07 · Gallery I: The Single Move

THE SHORTEST PATH — A History of the Art and Science of Problem-Solving

Five solves, four centuries, one pattern. Each of the people in this gallery faced a problem that looked, on the day they met it, like it demanded more of what had already failed — more travel, more staring at the sky, more scrubbing, more flying. Each instead found a single point of leverage that made the hard version of the problem unnecessary. Watch for the moment it happens in each story: it is almost always small, almost always fast, and almost always preceded by years in which nobody thought to look for it.

A Well at Noon — Alexandria and Syene, c. 240 BC

Eratosthenes of Cyrene ran the Library of Alexandria, the largest archive of the ancient world, and he had heard a piece of travelers' lore that most scholars would have filed away as trivia: in the town of Syene, far up the Nile at modern Aswan, there was a deep well into which the sun shone straight down to the water at noon on the summer solstice, casting no shadow at all (APS News, "Eratosthenes Measures Earth"). Nowhere in Alexandria did that happen. A vertical stick — a gnomon — planted in Alexandrian ground at the same hour on the same day always threw a shadow.

The obvious way to find out how large the Earth is would be to go around it, or at least a good long way across it, and measure. Nobody in the third century BC could do that. Eratosthenes realized he did not need to. If the sun's rays arrive at the Earth essentially parallel — a reasonable assumption, given the sun's enormous distance — then the difference between "no shadow at Syene" and "some shadow at Alexandria" is not a fact about two towns. It is a direct reading of the curvature of the Earth between them, expressed as an angle. He measured that angle at Alexandria: about 7.2 degrees, one-fiftieth of a full circle. He then needed only one more number — the ground distance between the two towns — which he obtained from bematists, professional pacers trained to walk a fixed stride and report distance by counting steps, who gave him roughly 5,000 stadia (APS News). Multiply 5,000 by 50, because the measured angle was one-fiftieth of the full circle the Earth's circumference would trace, and the Earth is 250,000 stadia around.

The honest version of this story has a hole in it, and the hole is worth naming rather than papering over. Eratosthenes' own treatise, On the Measurement of the Earth, is lost; everything known about his method comes down through a summary written some four hundred years later by the astronomer Cleomedes, who compressed and possibly simplified the original argument — the well may be a later illustrative flourish rather than Eratosthenes' own evidentiary step (Cleomedes, On the Circular Motions, discussed in the Eratosthenes literature). And the accuracy everyone likes to quote — "within 2 percent of the true value" — depends entirely on which ancient stadion you plug in for the unit, and no one today knows for certain which one Eratosthenes used. Values proposed for the stadion range from about 157 to 185 meters; on the low end the error comes out under 1 percent, on the high end it balloons past 15 percent (MAA, "Eratosthenes and the Mystery of the Stades"). The number is famous. The precision attached to it is a modern retrofit.

None of that touches the actual achievement, which was never really about hitting the right number. It was about realizing that the size of the entire planet could be read off two shadows and a walked distance, without a ship, a border crossing, or a single mile of travel beyond what the bematists had already covered for other reasons. Every other method available to the ancient world involved going somewhere. Eratosthenes' method involved standing in two places at once — one of them, notably, a place he never visited.

The move: the Earth was too big to walk; a shadow was small enough to measure.

The Sky Problem That Became a Clock Problem — London, 1714–1773

By the early eighteenth century, a ship at sea could find its latitude in an afternoon, from the height of the sun. It had no reliable way to find its longitude — its position east or west — and the consequence was routine catastrophe. In October 1707 a Royal Navy squadron under Sir Cloudesley Shovell, badly wrong about its position, wrecked on the rocks of the Isles of Scilly and killed as many as 1,500 sailors. Parliament responded in 1714 with the Longitude Act, which established a Board of Longitude and offered up to £20,000 — several million pounds today — to anyone who could determine longitude at sea to within half a degree (Royal Museums Greenwich, "Longitude found — the story of Harrison's timekeepers").

For decades, the Board and most of the scientific establishment treated this as an astronomy problem, because longitude is, mathematically, a time-difference problem, and the sky was the only clock anyone trusted at sea. The first serious proposal predated the Act by a century: Galileo Galilei, having discovered Jupiter's four largest moons in 1610, proposed timing their frequent eclipses as a celestial clock visible from anywhere on Earth. It worked beautifully from a fixed observatory on land and failed completely at sea — a telescope precise enough to catch a Jovian eclipse could not be held steady on a rolling deck, and Jupiter itself disappeared behind the sun for roughly ten weeks of every year. The method that displaced it, refined over decades by the future Astronomer Royal Nevil Maskelyne into the "lunar distance method," used the moon's position against the fixed stars as a natural clock face, cross-referenced against printed tables that took teams of human computers years to calculate (RMG, "Longitude found: Nevil Maskelyne and the lunar method"; explorersweb.com, "How We Learned to Measure Longitude at Sea"). It worked, eventually, but it demanded a trained navigator, clear skies, and roughly four hours of finicky calculation per fix. A century of the era's best astronomical minds had converged on the sky as the only place a solution could live.

John Harrison, a joiner and self-taught clockmaker from Lincolnshire, refused the sky. His reframe was almost embarrassingly literal: if longitude is a time-difference problem, then the instrument that solves it is not a telescope. It is a clock — one accurate enough to carry Greenwich time, unaltered, through months at sea, through temperature swings, through humidity, through the pitch and roll of a wooden hull. That had always sounded absurd; contemporary clocks lost or gained minutes a day on dry land. Harrison spent over three decades building larger and then, crucially, smaller mechanisms — H1, H2, H3 — before arriving in 1759 at H4, an instrument the size of a large pocket watch, weighing about 1.5 kilograms, that kept time at sea to within a few seconds a day (RMG; Worn & Wound, "The Harrison H4").

On a 1761–62 trial voyage to Jamaica, H4 gave a longitude accurate to about ten miles. That was well inside the prize's terms. It took Harrison until 1773 — helped, near the end, by the personal intervention of King George III — to be paid the full reward, because the Board, dominated by astronomers with careers invested in the lunar-distance method, kept moving the terms of judgment (RMG). The lunar method also won its own consolation payments; both approaches, in the end, went to sea. But it is the clockmaker's reframe, not the astronomers' refinement, that survives in every ship and aircraft today as the chronometer.

The move: everyone was improving the sky; the answer was in a pocket.

A Pump Handle in Soho — London, 1854

In late August 1854, cholera broke out in the streets around Broad Street in London's Soho district with a speed that shocked even a city inured to epidemics: over 600 deaths in the surrounding blocks within about ten days (London Museum, "John Snow: Cholera & the Broad Street pump"). The dominant medical theory of the day, championed by the statistician William Farr, held that cholera traveled through "miasma" — bad air rising from low, damp, organically rich ground; Farr had data from 1852 seemingly showing cholera deaths falling as elevation rose, and took that as confirmation (Farr, 1852 elevation report, discussed at brianaltonenmph.com).

John Snow, a physician who had already argued in print that cholera was waterborne, made a decision that looks obvious only in hindsight: he stopped treating "the neighborhood" as the unit of analysis and started treating each address as a data point. He went door to door recording where the dead had lived and, crucially, where they drew their water. Plotted on a street map, the deaths clustered tightly around a single public water pump on Broad Street (UCLA Department of Epidemiology, John Snow Archive). On September 7, Snow persuaded the local council to remove the pump's handle.

The map itself is more famous than its timing deserves: the polished version most textbooks reproduce, with its stacked bars of death marching down Broad Street toward the pump, was not the tool Snow used to find the pattern during the outbreak — he traced the cluster first through door-to-door interviews — but a version he prepared afterward and presented to the London Epidemiological Society that December, as an argument for a theory he had already reached by legwork (Esri, "Something in the water").

Here the popular version of the story gets ahead of the evidence, and Snow's own account is more careful than his admirers. By the time the handle came off, the outbreak was already sharply declining — most of the deaths had already occurred, and case counts had been falling for days. Snow himself later wrote that he could not say the removal had much effect on an epidemic already burning out (Esri, "Something in the water: the mythology of Snow's map of cholera"). The pump handle is a legend of clean cause and effect; the truth is messier and, for a book about method, more instructive: the map's real power was not stopping this outbreak but making the mechanism visible for the first time — later investigation found the well beneath the pump had been contaminated by a leaking cesspit only feet away, including waste from an infant who had contracted cholera days earlier.

Snow's strongest evidence, in fact, came from a second and quieter piece of work often called his "grand experiment." Two water companies, Lambeth and Southwark & Vauxhall, supplied overlapping streets of south London — customers on the same block, often the same building, could be served by either. Until 1852 both companies drew water from a badly polluted stretch of the Thames; Lambeth then moved its intake upstream, above the worst sewage discharge, while Southwark & Vauxhall did not. In the 1854 epidemic, houses supplied by Southwark & Vauxhall died of cholera at eight to nine times the rate of houses supplied by Lambeth — a comparison Snow could make because everything else about the two populations was, for practical purposes, identical (UCLA John Snow Archive). It took over a decade for Farr, and the medical establishment generally, to concede the point; Farr himself finally credited water as the primary vector only in 1866, eight years after Snow's death (Wiley Significance, "London cholera and the blind-spot of an epidemiology theory").

The move: the city was hiding a mechanism behind its geography; an address book found it.

Two Doors, One Corridor — Vienna, 1847

The Vienna General Hospital's maternity service ran two adjacent clinics that admitted patients on alternating days, which made them, unintentionally, one of history's cleanest natural experiments. Clinic 1 was staffed by physicians and medical students, who spent their mornings performing autopsies before walking to the maternity ward. Clinic 2 was staffed entirely by midwives, who never touched a cadaver. For years, women begged to be admitted to Clinic 2, sometimes preferring to give birth in the street rather than risk Clinic 1 — because everyone on the ward already knew, without knowing why, that Clinic 1 killed. Mortality from puerperal ("childbed") fever there ran as high as 18 percent of mothers in some months; Clinic 2's rate was a fraction of that (Science History Institute, "Ignaz Semmelweis").

Ignaz Semmelweis, a Hungarian obstetrician working in Clinic 1, tried the theories on offer — overcrowding, diet, even the shame of male examiners — and each collapsed against some counterexample. The break came in 1847, when Semmelweis's friend and colleague Jakob Kolletschka died after cutting his finger during an autopsy and developed a wasting illness whose autopsy findings looked identical to the ones killing mothers on the ward. Semmelweis's insight was to treat that as data, not tragedy: the doctors themselves, hands fresh from dissecting corpses, were carrying "cadaverous particles" from the morgue to the delivery bed. The two clinics were not different in staff skill or patient population. They differed in exactly one variable — what the attendants' hands had touched before they touched the mother — and Semmelweis had, without designing it, an experiment that isolated that variable perfectly.

He ordered doctors and students to scrub with a chlorinated lime solution before every examination, strong enough to remove the smell of decay that soap could not. The mortality rate in Clinic 1 fell from 18.3 percent in April 1847 to 2.2 percent by June; over 1848, the first full year of mandatory chlorine washings, the clinic's rate stood at 1.27 percent, converging with Clinic 2's (Science History Institute). By any modern standard this was a clean, replicated, dramatic result: one intervention, one variable, a fivefold to eightfold drop in deaths.

The story does not end there, and the second half is the part every account of "solving" needs and most omit. Semmelweis never published his reasoning in a form the medical establishment found persuasive; he had no germ theory to explain why chlorinated lime worked, only the correlation, and he defended it with an abrasiveness that alienated the senior figures who might have carried the idea forward. Vienna's obstetric establishment largely rejected the practice after he left the hospital in 1849; childbed fever deaths at his former clinic rose again. Semmelweis grew increasingly bitter and erratic over the following years, was committed to an asylum in 1865, and died there weeks later, reportedly from an infected wound — apparently the very kind of infection his hand-washing rule was designed to prevent (Wikipedia, "Ignaz Semmelweis"; Linda Hall Library, "Scientist of the Day: Ignaz Semmelweis"). His result was vindicated within two decades, once Pasteur and Lister supplied the germ theory that explained it. The lesson this book states plainly, because Semmelweis paid for it with his career and arguably his life: solving a problem is not complete when the answer is correct. It is complete when the answer is transmitted — accepted, adopted, and used by people who were not there for the discovery. A right answer that dies with its discoverer has not finished being solved.

The move: the variable was the hands, not the ward; and finding it was only half the job.

The Six-Foot Box That Replaced the Sky — Kitty Hawk, 1901–1902

By the end of the 1901 flying season, Wilbur and Orville Wright were, by their own later account, close to giving up. Their new glider, built using the most trusted aerodynamic data available — tables of lift published by the German gliding pioneer Otto Lilienthal, refined through hundreds of his own flights — should have flown well. It generated only about a third of the lift the tables predicted (NASA Glenn Research Center, "Wright 1901 Wind Tunnel").

The instinct of most engineers in that position is to keep flying, adjusting, and hoping the next glider closes the gap — to keep searching inside the expensive, dangerous, weather-dependent world of an actual flight test at Kitty Hawk, North Carolina, four hundred miles from their bicycle shop in Dayton, Ohio. The Wrights instead concluded that the data itself, not their execution, was the problem, and that no amount of further gliding at Kitty Hawk would tell them why. They went home and built a wind tunnel: a wooden box roughly six feet long, with a fan at one end driven by a belt off the gas engine that powered their shop tools, and — the genuinely novel part — two homemade balances that could measure the lift and drag of a small model wing directly, converting airflow into a needle deflection they could read through a viewing window (Wright-House, "Doing it Wright").

Over the fall of 1901, working from strips of hammered and soldered sheet steel, they built and tested between one and two hundred miniature wing shapes, changing camber, aspect ratio, and section one variable at a time, then narrowed to roughly thirty designs for detailed, repeated measurement (NASA Glenn Research Center, "Wright 1901 Wind Tunnel"). This was the actual reframe, and it is easy to understate how large it was: the problem stopped being how do we fly and became how do we measure lift, cheaply, indoors, a hundred times a day, without risking a neck. An afternoon in the wind tunnel replaced a season at Kitty Hawk. Along the way they discovered that "Smeaton's coefficient," a constant used throughout aeronautics for over a century to convert air pressure into aerodynamic force, was itself wrong: the accepted value of 0.005 should have been closer to 0.0033, an error of roughly 40 percent sitting underneath every lift calculation the field had ever made, Lilienthal's included (NASA Glenn Research Center, "1901 Wind Tunnel Tests"). Modern aerodynamicists have since confirmed the Wrights' revised figure as substantially correct.

The 1902 glider, designed from wind-tunnel data instead of Lilienthal's tables, flew almost exactly as their calculations predicted. It is worth noting what the Wrights treated as solved first: not powered flight, but control — the 1902 glider's three-axis system of wing-warping, forward elevator, and rear rudder gave a pilot command over roll, pitch, and yaw before an engine was ever added, on the reasoning that an uncontrollable aircraft with a powerful motor is simply a fast way to crash (Smithsonian National Air and Space Museum, "Researching the Wright Way"). Power came in 1903, once control and lift were already answered questions. The famous first flight at Kitty Hawk was, by the time it happened, almost an anticlimax — the hard problem had already been solved, in a six-foot box, more than a year earlier.

The move: the sky was the wrong instrument; a box on a workbench was the right one.

The Move

Move 07 — Shrink the experiment: find the cheap measurement that stands in for the expensive world.

Every solver in this gallery faced a version of the same trap: the real problem — the size of the planet, a ship's position, a city's water, a ward's mortality, an aircraft's lift — was too large, too slow, too dangerous, or too far away to test directly. Each found a small, fast, repeatable stand-in that captured the part of the world that actually mattered and threw away the rest.

• Ask what the expensive world and the cheap surrogate share. A shadow shares an angle with the Earth's curvature. A wind-tunnel model shares a coefficient of lift with a full-scale glider. The surrogate is only useful if that shared quantity is the one you actually need.
• Isolate one variable, even when you cannot design the experiment yourself. Semmelweis did not build his two clinics; he recognized that an accident of hospital scheduling had already isolated the variable for him. Snow's two water companies were an accident of corporate history. Half the skill in this move is noticing a natural experiment already running.
• Distrust inherited data before distrusting your own execution. The Wrights spent a season assuming their gliders were built wrong before they considered that Lilienthal's tables — the entire field's shared foundation — might be wrong instead. Reframing often starts with refusing to blame yourself for someone else's bad number.
• A correct answer is not a finished one. Semmelweis proved his case with numbers as clean as any in this book and still lost, because proof and persuasion are different problems with different solutions. Carrying an answer into other people's hands is part of the work, not an epilogue to it.

Next — Chapter 08 · Gallery II: The Modern Solves: five more solves, all inside living memory, in which the single move is not a cheap measurement but a hidden constraint — the one condition, found and honored or found and broken, that decides everything else.