The Midnight Buzz: The Hundred-Year Evolution of Classical Computing

The air in the basement of the University of Pennsylvania’s Moore School of Electrical Engineering was thick with the smell of ozone and overheated vacuum tubes on the night of December 10, 1945. Outside, a light snow dusted the Philadelphia cobblestones, but inside, a crowd of engineers, military officers, and the six young women who had spent weeks rewiring the machine’s 6,000-plus switches held their breath as the last line of code for the test run scrolled across the machine’s printer. The task was to simulate the thermonuclear reaction at the core of the first U.S. hydrogen bomb, a calculation that would have taken a team of human “computers” (the term for people who did math by hand back then) three months to finish. ENIAC, the Electronic Numerical Integrator and Computer, crunched the numbers in 30 seconds. When the final result printed out—a set of critical mass parameters that matched theoretical predictions to within 1%—the room erupted in cheers. No one in the room, not even the machine’s designers, could have imagined that this 30-ton, 167-square-meter behemoth, with its 17,468 glowing vacuum tubes and 150 kilowatt power draw, would be the first step toward the smartphone in your pocket or the supercomputer that models climate change today.

ENIAC was not the first calculating machine, of course. Human beings have been building tools to help us count and compute for more than 5,000 years. The earliest known counting tools were clay boards used by Sumerian traders around 3400 BCE, where they would arrange small stones in grooves to track grain shipments and tax payments. By 200 BCE, Chinese mathematicians had developed the suanpan, or abacus, a frame of rods with sliding beads that could perform addition, subtraction, multiplication, and division far faster than any mental calculation. The suanpan remained a dominant computing tool across East Asia for more than 2,000 years, and is still used in some parts of the world today. In the 17th century, French mathematician Blaise Pascal built the first mechanical adding machine to help his father, a tax collector, avoid arithmetic errors. Powered by a system of gears, Pascal’s machine could only add and subtract, but it proved that calculation could be automated. A few decades later, German polymath Gottfried Wilhelm Leibniz built a more advanced “stepped reckoner” that could perform all four basic arithmetic operations, and he was the first to formalize the binary number system—using only 0s and 1s to represent all numbers—arguing that “the 0 and 1 are like the ‘nothing’ and ‘something’ of creation, from which all numbers can be born.” Leibniz even noted the similarity between binary digits and the yin-yang symbols of the Chinese I Ching, a cross-cultural connection that feels almost prescient in hindsight.

The true leap toward modern computing came in the 19th century, driven not by war or trade, but by the need

The true leap toward modern computing came in the 19th century, driven not by war or trade, but by the need to eliminate fatal errors in British naval navigation. In 1831, a miscalculation in the Royal Navy’s official *Nautical Ephemeris*—the annual book of star positions and tidal tables used to guide ships across the Atlantic—led a troop transport to run aground off the coast of Ireland, killing more than 200 soldiers. The error was not a one-off: human “computers,” the term then used for the teams of mathematicians who did all arithmetic by hand, regularly made mistakes in the tens of thousands of calculations required to produce the *Ephemeris* each year. Charles Babbage, a Cambridge mathematics professor who had already built a small mechanical calculator to automate tax tables for his father, saw the disaster as a call to action. If a machine could be built to perform these calculations automatically, he reasoned, it would not only save lives but also free human computers to do more creative work.

Babbage’s first project, the Difference Engine, was designed to automate the calculation of polynomial functions used for the *Ephemeris*. Built from thousands of precision brass gears, cranks, and levers, it used the mathematical principle of finite differences to eliminate the need for multiplication and division, reducing all calculations to a series of additions. After a decade of work and a £17,000 government grant (equivalent to roughly £2.2 million today), Babbage completed a working prototype of a smaller version of the engine in 1822, which could calculate polynomials up to 6 decimal places with perfect accuracy. But when he asked for funding to build a full-scale version that could calculate 20-digit polynomials, the British government balked at the rising costs and technical complexity, withdrawing support in 1842. The half-built full-scale engine was scrapped, its parts sold for scrap; only a small portion of the prototype survives today, on display at the London Science Museum. Babbage was undeterred. If a machine could only calculate fixed polynomials