Recombination: The Value of General-Purpose Technology Lies in the Ruins of Old Forms

——The First Act of the Technological Revolution Trilogy

Every new general-purpose technology is initially dressed in the old clothes of the previous era: films imitate plays, telephones imitate telegrams, television imitates radio, and websites imitate brochures. True revolution does not begin with the invention of a technology, but with the moment someone sheds that old skin.

A Law of Iron: New Technologies Always Mimic Old Forms

Reviewing 140 years of technological history, you will find a nearly unbroken rule: For the first ten to twenty years after a new general-purpose technology appears, it is used to mimic the form of the old. People are not lacking in enthusiasm—quite the opposite, each time it is a grand rush to embrace it—but they lack the imagination to envision a new form. Enthusiasm is more than enough to replicate the old world; it is woefully insufficient to create a new one.

Electricity mimics steam. The central power station was built in 1882, but it was not until 1900 that the power from electricity accounted for less than 5% of the mechanical power in American factories. The first “electrical factories” simply replaced steam engines with large motors, continuing to use central drive shafts and belts to power the entire factory—everything else remained the same. Electricity was merely more expensive steam. The real benefits only came when “single-machine drive” became widespread: each machine had its own motor, allowing factories to be laid out according to process flow rather than power transmission, making assembly lines possible. Economic historian Paul David found that it took about forty years from the technology’s availability to a productivity surge.

Film mimics theater. For the first dozen years after the invention of the camera, most films were simply a fixed camera in the fifth row of a theater filming a live stage play: fixed camera angles, wide shots, scene changes. It was not until Griffith and others invented close-ups, montage, and parallel editing that people realized the camera was not “a cheaper theater seat”—it could shift perspectives, compress time, and get close to a face. Film became an independent art when it stopped imitating theater.

The telephone mimics telegraph and newspapers. In 1893, Budapest introduced “telephone news”: subscribers would listen to the broadcast of news and stock prices at a fixed time. The Bell system in America positioned the telephone as a serious business tool, with ads showing orders and market data, and companies discouraged users from “chitchat,” seeing it as a waste of lines. In reality, the original use of the telephone was discovered by users—especially the neglected homemakers in the business world: maintaining relationships and social connections. It took decades for the industry to recognize that the largest market for the telephone was not in transmitting information but in providing companionship.

Television mimics radio, and websites mimic printed materials. Early television was simply “broadcast with pictures,” replicating radio dramas on screen; it was not until the advent of live sports, news broadcasts, and sitcoms that television became television. Ninety years later, it was the same: the first websites were mocked as “brochureware,” with newspapers simply moving their print versions to the web, bleeding to death; the banner ad, born in 1994, mimicked print ads and was sold by position—until Google’s AdWords created ads that followed search intent rather than position. Only then did the internet’s largest business model emerge: one structurally impossible in print.

The mobile internet repeats the cycle. In the WAP era, operators squeezed web pages into small screens, struggling for nearly a decade. The true mobile species look like this: Uber = GPS + mobile payments + idle vehicles + real-time matching—none of the four elements were invented by Uber, but the combination of “ride-hailing” was impossible in the PC era. Similarly, VisiCalc in 1979 did not invent any hardware; it reorganized “calculation” into “what-if scenarios”: accountants could ask “What if the interest rate were 8%?” and see the entire table recalculate automatically. Countless people bought an Apple II to run this software—personal computing’s first killer app was a reorganization.

This law does not just rule commerce. It ruled the battlefield—and on the battlefield, the cost of being wrong is not discounted.

Gunpowder served kings for centuries until the revolution taught it to speak. After gunpowder was introduced to Europe, it was long confined to the old military attire: mercenaries, line tactics, noble officer corps—just louder bows. In 1793, the French Republic, besieged both at home and abroad, issued a general mobilization order, creating something an old dynasty could not: citizen soldiers. The state no longer paid to hire soldiers but made “every man a soldier” part of citizenship—recruits were now nearly infinite. Napoleon then rewrote war around this new organization: the corps system, local requisitioning, and using mobility to replace supply lines. The opposing old monarchies had the same cannons and gunpowder, but lacked the “citizen” organizational form—they were not defeated by the new weapons, but by the new social structure around those weapons. After the disastrous defeat at Jena in 1806, the Prussian reformers’ conclusion was starkly clear: to defeat this army, one must first replicate that reorganization.

Sixty years later, Prussia became the reorganizer. In 1870, France and Prussia had roughly contemporary railroads and telegraphs. Moltke’s General Staff wove mobilization plans, railway timetables, and staff officer corps into a machine—railway departments became one of the most powerful departments in the entire army, the war was fought on paper long before it was fought on the ground. France’s railways were “faster carriages”: mobilization was a spontaneous performance, soldiers could not find their units at the station, officers searched files on the platform. The outcome at Sedan was decided before the war began—it was the victory of reorganizing around new technology over trying to stuff new technology into old organizations. A military historian’s summary can be given to every CEO: Prussia won with the General Staff, and the General Staff was an organizational invention, not a weapon. (For how this machine later turned around and held its master hostage, see the chapter “The Timetable War”—no matter how well the reorganization answer is written, it will become the prompt for the next exam question.)

Why Everyone Answers Wrong: Mimicry is Not Stupidity, but Gravity

Mimicking the old form is not because that generation was stupid, but because three “gravities” are at work simultaneously:

Cognitive Gravity. Humans can only understand the unknown through what they already know. The first car was called a “horseless carriage,” the first films were called “moving pictures,” and to this day, broadcast and television news anchors are still called “anchors.” Language reveals our thinking: the names we give to new things are like old clothes draped over them.
Asset Gravity. The factories, transmission shafts, and skilled workers of steam-powered plants are still not fully depreciated. Reorganizing around new technologies means writing down the value of still-profitable assets—on the books, this is “destroying value,” but strategically, it is the only way forward.
Organizational Gravity. Every position, reporting line, and KPI in old processes grew out of constraints around old technologies. Reorganizing processes means reallocating power, and an organization’s immune system will precisely kill it.

This explains a counterintuitive phenomenon: the native form is almost always found by outsiders or marginal figures—Griffith, unbound by the theater; the homemakers overlooked by Bell; and two young men in a garage. They are not smarter, but they are free from the gravitational pull of the status quo.

How to Find the Native Form: Three Actionable Steps

First, Change Your Question. A mundane question is: “How can this technology make what I do now cheaper?” A native question is: “What impossible things does this technology make possible?” The answer to the former might be a 30% cost reduction, but the latter could be assembly lines, montage, and AdWords. A practical thought experiment: Assume all old constraints are gone (power must be centralized, information must be printed, intelligence must be provided by humans), and design from scratch. Would you still make it the same way? Any answer of “no” is an entry point for reorganization.

Second, Watch Users’ Misuses. The social uses of the telephone, the text messages originally used by engineers as test channels, and businesspeople using Excel as a database—native uses often appear first in users’ hands in unconventional ways. Users’ workarounds, hacks, and “misuses” are the original texts of the new forms yet to be named.

Third, Beware of “Imitation Period Prosperity.” Imitating old forms often works in the short term—“broadcast with pictures” did have viewers, and “brochure websites” did save on printing costs. This mild success is the most dangerous, as it makes organizations believe they have completed their transformation. The standard is cold: If your new technology can be described as “cheaper/faster old stuff,” you are still in the imitation period.

For the Age of Compute

Apply this iron law to today: What is most of today’s “AI transformation”?—Putting a model into a step of the old process to make it cheaper. That’s like electrifying a steam factory, like a “horseless carriage,” like a brochure website.

The real question should be: When “intelligence” becomes nearly free, which design assumptions underlying our processes, products, and organizational forms—assumptions built around intelligence being scarce and expensive—have now been destroyed? Today’s company structures—layered reporting, functional division, approval chains—are fundamentally built around the constraint of “limited and expensive human cognitive bandwidth,” just as steam factories were built around a single transmission shaft. When this constraint is loosened, the question is not “Can AI help someone write a weekly report?” but “Does the structure that requires a weekly report still hold?”

The answer in the age of electricity took forty years to emerge. This time, the forty-year clock starts now—and the only way to shorten it is to stop spending time electrifying steam-powered factories.