastructure did not discriminate by scale.

**Standardised components.** The container itself was the standard — a steel box with standardised corner castings that could be gripped by any crane, locked onto any chassis, and stacked on any ship. This was the interchangeable part of global logistics.

**Democratised access.** Before containerisation, international trade was dominated by large companies that could afford to manage the complexity of break-bulk shipping — dedicated forwarding agents, warehouse space at multiple ports, relationships with longshoremen. After containerisation, a small factory in Vietnam could ship goods to a retailer in Birmingham with no more logistical infrastructure than a telephone and a container booking. The platform had made global trade accessible to anyone.

The parallels with [The Great Offshoring](/articles/the-great-offshoring-how-the-worlds-factory-moved-east) are direct. The movement of manufacturing from West to East was enabled not by cheaper labour alone, but by the container's elimination of transport friction. Without the platform, the cost advantage of Asian wages would have been eaten by the cost disadvantage of Asian distance.

## The Pattern

Strip away the specifics — the galleys, the muskets, the steel boxes — and the same structure emerges every time.

**Before the platform:** Production is fragmented, bespoke, and slow. Each producer must build or maintain their own infrastructure. Quality depends on the skill of individual craftsmen. Scale is limited by the capacity of individual workshops. Entry barriers are high because new entrants must replicate the infrastructure of incumbents before they can produce a single unit.

**The platform emerges:** Someone builds shared infrastructure — facilities, standards, and processes — that decouples capability from individual producers. Components become standardised and interchangeable. The infrastructure is open to anyone who meets the standard.

**After the platform:** Speed increases dramatically. Cost falls. Quality becomes consistent rather than variable. Small players gain access to capabilities previously reserved for giants. The industry restructures around the platform, and the producers who refuse to adopt it gradually disappear.

**The resistance is always the same.** Incumbents who have invested in their own infrastructure resist the platform because it commoditises their advantage. Skilled workers resist because it reduces the premium on their individual craft. Regulators resist because standardisation requires coordination they are not equipped to provide. The resistance delays adoption — sometimes by years, sometimes by decades — but it never prevents it. The economics always win.

This pattern is not an abstraction. It is a predictive model. If you can identify an industry that is still in the "before" phase — fragmented, bespoke, slow, with high entry barriers driven by infrastructure costs — you can predict with considerable confidence that a platform will eventually emerge to transform it. The question is not *whether*, but *when* and *who builds it*.

Consider what Vaclav Smil observes in *How the World Really Works*: the modern economy rests on a remarkably small number of foundational materials and processes — steel, concrete, plastics, ammonia — and the industries that produce them were all, at one point, fragmented and artisanal before being transformed by standardised, shared-infrastructure approaches. The pattern is not limited to transport or weapons. It is fundamental to how industrial economies evolve.

## The Pre-Platform Present

Hardware manufacturing today sits precisely where these industries sat before their platform moments.

A hardware startup developing a new physical product must, in effect, build its own Arsenale before it can build its first ship. It needs access to mechanical design, electrical engineering, firmware development, thermal analysis, regulatory certification, supply chain management, prototype fabrication, and manufacturing process engineering — and it needs all of these simultaneously, because in hardware, unlike software, the disciplines are coupled. A change to the enclosure affects the thermal design, which affects the PCB layout, which affects the firmware, which affects the regulatory submission.

As we explored in the previous article in this series, this coupling is the structural reason hardware startups are slow. But the coupling is not the root cause. The root cause is that each startup must independently acquire or build the infrastructure — the expertise, the equipment, the relationships, the processes — needed to navigate these coupled disciplines. There is no shared ropewalk. There is no standardised gauge. There is no intermodal container.

The contract manufacturers exist, certainly. The specialist consultancies exist. But they are fragmented, disconnected, and bespoke. Each engagement is negotiated from scratch. Each supplier has its own standards, its own processes, its own timelines. Coordinating between them falls entirely on the startup — which, by definition, does not yet have the institutional knowledge to do so efficiently. It is, in effect, break-bulk shipping: every piece handled individually, every transition a source of friction, delay, and cost.

The irony is acute. The technology sector — which prides itself on disruption, platform thinking, and the elimination of friction — has left its own manufacturing largely untransformed. Software solved this problem decades ago. Hardware has not. The next article in this series will examine exactly how software solved it, and what that solution reveals about the platform that hardware needs.

## The Lesson History Keeps Teaching

Venice's Arsenale operated for over seven hundred years. The American system of manufactures that emerged from Springfield became the foundation of modern industrial production. Containerisation reshaped the entire global economy in under three decades. In each case, the platform's creators did not build a better product. They built the infrastructure that allowed *everyone* to build better, faster, and cheaper.

The pattern is not subtle. It is not ambiguous. It has repeated with such regularity, across such different domains and centuries, that denying its applicability to any given industry requires a specific argument for why *this time is different*. History suggests that "this time is different" is the most expensive sentence in the English language.

The shared infrastructure always wins. The only question is how long the resistance lasts — and how much advantage accrues to those who see it coming before their competitors do.