Four Accelerations That Compressed Time and Space: From Rails to Self-Driving

The coal and cotton of the Industrial Revolution had to be moved, but canals were slow and froze in winter. Colliery engineer George Stephenson (1781–1848) bet on a faster route: let the steam engine run itself along rails. The Stockton & Darlington Railway he built in 1825 was the world's first public steam railway.

The Rainhill Trials of 1829 were the decisive verdict — rival locomotives competed head to head, and the Stephensons' "Rocket" won at roughly 48 km/h, cementing the steam locomotive as the mainstream design. The next year the Liverpool & Manchester Railway opened as the first purpose-built steam trunk line. Historian Wolfgang Schivelbusch, in The Railway Journey (1977), wrote that the railway "annihilated space and time": passengers crossed landscapes at unprecedented speed, and even perception was reshaped. More far-reaching still — to coordinate timetables, in 1883 America's private railway companies carved the country into four standard time zones. The unification of time was accomplished by railroads, not governments.

Once an infrastructure locks in a standard (gauge, format, connector), latecomers struggle to dislodge it even with a better design. Today's charging-plug wars and communication-protocol battles replay the same "first-mover lock-in" path dependence.

In the early 1930s, airlines survived almost entirely on government mail subsidies — carrying passengers alone lost money. American Airlines president C. R. Smith wanted an aircraft that could keep a company alive "on ticket sales alone," and turned to aircraft designer Donald Douglas.

In 1935 the DC-3 first flew: it carried 21 passengers at roughly 300 km/h, rugged and reliable. For the first time, an airline could profit on passenger service without mail subsidies. By 1939 the DC-3 carried about 90% of all U.S. air traffic. In WWII, its military version — the C-47 — was mass-produced by the thousands as the backbone of Allied airlift. Its triumph rested on no single dazzling invention, but on a combination: all-metal monoplane, variable-pitch propellers, dependable engines — like the medieval heavy plow, a victory of the "boring combination."

What decides a technology's fate is often not its performance ceiling, but when it crosses the profitability threshold — the moment it goes from "surviving on subsidies" to "earning its own way." Today's electric cars and solar power sit right on, or just past, that line.

In 2004, the State Council's Medium- and Long-Term Railway Network Plan set out the blueprint for high-speed rail. China chose a path of "import–digest–absorb–reinnovate": between 2004 and 2006 it negotiated with Kawasaki, Siemens, Alstom, and Bombardier, on the condition of market for technology — vast orders in exchange for core blueprints.

In 2008, the first 350 km/h line — the Beijing–Tianjin intercity — opened just before the Olympics, and network-building surged. But on July 23, 2011, two trains collided on the Wenzhou line, killing 40 people and shocking the nation — a decision point: halt here, or fix and continue? (Railway minister Liu Zhijun had already fallen to corruption charges that February.) China chose to slow down, rectify, and press on. By 2024, high-speed track in operation reached about 45,000 km — roughly 70% of the world's total, firmly the longest network on Earth.

The "market for technology" catch-up model mirrors today's industrial-policy contests over electric cars and AI: trading scale and speed for a narrowed technology gap, while wagering thinned safety margins and debt as the price.

In 2004, not a single vehicle finished the U.S. DARPA autonomous-car challenge; the next year, Stanford's Sebastian Thrun and his "Stanley" won, igniting the field. Google (later spun off as Waymo) took the baton and pushed it onto real roads.

On the night of March 18, 2018, in Tempe, Arizona, an Uber self-driving test car struck and killed Elaine Herzberg, who was wheeling a bicycle across the road — the world's first autonomous-vehicle pedestrian fatality. The investigation surfaced disturbing details: the system had detected her, yet failed to brake in time due to a logic flaw, while the safety driver was distracted. This became a decision point: who is responsible — the algorithm, the company, the safety driver, or the victim? Uber suspended testing, and the whole industry was forced to rethink the design of its "safety fallback."

This is the shared question for every AI system reaching the real world: when a machine makes a consequential decision on a person's behalf, how is responsibility distributed among the algorithm, the deployer, and the user? Self-driving is merely the first domain to drag the abstract "trolley problem" into a real courtroom.