The Map of Cloth Was Never Drawn by the Market

In 1907, Aurel Stein dug a bundle of undelivered Sogdian letters out of a Han-dynasty watchtower ruin west of Dunhuang, written around 313–314 CE. The sender, Nanai-vandak, reported to his principals in Samarkand: Luoyang had been burned (the Yongjia disaster), the Indians and Sogdians stranded there had starved, and the money could not be recovered. These are the earliest first-hand commercial documents from the Silk Road.

After reading through the contracts and ledgers excavated at Turfan and Dunhuang, Valerie Hansen's The Silk Road: A New History (2012) reaches a counterintuitive conclusion: trans-Eurasian long-distance bulk trade was small, and the overwhelming majority of transactions were short-range, low-value and local. What actually fed the oasis markets was Tang fiscal policy — the court garrisoned Xizhou, Tingzhou and Anxi, and paid wages and procurement largely in bolts of silk. In the Turfan documents silk functions directly as a unit of account: portable, divisible, storable, better suited to the frontier than bronze coin. It was state military spending that sustained the market along that road. After the An Lushan rebellion of 755 pulled the Hexi and Longyou garrisons east and cut the fiscal flow, monetized exchange in the oases contracted with it.

Integrated circuits in the 1960s: Apollo and the Minuteman program bought virtually the entire output, and the civilian market grew later out of the unit costs that volume had pushed down.

In 1860 British cotton spinning and weaving employed roughly 440,000 people, and about 80% of the raw cotton came from the American South. In April 1861 the Civil War began, the Union blockaded Southern ports, and supply broke — what Sven Beckert's Empire of Cotton (2014) calls a forced rewiring of the global cotton network.

By the end of 1862 roughly 60% of Lancashire's mill workers were unemployed or on heavily reduced hours; cotton went from about 6.5 pence a pound in 1860 to over 30 pence in 1864. The price signal travelled down imperial arteries: Berar in central India, the Egyptian Nile Delta and northeastern Brazil switched rapidly into cotton. The Egyptian khedive Isma'il Pasha (r. 1863–1879) multiplied cotton exports within a few years and, pledging future cotton revenue as collateral, borrowed heavily in London and Paris to build railways. In 1865 the war ended, American cotton returned, and prices collapsed. Egypt's debts, however, had been signed at the high price. In 1876 Egypt declared itself unable to pay; Britain and France took over its finances; in 1882 British troops occupied the country. A war in North America, transmitted along the single wire of the cotton price, pulled Egypt into colonial status.

The 2011 Thai floods took out roughly a quarter of world hard-drive capacity; once the price spike passed, the capacity that had moved to the Philippines never moved back.

Around 1764 the Lancashire weaver James Hargreaves built the spinning jenny in Blackburn, letting one person work eight spindles at once. In 1768 his workshop was smashed by spinners who feared for their jobs; he moved to Nottingham and patented the machine in 1770. The following year Richard Arkwright opened his water-powered mill at Cromford, and in 1779 Samuel Crompton's mule merged the two lines. One fact usually skipped over: at the same time Indian artisans out-spun the British by hand, and Bengali muslin was the global benchmark. The skill was not a British monopoly.

Robert Allen's The British Industrial Revolution in Global Perspective (2009) explains it through relative prices: eighteenth-century London had among the highest real wages in the world, and British coal at the pithead was extremely cheap. Substituting machines and coal for hands therefore paid in Britain and did not pay in India or China, where wages were a fraction of that. Allen estimates the jenny's return on investment as solid in Britain, marginal in France, and negative in India — the same machine is a completely different business decision under different factor prices. There was a demand-side layer too: the Calico Acts of 1700 and 1721 barred Indian printed cottons from the British market, reserving protected space for domestic spinners.

The highest industrial robot densities are in South Korea, Japan and Germany — high-wage economies automate first. The sequence of AI deployment is likewise set by the unit price of the labour being displaced, not by model capability.

The Multi-Fibre Arrangement (MFA), in force from 1974, set country-by-country quotas on textile imports into developed markets, intended to protect European and American producers. The side effect showed up immediately: once South Korea, Hong Kong and Taiwan had used up their quotas, orders had to go looking for countries that had no quota yet.

In 1978 Bangladesh's Desh Garments signed an agreement with South Korea's Daewoo and sent 130 Bangladeshi workers to a Busan factory for six months of training in cutting, sewing, quality control and export documentation. Back home, those workers left one by one to start their own factories — the World Bank's Yung Whee Rhee called it the "catalyst model" of development. Bangladesh went from a handful of garment plants to hundreds; by the 2010s garments were over 80% of its exports, employing some four million people, roughly 80% of them women. The bill came due on 24 April 2013: the eight-storey Rana Plaza outside Dhaka collapsed, killing 1,134 people. Visible cracks had appeared the day before; the shops and bank on the lower floors evacuated, while the garment workers were told to report as usual. The Accord on Fire and Building Safety in Bangladesh, signed afterwards, is a rare legally binding transnational labour agreement.

Export controls and tariffs are producing the same unintended maps: the restriction lands in one place, the capacity grows in another. And third-party auditing, now carrying great hopes in AI governance, is worth judging first on its track record in garments.