What Decides a War Is Usually Not the Fighting

In April 1592 Toyotomi Hideyoshi sent roughly 150,000 troops across the strait. Seoul fell within three weeks, Pyongyang by mid-summer — the land campaign met almost no resistance from an army carrying arquebuses and a century of civil-war experience. Hideyoshi's plan required land and sea to advance together: the army marching north up the peninsula, the navy shipping grain and powder up the west coast to the front. This was not a support arrangement; it was the premise the whole strategy rested on. Yi Sun-sin (1545–1598), naval commander of Left Jeolla Province, had spent the pre-war months building turtle ships and mounting cannon — one of the few Korean commanders actually ready.

On 14 August 1592, off Hansan Island, Yi lured the Japanese fleet into open water with a crane-wing formation and destroyed or captured more than fifty ships in a single engagement — Korean panokseon were larger and carried more guns, and Yi settled the fight outside the boarding range the Japanese excelled at. Japan abandoned the west-coast supply route; everything now had to travel overland. Konishi Yukinaga's army, stuck in Pyongyang without grain or powder, stalled. In early 1593 the Ming general Li Rusong retook Pyongyang and the Japanese fell back to Seoul.

In distributed systems, killing the primary node is often less effective than cutting the replication link. Today's technology controls follow the same logic: not competing head-on at the product layer, but severing the equipment and toolchain supply lane.

Meiji-era Japan's forces were plagued by beriberi: numbness in the legs, heart failure, sometimes sudden death. Takaki Kanehiro (1849–1920), a naval surgeon trained at St Thomas's Hospital in London, noticed a statistical fact — cases clustered among lower-deck sailors living on polished white rice, while officers with varied diets were almost never affected. He suspected food, but could produce neither a pathogen nor a mechanism.

When the warship Ryujo returned from a long cruise in 1883, 169 of her 376 men had beriberi and 25 were dead. Takaki persuaded the navy to run a control: in 1884 the Tsukuba sailed the same route with bread, meat and condensed milk added to the ration — 14 cases, no deaths. Naval beriberi effectively disappeared. The army's surgeon general Mori Rintaro (the writer Mori Ogai) and the Tokyo Imperial University faculty held to German bacteriology, insisted beriberi was infectious, and dismissed Takaki's work as «experience without theory». During the Russo-Japanese War the army recorded over 200,000 cases and roughly 27,000 deaths, against roughly 47,000 killed in combat — a disease preventable by diet consumed more than half as many men again as the enemy. Thiamine was not isolated until 1910.

Proposals with strong data and no clean mechanism are routinely rejected in review as «unexplainable». The more successful a paradigm is, the better equipped it is to dismiss evidence from outside itself — that is not stupidity, it is the paradigm working as designed.

Francis Lieber (1798–1872) was born in Berlin, fought and was wounded at Waterloo as a teenager, left Prussia under political suspicion, reached America in 1827 and became a professor at Columbia College. The Civil War came into his own household: his eldest son died fighting for the Confederacy, another lost an arm fighting for the Union. In 1862 the Union general-in-chief Halleck asked him to draft rules for land warfare — the urgent problems being guerrillas, plunder, and what happened to captured Black soldiers.

On 24 April 1863 Lincoln issued General Orders No. 100, 157 articles, known ever since as the Lieber Code: torture and abuse of prisoners forbidden, non-combatants and private property protected, and captured Black soldiers explicitly entitled to prisoner-of-war status — aimed squarely at the Confederate practice of treating them as rebel slaves. Yet Articles 14 to 17 established «military necessity»: measures indispensable to securing the ends of war are lawful, up to and including starving civilians in a besieged town. It is the first written law of war and also the first formal endorsement of military necessity; the Brussels Declaration of 1874 and the Hague Conventions of 1899 and 1907 all took it as their model.

Every compliance framework that draws a boundary must also write an escape hatch — break-glass privilege escalation, the exception path in an approval workflow. What gets used again and again is usually the exception.

In late-eighteenth-century French expeditions, scurvy and spoiled food routinely killed more men than the enemy did. In 1795 the French government offered 12,000 francs for a method of preserving food. Nicolas Appert (1749–1841), a Paris confectioner and cook, spent more than a decade arriving at one — bottle it, seal it, heat it in a water bath. He had no idea why it worked; Pasteur would not explain that until the 1860s. Appert published his method and collected the prize in 1810.

That same year Peter Durand patented the tinplate canister in England; by 1813 Donkin & Hall were producing in London, with the Royal Navy and the army as their customers. But one detail betrays the true speed of «spillover»: the can opener did not appear until 1855 (Robert Yeates in England) and 1858 (Ezra Warner in America). For more than forty years cans were opened with bayonets, chisels and rocks — tins brought back from Parry's 1824 Arctic voyage carry the printed instruction to cut round the outer edge with a chisel and hammer.

A jump in model capability is not a jump in productivity. What blocks deployment is rarely the core technology; it is permissions, data interfaces and approval workflows — the can openers.