The Adoption-Adaptation Gap: 12,000 Years of Technology Outrunning Civilization
Two Months
That’s how long it took ChatGPT to reach 100 million users. No product in history had moved with such speed.
The telephone took 75 years to reach 50 million people. Radio took 38 years. Then TV, 13 years. And in our lifetimes, we’ve only seen the tightening of this cycle.
Internet, phones, apps, and now AI all the way down to two months.
Sure, adoption works in mysterious ways, but one clear trend is apparent. It’s getting faster. But that’s not the most interesting trend in this article.
But there’s a more interesting range… 50 to 150 years.
That’s how long it has historically taken for civilization, its institutions, its culture, and its philosophies to adapt to major technological transitions. Not just adopt the technology, but adapt to what it means and build the frameworks that let societies absorb it without tearing themselves apart.
“Adapt” is worth defining precisely because it’s the crux of the argument. Adoption is when you use the thing. Adaptation is when your society rebuilds its governing logic around the thing.
Including what it makes and what it destroys.
We adopted smartphones in about five years. We still haven’t adapted. Our democratic institutions weren’t designed for algorithmically amplified outrage cycles. Our mental health infrastructure wasn’t designed for adolescent social comparison at scale. Our civic epistemics are running on hardware shaped by print-era democracy operating inside a social media environment.
That’s adaptation lag. It’s the gap between the technology in your pocket and the civilization designed to survive it.
Mobile gaming forced me to think about the adoption of technology. In 2013, I remember helping set up user acquisition campaigns on Facebook ads and seeing firsthand the access to detailed information on our “targets.” In those days you could hyper-target using anything available on profiles, which is nearly everything. Even in those early days, it was obvious we were living within a digital wild west without any guardrails. Typical in the early adaptation phase of new technology.
Decades later, we’re still culturally grappling with the introduction of such heavy digital consumption and the aftermath of terminal connection.
History has shown us that adaptation takes fifty years on the fast end.
Now look at these two numbers together. Adoption: two months. Adaptation: minimum fifty years. That gap has been present in every major technological transition in human history.
It has never been this wide.
Contents
- Two Months
- The Two Clocks
- Twelve Thousand Years Ago: The Agricultural Revolution
- Five Thousand Years Ago: The Bronze Age Collapse
- The Strangest Thing That Ever Happened
- Twenty-Five Hundred Years Ago: Classical Empires and the Limits of Scale
- Thirteen Hundred Years Ago: The Islamic Golden Age and Its Fracture
- Six Hundred Years Ago: Print, Gunpowder, and 200 Years of Religious War
- For Science!
- Two Hundred and Fifty Years Ago: The Industrial Revolution and the 166-Year Lag
- A Hundred and Fifty Years Ago: Electricity and the 70-Year Labor War
- Eighty Years Ago: Nuclear, Mass Media, and the Permanent Standoff
- Thirty Years Ago: The Internet and Its Governance Vacuum
- The One That Worked
- Every Single Time
- Back to Two Months
- K=0.73
The Two Clocks
Every major technological transition runs two clocks simultaneously.
Adoption and adaptation.
The adoption clock is easy. It measures how quickly a new technology spreads through society.
The adaptation clock is harder to read: it measures how quickly institutions, culture, and philosophy catch up to what that technology makes possible and, maybe more importantly, what it destroys.
Now, let’s apply these clocks in real terms.
The Agricultural Revolution took 5,000 years to spread globally. 500 years for the self-reflective moral philosophy of the Axial Age to distribute. Skip forward, and you have the printing press: 50 years to restructure European access to information. Steam engine: 50 years to industrialize Britain. Electrification: 30 years to reach mass adoption. Internet: 7 years. Smartphones: 5 years. ChatGPT: a few months.
That’s the adoption clock, and just like the conveyor belt of our favorite TikTok trends, it keeps accelerating.
Now the adaptation clock in short order. It took 3,000 years for agricultural civilization to develop stable philosophical frameworks. The Axial Age spanned 400 years to spread a new morality. Then, 200 years for the printing press to produce the Peace of Westphalia. 150 years for the steam engine to produce functional labor protections. 75 years for electrification to produce the New Deal. The internet loop is still open after 30 years. And now, AI’s adaptation clock: unknown.
3,000. 400. 200. 150. 75. ???.
| Loop | Era | Adoption | Adaptation Lag | Status |
| Agricultural Revolution | ~10,000 BCE | ~5,000 years | ~3,000 years | Resolved |
| Bronze Age Empires | ~3,500 BCE | ~1,000 years | Never | Collapsed |
| Axial Age | ~800 BCE | ~600 years | ~400 years | Resolved |
| Classical Empires | ~500 BCE | Decades | ~500 years | Collapsed |
| Islamic Golden Age | ~700 CE | ~100 years | ~200 years | Resolved |
| Print + Gunpowder | ~1440 | ~50 years | ~200 years | Resolved (1648) |
| Scientific Revolution | ~1550 | ~100 years | ~200 years | Resolved (~1780) |
| Industrial Revolution | ~1760 | ~50 years | ~150 years | Resolved (1930s) |
| Electricity + 2nd Industrial | ~1870 | ~30 years | ~70 years | Partially resolved |
| Nuclear + Mass Media | ~1940 | ~5-20 years | ~80 years (ongoing) | Open |
| Digital + Internet | ~1993 | ~15 years | >30 years (ongoing) | Open |
| AI | ~2022 | ~2 years | ??? | ??? |
For most of human history, these two clocks moved at roughly the same speed. Agriculture spread over millennia. The philosophical frameworks that made sense of agricultural civilization, divine kingship, and the cosmological legitimacy of the ruling class and the state (as natural and inevitable) developed over roughly the same timeframe.
Slow technology with slow adaptation. The clocks broadly matched.
Now that synchronization is gone. Stewart Brand named this pace layering: civilization runs on multiple clocks at different speeds, with technology fastest and culture slowest. The relative desynchronization is what’s new.
And every loop runs the same three steps. A technology creates new surplus or new information asymmetries. It concentrates something, food, force, knowledge, and/or coordination capability, that was previously dispersed.
The existing institutions, built for the previous regime, fail to manage the new distribution of power. Then a philosophical revolution produces the new legitimating framework. This usually comes from the people who “lost” in the transition, becoming the threat that shapes how it is resolved. Look familiar?
Surplus, then institutional failure, then philosophical reinvention.
All propelled by Schumpeterian creative destruction as the engine. What’s new in the AI iteration is the speed at which destruction outruns creation.
The losers push back, often violently, philosophy absorbs the critique (eventually), and a new stability emerges. Then the next technology arrives. Faster and more supernatural.
Now, let’s walk through these cycles, starting at the beginning.
Twelve Thousand Years Ago: The Agricultural Revolution
Agricultural Revolution (~10,000 BCE). Adoption: ~5,000 years. Adaptation: ~3,000 years.
Agriculture spread across the world over roughly 5,000 years, emerging independently in at least seven regions: the Fertile Crescent, China, Mesoamerica, the Andes, West Africa, New Guinea, and eastern North America.
Each one discovered you could grow more calories than you could hunt or gather. Then they stopped moving. That’s the load-bearing innovation with everything else built on top of it.
Here’s the paradox: early agricultural populations were measurably worse off as individuals. Archaeological health data is consistent and somewhat alarming.
Average stature among early farmers was 3-4 inches shorter than that of the hunter-gatherers they replaced. Dental issues, essentially absent in forager populations, were common in agricultural ones. Infectious disease rates rose dramatically with sedentism and population density. Even individual life expectancy may have decreased.
Why did agriculture spread if it made people individually worse off? Competition.
Groups that farmed could support larger populations per square mile. Larger populations meant greater resource allocation, higher specialization, and, of course, larger armies. Agricultural societies could absorb population losses and keep fighting (literally and metaphorically). Forager bands couldn’t.
Group-level dominance was the only metric that mattered, and agriculture had it decisively.
Storeable surplus restructured everything downstream. It created a class of people who could live without farming: priests to manage the cosmological legitimacy of planting cycles, warriors to defend the stores, and administrators to track the accounts.
The earliest surviving writing, Sumerian cuneiform, includes inventory, grain stocks, labor allocations, and simple receipts.
The philosophical response followed the institutional reality: divine hierarchy, kingship as cosmic order, and gods who owned the land and chose the ruler. It fit perfectly with the technology that produced it. If your survival depended on natural cycles you couldn’t control, you built a worldview organized around appeasing those cycles.
And then there’s the Y-chromosome bottleneck. Somewhere between 6,000 and 4,000 years ago, the Y-chromosome diversity of the human species collapsed dramatically. Female lineages kept diversifying normally, but male lineages didn’t.
The leading explanation, from a 2018 Nature Communications paper by Zeng et al., is patrilineal clan warfare: mobile pastoral groups raiding agricultural surplus and eliminating rival male lines. The genetic record we carry in our Y-chromosomes may be the founding document of a pattern that never stopped: losers of the age pushing back.
There it is, roughly seven thousand years from the first harvest to codified law.
That’s our first run through. Now each cycle is a compression of our two clocks.
Five Thousand Years Ago: The Bronze Age Collapse
Bronze Age Empires and Their Collapse (~3,500 BCE). Adoption: ~1,000 years. Adaptation: never completed before the collapse.
Bronze requires copper and tin, which rarely occur naturally in the same location. The Bronze Age was, at its core, a coordination technology: an empire was the structure needed to source, transport, smelt, and distribute a metal whose components were often thousands of miles apart.
The result was the first genuinely complex civilization. Sumer, Egypt, Mycenae, the Hittites, and Shang China: all organized around palace economies, centralized redistribution of agricultural surplus, professional militaries, and legal codes. With divine kingship as the philosophical system legitimizing the whole arrangement.
Then, around 1,200 BCE, in roughly fifty years, almost every Bronze Age palace economy in the Eastern Mediterranean collapsed simultaneously.
We have their last letters.
A tablet excavated from Ugarit, dated around 1185 BCE:
*“My father, behold, the enemy’s ships came here… The cities were burned, and they did evil things in my country.” *
Another from the same archive:
*“There is no time. Send whatever you can. The enemy is killing us.” *
Emergency dispatches that never reached their destinations, because there was nowhere to send them.
The Bronze Age Collapse is the loop running in reverse: existing institutions over-extended and collapsed faster than any adaptation could follow. The highly interdependent palace economy system, the same integration that had made it powerful, made it fragile. When one node failed, the failure cascaded.
What emerged from the reset was iron technology: cheaper, more distributed than bronze, accessible without a palace economy. Anyone could smelt iron, and most importantly, no empire was required. The technology that replaced the Bronze Age’s coordination infrastructure was its opposite: distributed and scalable without central control. (A theme you all know)
The collapse of one loop created the conditions for the next.
New technology concentrates surplus. The displaced attack it.
Step one: bronze coordination, enabling the first complex states. Step two: interdependency so tight that when one node failed, everything cascaded. Step three: never arrived. The loop collapsed before philosophy could close it. Iron filled the vacuum. Distributed, accessible, no palace required.
Sometimes the next loop starts from rubble.
The Strangest Thing That Ever Happened
The Axial Age (~800 BCE). Step three of the agricultural and Bronze Age loops, arriving ~3,000 years late as a philosophical reinvention rather than a new technology.
Over roughly 600 years, across four completely disconnected civilizations, reflective moral philosophy emerged simultaneously. Buddha (~500 BCE) in northern India, Confucius (~500 BCE) in China, Socrates (~470-399 BCE) in Greece, and the Hebrew prophets in Judea (930-580 BCE).
None knew the others existed.
All made the same fundamental move: from embedded cosmology to reflective ethics. From “this is how the world works” to “this is how the world should work.”
Before the Axial Age, morality was about correct ritual action toward the gods. A “did you perform the sacrifice correctly?” righteousness. After this Age, morality evolved about inner disposition and universal obligation: how should you treat another person, any person, regardless of their relationship to your tribe or your gods?
That shift is the foundation of what we now call ethics, and it arrived near-simultaneously across four unconnected civilizations.
Same fundamental question, different answers, arrived at independently across four civilizations. Each tradition made the same move differently.
- Buddha’s First Noble Truth established that suffering is universal, not just personal: the framework for compassion as a category that crosses tribal lines.
- Confucius’ ren (benevolence) made human-heartedness the foundation of right governance.
- Socrates’ claim that “the unexamined life is not worth living” established individual reflective consciousness as the focus of philosophical inquiry.
- The Hebrew prophets introduced the radical idea that God cares about justice, not just worship: how you treat others matters more than how you perform the sacrifice. Karl Jaspers named this the Axial Age in 1949. He never fully explained it, and still today no one has.
The two strongest explanations are complementary. The Iron Age agricultural surplus had freed the first tangible intellectual class, giving it the unprecedented benefit of free time. You can’t have philosophy without people whose survival is so secure that they have time to sit and think.
And all four zones featured the same political structure: multiple small competing states, no single empire imposing orthodoxy, and intellectuals moving between courts.
Twenty-Five Hundred Years Ago: Classical Empires and the Limits of Scale
Classical Empires and Their Decline (~500 BCE). Adoption: decades. Adaptation: institutions never scaled to match.
The Axial Age philosophies became state technologies fast.
Then the over-extension problem hit.
Rome’s institutions were designed to manage a city-state and a modest regional empire. By 200 CE, they were managing a continent-spanning system whose coordination costs exceeded the surplus they could extract. The debasement of the denarius tells the story: 95% silver under Augustus (14 CE), under 5% silver by the 270s CE. (Frighteningly similar to the last century of USD)
When you can’t collect enough tax to pay the army, you debase the currency.
When you debase the currency, prices rise. When prices rise faster than wages, you lose the soldiers. Diocletian’s Edict on Maximum Prices (301 CE) tried to fix prices across the empire, producing immediate black markets and supply shortages.
Price ceilings produce shortages, get worked around, and don’t fix the underlying problem. Same in 301 CE, same in 2026. The institutional responses to structural problems became new structural problems.
The fall of Rome was the slow kind: a century-long step-two failure where institutions designed for one scale couldn’t manage the continent-spanning system they’d ended up with.
Augustine’s City of God (413-426 CE) was written as Rome was actively falling, and it did something philosophically essential: it reframed the collapse. The earthly city was always going to be impermanent.
The heavenly city was the real one. The argument held that medieval Christendom needed something to legitimate the institutions that came after Rome. Augustine had already built it.
What emerged from the collapse was a cultural technology designed for preservation: the monastery. Monasteries were the backup drives for classical knowledge. Monte Cassino, Lindisfarne, the Irish monasteries: without them, Europe loses Virgil, Cicero, Augustine, and most of what came after.
Augustine wrote City of God while Rome was actively falling. Sometimes step three is just a lifeboat. Don’t lose what we already know. Hold until something else can emerge. Even if it’s a slow decay over five hundred years.
Thirteen Hundred Years Ago: The Islamic Golden Age and Its Fracture
Islamic Golden Age (~700 CE). Adoption: ~100 years. Adaptation: ~200 years.
The Islamic world between 700 and 1300 CE became the dominant knowledge coordination layer on Earth.
The coordination technology that enabled it: paper. Chinese in origin, paper reached the Islamic world around 750 CE via prisoners captured at the Battle of Talas. The shift from papyrus and parchment dropped manuscript production costs by an order of magnitude. Knowledge that had required a scriptorium suddenly required a market stall.
What followed was specific and transformative: Arabic numerals replaced Roman ones (try long-dividing MMDCCXVIII by XLVI). These numbers made the formulation of algebra much easier. And finally, optics and the early scientific method it needed.
Ibn al-Haytham’s Book of Optics (1011 CE) established the experimental method 600 years before the European scientific revolution. The House of Wisdom in Baghdad, founded around 830 CE, was built explicitly to accumulate and synthesize knowledge from multiple civilizations: Greek, Persian, Indian, and Syrian.
The Renaissance ran through Baghdad and Córdoba long before it ever reached Florence.
The Toledo School of Translators (1100s-1200s) was the specific mechanism by which Islamic science re-entered European universities. Gerard of Cremona alone translated 87 works from Arabic into Latin, including Euclid, Ptolemy, Avicenna, and al-Kindi. European universities were teaching from Arabic translations of Greek texts that had been lost to Europe for centuries.
Then the loop started closing. Al-Ghazali’s Incoherence of the Philosophers (1095 CE) argued that rationalist philosophy was incompatible with Islamic faith, targeting the Aristotelian tradition on which the Golden Age had been built.
The Mongol sack of Baghdad in 1258 CE closed the loop violently. The House of Wisdom was destroyed, with reports that the Tigris ran black for days from the ink of dissolving books.
The knowledge that survived, survived because of European translators.
I guess Al-Ghazali won the argument. The Mongols ended the institution, and step three never arrived within Islam. It exported itself to Europe through translators and got credited to the wrong civilization. Two hundred years, and Florence got the Renaissance that Baghdad built.
Six Hundred Years Ago: Print, Gunpowder, and 200 Years of Religious War
Print and Gunpowder (~1440). Adoption: ~50 years. Adaptation: ~200 years.
Two monopolies fell within two centuries of each other, and together they ended the medieval system.
Today’s trend of asymmetric warfare follows the gunpowder script: democratized violence at a fraction of the previous cost. Traditionally, a trained knight represented roughly a decade of investment from childhood. A musketeer required weeks and changed the brute economics applied to violence.
Artillery made castle walls economically unviable: the cost of a cannon battery exceeded what any individual feudal lord could sustain. Only centralized royal treasuries could finance the new warfare. The military monopoly that sustained feudal nobility collapsed with nation-states filling the vacuum. Step two, arriving fast.
The press dissolved the Church’s information monopoly. In 1440, books were hand-copied manuscripts accessible only to the clergy and the wealthy. Within fifty years of Gutenberg, twenty million books were in circulation across Europe. Martin Luther’s 95 Theses spread across Germany in two weeks in 1517. The first event in history that functioned like viral content distribution.
Luther himself noted: “They are printed and circulated far beyond my expectation.” Classic humility when going viral.
Printers made commercial decisions to publish in vernacular languages rather than Latin: larger markets and much better margins. Those standardized print languages became the cultural substrate for nationalism. Readers developed an imagined sense of community with others reading the same language, whom they’d never met.
Benedict Anderson called this “imagined communities.” What you were reading shaped who you thought “we” were.
Then there’s Abbot Johannes Trithemius. In 1492, he wrote In Praise of Scribes, a comprehensive defense of manuscript hand-copying against the printing press. His argument was serious: print would democratize error, undermine spiritual discipline, and flood the culture with low-quality text. He wasn’t entirely wrong. Just very, very early.
He had it typeset and distributed.
The Ottoman Empire watched all of this and made a deliberate choice. The Sultans delayed the adoption of the press for nearly 200 years after Gutenberg, not out of ignorance but out of institutional calculation. The scribal class and the religious establishment’s interpretive monopoly were too valuable to risk.
The first Ottoman press didn’t operate until 1727. Two centuries of opting out, followed by two centuries of paying for it when modernizing rivals arrived with the benefits of mass literacy. A smart bet for the upper class at the time, but catastrophic across two centuries. We’re still bad at telling those apart in advance.
The adaptation timeline: 1440 Gutenberg. 1517 Luther’s 95 Theses. 1524 German Peasants’ War (100,000 dead). 1534 English Reformation. 1555 Peace of Augsburg. 1618 Thirty Years’ War begins.
Finally, in 1648, the Peace of Westphalia created a new organizing principle: territorial sovereignty. The religious question stayed open, but the infrastructure for living alongside it was the actual deliverable.
Two hundred and eight years after Gutenberg, Westphalia gave Europe an organizing principle that let competing powers coexist without agreeing on God. That’s step three working as designed: the original question remains unresolved, the infrastructure for coexisting with it is built, and civilization continues.
For Science!
Scientific Revolution and Enlightenment (~1600). Adoption: ~100 years. Adaptation: ~150 years.
The period reads like a Hall of Fame lineup, and for good reason. It laid the fundamental groundwork of all our modern scientific frameworks.
Bacon (empiricism), Descartes (rationalism), Newton (mechanical universe), Locke (individual rights and consent of the governed), Voltaire, and Hume.
The press enabled more than the Reformation. The cumulative, correctable, cross-border scientific method also rode in on it. Knowledge became something you could publish, critique, refute, and build on. Truth became a shared project rather than a revealed one.
The Scottish Enlightenment is worth isolating because it’s where philosophy and engineering converged most visibly. Adam Smith, David Hume, and James Watt were contemporaries in the same city, publishing within decades of each other.
Smith’s Wealth of Nations (1776) and Watt’s steam engine (1769) landed in the same intellectual environment. Same city, same generation, within the same conversations.
The Enlightenment was geographically as well as philosophically tight.
The specific institutional products were consequential: the 1776 Declaration of Independence, the 1789 US Constitution, the 1791 Bill of Rights, and the 1807 British Slave Trade Act. Each required an underlying philosophical framework, individual rights, consent of the governed, and the state as contract rather than divine inheritance, which the printing press had spent 150 years creating.
Step three of the print loop. One hundred fifty years late. The printing press generated the Enlightenment, not the other way around. Then industrialization arrived, and the clock reset.
Two Hundred and Fifty Years Ago: The Industrial Revolution and the 166-Year Lag
Industrial Revolution (~1760). Adoption: ~50 years. Adaptation: ~150 years.
The steam engine severed manufacturing from geography. Factories concentrated in cities. And in doing so, created something that hadn’t existed before: the wage laborer.
Before industrialization, artisans owned their output. A cobbler owned the shoes. A weaver owned the cloth. Craft was identity, skill was capital, and the product was yours. The factory system stripped this. Workers sold their time and owned nothing of what they made.
The Luddites (1811-1816) were skilled artisans whose decade-long apprenticeships had been rendered worthless overnight. The technophobe label was a later misreading. Parliament responded with the Frame-Breaking Act, making machine destruction a capital offense. Twelve thousand troops were deployed to the affected regions.
The state treated this as a civilizational threat.
Because in their time, it was.
Friedrich Engels spent 1843 and 1844 in Manchester, the first industrial city on Earth, and documented the average life expectancy between the working-class (28 years) and the upper class (38 years).
Different worlds inside the same city, but the results speak for themselves. The Industrial Revolution was measurably killing people faster than the previous agricultural economy had. The economics didn’t penalize this because the political system couldn’t afford to.
This single structural change produced the three dominant political philosophies of the 19th century with almost robotic clarity.
Liberalism named the bourgeoisie’s interest as universal. Socialism organized the proletariat to claim its own. Conservatism defended an aristocracy already losing the fight to both.
Three classes, three philosophies, three accounts of who got robbed. All three are still arguing about it 200 years later.
Then the second wave of the Industrial Revolution, which we’ll see next, was brought on by electricity. Essentially resetting the board without any conclusion for adaptation and simply waiting for governments to solve for both.
The philosophical counter-movement: Romanticism. Wordsworth and Coleridge against the dark satanic mills and nature as a redemptive force (sound familiar). Imagination over the rationality of industrialization. Romanticism’s critique survived as the seed of every later movement that tried to recover what industrial rationality had destroyed.
One hundred sixty-six years. Watt’s engine to the Wagner Act. The gap between adoption and adaptation.
A Hundred and Fifty Years Ago: Electricity and the 70-Year Labor War
Electricity and the Second Industrial Revolution (~1870). Adoption: ~30 years. Adaptation: ~70 years, incomplete when WWII hit.
In 1910, roughly 10% of American homes had electricity. By 1940, that number was 90%. Thirty years of infrastructure buildout, transforming the basic structure of daily life beyond having light in the house.
Electricity restructured time itself, with artificial light extending the average workday beyond sunset. Frederick Winslow Taylor introduced scientific management in 1911, the same year as the Triangle Shirtwaist fire. The fire killed 146 garment workers, mostly young immigrant women, locked inside a factory with no fire escapes.
The owners were acquitted, but the deaths catalyzed the International Ladies’ Garment Workers’ Union and a wave of labor legislation.
The continued pattern of adaptation paid for in lives.
Soon after, Ford’s assembly line (1913) combined electricity and Taylorism to strip craft from production. A skilled machinist’s knowledge became worthless overnight, and output per worker multiplied.
For a time, wages stayed flat.
The Fordist bargain, paying workers enough to buy the goods they were producing, was Henry Ford’s recognition that the system was eating its own market. Adaptation from within, not governance, not philosophy, but industrial logic catching up to industrial reality. (I should note Ford, in shrewdness, wanted his own workers to prop up demand for their new vehicles.)
The philosophical response went in two directions.
- Pragmatism (Dewey, James) concluded that truth is what works, so philosophy as engineering: if it ships, it’s true.
- Freud, who proposed that a civilization running on industrial rationality was producing an interior life it couldn’t manage. The unconscious as the residue of everything repressed to keep the machine running. The political response was more violent. Fascism and Stalinism both emerged from the same structural conditions: the second industrial transformation arrived faster than democratic institutions could adapt.
The ideology gap was filled with movements that offered coordination through force rather than through the slower machinery of liberal governance. Germany’s Weimar Republic had excellent institutions on paper. Excellent on paper means very little when theory is tested. The republic didn’t have the social fabric to metabolize simultaneous economic collapse and technological displacement.
Fascism and Stalinism are what institutional failure looks like when ideology fills the gap before governance can.
Lastly, the New Deal was step three. It just took seventy years after the machines. Enacted by Franklin D. Roosevelt, the New Deal aimed at providing Relief, Recovery, and Reform during the Great Depression. It significantly expanded the government’s role in the economy and created a social safety net, but it never reached its logical conclusion with the interruption of WWII.
Eighty Years Ago: Nuclear, Mass Media, and the Permanent Standoff
Nuclear and Mass Media (~1940). Adoption: 5-20 years. Adaptation: still running.
Nuclear is where the list gets serious.
Every transition before this one had real costs, but nothing to this scale. Sure, displacement, economic orders coming apart, and skill sets going obsolete overnight were painful (sometimes for generations). Nuclear sits in a different category.
The cost to the species.
The Manhattan Project went from theoretical to operational in three years. Three years from “this might be possible” to a working bomb. That’s an adoption clock that should make you uncomfortable.
The only governance response that exists for nuclear weapons is Mutually Assured Destruction (MAD): two parties, each with the authority to end civilization, agreeing not to, because doing so would end their own civilization simultaneously.
Not exactly ideal governance, but more of a standoff. Eighty years later, it remains the primary mechanism for managing the most dangerous technology humanity has ever built.
Mass media arrived in the same breath. Radio gave politicians something they’d never had before: a way to speak directly into millions of homes at once, in an intimate register, and bypassing every editor and gatekeeper that had previously shaped political messaging.
Most famously, Roosevelt’s fireside chats rebuilt public trust in the federal government during the Depression. Most infamously, Hitler’s radio addresses produced something else entirely. Same technology, same adoption speed, but wildly different outcomes, depending simply on who held the microphone.
Then, on September 26, 1960, the Nixon-Kennedy debate rewrote the rules again. Radio listeners thought Nixon won. Television viewers thought Kennedy won. Same words, same arguments, same debate. It was the first time in history that separate media produced different political realities. The image replaced the sentence as the primary unit of political legitimacy on that evening.
Recent elections highlight that our democratic institutions still haven’t finished adapting.
The philosophical response was immediate (also one of my favorite periods), with existentialism (Sartre, Camus, Beauvoir) emerging from a civilization that had produced Auschwitz and Hiroshima within the same decade.
The question it kept asking wasn’t rhetorical but dialectical.
How do you create meaning in a world that could be ended by a political decision made by people you’ve never met?
Absurdism and Postmodernism are the conclusion that no framework for meaning, or the institutions that built it, could be fully trusted. Philosophy born out of survival more than academia, acting as an emergency framework for living after the initial adaptation failed.
When two fingers are hovering over buttons, and the powers that be are calling it a doctrine, I can imagine anyone getting a little existential.
So, that’s the governance for the most dangerous technology humanity has ever produced. The philosophical reinvention arrived: existentialism, postmodernism, but only in the cultural register. In the political register, we never moved past the standoff.
We’re still figuring it out eighty years later.
Thirty Years Ago: The Internet and Its Governance Vacuum
Digital and Internet (~1993). Adoption: 15 years. Adaptation: still running.
Thirty years in, and we still don’t know how to govern the epistemic commons the internet created.
Actually, we’ve given it the old college try. Section 230 of the Communications Decency Act, passed in 1996, was the first major institutional response.
It was twenty-six words…
“*No provider or user of an interactive computer service shall be treated as the publisher or speaker of any information provided by another information content provider.” *
Those twenty-six words shaped the entire social media era by shielding platforms from liability for user content. Written three years after the web browser existed. For the scale that arrived fifteen years later, it was like using an 1880s building permit to regulate a nuclear plant.
The EU’s GDPR arrived in 2018, twenty-five years after the web went public. The US still doesn’t have a federal equivalent. Companies had to rush to police themselves before the government stepped in. Led by Apple in 2021, when it released its own App Tracking Transparency (ATT).
This is the adaptation clock doing exactly what it does.
Google solved the access problem by giving everyone access to every piece of information ever digitized, reachable from nearly anywhere. Genuinely transformative. For the first time in human history, every fact was searchable from a desk with a monitor. That was step one. Information asymmetry dissolved.
Then social media solved the distribution problem and created a different one: engagement optimization as the de facto governance system of public discourse.
The echo chamber is what optimizing for engagement at scale produces. The system is working exactly as designed. When you optimize for engagement, you get outrage, fear, and tribal confirmation, because those are the behavioral impulses most reliably triggered at scale.
The algorithm found this and kept running.
Like the gaming industry, most businesses were using this information to target their products and services. Cambridge Analytica flipped this use case, using the platform’s own targeting infrastructure against its users. Nothing about it was a hack. The data had been there for years, available to anyone who knew how to query it. The surprise was the application: aiming consumer-grade targeting infrastructure at voters instead of consumers. That’s step two, arriving without a law to break.
The smartphone was the second wave within the same loop. The web reached roughly a billion people over fifteen years. The smartphone put a persistent ambient connection in the pockets of three billion more people in ten. Two adoption events within one loop, each compressing faster than the last.
The printing press produced the Reformation and the Enlightenment. The steam engine produced liberalism, socialism, and conservatism. The internet produced…
We’re still mid-loop. We don’t even have a name for what it produced. Maybe the digitalization of reality? Your guess is as good as mine.
And that’s the adaptation failure in one sentence.
No step three. The surplus shifted mid-loop: information democratized, then attention replaced it as the concentrated resource before anyone could govern the first one. The next loop started before this one closed.
We’ve now walked the pattern twelve thousand years deep. Let me show you the one time it actually worked.
The One That Worked
The loop needs a counterexample before we get to AI. Otherwise, this reads as fatalistic.
The Meiji Restoration (1868-1912) is the most instructive successful transition in the record. Japan went from feudal isolation to industrialized nation-state in roughly 40 years.
The Tokugawa regime had maintained deliberate isolation for 250 years. When Commodore Perry’s 1853 naval demonstration made that model existentially untenable, Japan chose rapid adoption over continued resistance.
The Iwakura Mission (1871-1873) was the spearhead.
Fifty Japanese officials spent 18 months traveling across the United States and Europe. They studied everything. How a post office works, how banking is structured, how a modern military is trained, and how a public school system is administered. They came back and built versions of what they’d seen.
The specific institutional adoptions followed with unusual speed: universal public education under the Gakusei law (1872), land tax reform creating modern property rights (1873), the Bank of Japan (1882), the Meiji Constitution (1889), a conscript army replacing the samurai class (1873-1890). Reform went further here than in any contemporary Western case. The samurai class was abolished outright, the warrior caste broken.
The Meiji slogan was Wakon Yosai, “Japanese spirit, Western learning.” The adaptation was deliberate, state-directed, and expensive. Not without violence (the Satsuma Rebellion of 1877, 40,000 dead). But the civilization didn’t collapse. The loop completed in 40 years rather than 200.
What Meiji suggests: the adaptation gap is a function of institutional choice. Societies that recognize the pattern and steer deliberately can compress the adjustment period.
But the Meiji story has a cost it tends to omit: the same centralized nationalism, the same institutional efficiency, the same state-directed modernization that survived the industrial loop became the engine of Imperial Japan’s military expansion thirty years later. The adaptation solved one problem and opened the next loop.
The adaptation clock reset immediately.
What would a modern Meiji moment look like for AI? The Iwakura Mission is the template, and it seems the US and China are trying to deploy it, but something seems off.
As we’ve seen, getting this right is an uncommon part of this loop.
Every Single Time
Before going further, a pattern worth naming explicitly.
Every technology in this list generated a cultural panic that recycled the same structure, regardless of whether the underlying concern was justified.
The Lesewut (“reading fever”) was a medical diagnosis applied to novel-reading in the 1740s through 1860s. Women were considered especially vulnerable. In the 1860s, British Medical Journal articles documented “railway spine,” a supposed neurological condition caused by train vibration and speed. The telegraph prompted George Beard’s 1881 treatise “American Nervousness,” diagnosing the first technology-induced stress disorder. The telephone, according to 1920s critics, would destroy family communication by replacing face-to-face interaction with impersonal remote connection. Radio caused violence. Comics caused juvenile delinquency, Television rotted minds, and Video games caused mass shootings.
In 2011 the US Supreme Court, reviewing violence studies on video games, noted: “These studies have been rejected by every court to consider them, and with good reason.”
In 1836, John Stuart Mill wrote of “a state of society where any voice, not pitched in an exaggerated key, is lost in the hubbub.”
Mill was describing the newspaper. The complaint runs two centuries deep before Twitter ever existed. The pattern repeats with every cycle. The technology changes, the moral panic finds it, and a generation later nobody remembers what the fuss was about.
The pattern: utopian adoption, moral panic from incumbent gatekeepers, poorly designed regulation, normalization, nostalgia. Then the old technology becomes the baseline of decency against which the next one is judged dangerous. The printing press was Sodom. Now it’s literature. Television was rotting minds. Now it’s prestige drama.
There’s a mechanism here worth naming.
Vilification.
And it consistently comes from two groups. Incumbent gatekeepers who stand to lose power: the scribal class against print, the music industry against MP3s. And people whose lives were disrupted by the transition: the Luddites lost their livelihoods, which is plenty of reason. Both voices contain a real signal. Both consistently misframe the threat as categorical and unique, when the pattern is consistent and recurring.
That consistent failure matters enormously when evaluating AI alarm. And it doesn’t settle the question.
Back to Two Months
So, ChatGPT.
Two months and 100 million users.
I’ve been sitting at the knife-edge of technological applications for nearly two decades, and it’s become crystal clear that AI is different. It’s much more than the typical hype-cycle bs. The continuation of our trend through democratization has moved from systems, politics, and information and wedged itself into our personal space.
Every AI system I’ve built has reinforced this for me. The agents are better at skills than I am. Right now, they can do nearly anything with the correct guidance (and patience), but they’re not better at knowing what matters… yet.
Remember, if this were a child, the parents would still be telling you the child’s age in months. It’s early. But that doesn’t excuse it from the cycle’s conditions.
The vilification cycle is already running. Anthropic as enemy of the state, Sam Altman before Congress, the EU AI Act, and of course, datacenters. Not to mention AI 2027, Turing Award laureates publishing probability-of-doom estimates, and frontier lab employees quitting to sound the alarm without a PR team chasing after them. The historical pattern of over-panicking gives us grounds for skepticism.
But here’s where the comparison to every prior technology breaks down.
Every previous technology panic was about what a technology does to human behavior. Reading makes women hysterical. Comics cause delinquency. Video games cause violence. All claims about the effects on how individual people act.
The serious AI safety argument, from the people who built these systems, is a different category of claim. This one is about whether humans stay relevant to the decisions that matter… at all.
Behavior and agency aren’t the same concern. The historical pattern of over-panicking about behavioral effects doesn’t tell you much about whether a categorical threat to agency is real.
There’s also a loop-within-the-loop nobody is governing. AI is being used to accelerate AI research. Beyond solving protein folding, AlphaFold compressed a research timeline that would have taken another decade of human scientific labor into months.
AI is already shortening the gap between AI milestones. The adoption clock may be self-accelerating in a way no previous technology was: the technology is improving its own rate of improvement. AI accelerates the entire stack, and this includes the parts we wouldn’t choose to accelerate.
Three hyperscalers control roughly 60% of cloud AI infrastructure. A hundred companies control roughly 40% of global private AI investment. The data shows the concentration pattern of an installation phase. Three companies hold the keys to reasoning capability for the species. Whatever governance gets built has to be built around that fact. The bubble and the golden age are both still ahead.
Carlota Perez’s framework on technological revolutions and financial capital maps this exactly: the installation phase concentrates, the deployment phase overshoots, and the regulatory response always lags.
The governance gap is specific. The EU AI Act doesn’t adequately cover general-purpose AI systems. US executive orders on AI don’t survive administrations. There is no international coordination body with meaningful authority over frontier AI development. The institutions being built to govern this technology were designed for a previous technological regime.
Institutional inertia at an industrial scale.
Again, we’re at month thirty. The adoption clock: two years. The adaptation clock: the better part of a century, minimum.
That gap has a name now.
K=0.73
Let’s zoom all the way out.
In 1964, Soviet astronomer Nikolai Kardashev proposed measuring civilizational progress by energy. A Type I civilization harnesses the full energy of its home planet. Type II, its star. Type III, its galaxy.
Carl Sagan extended this into a continuous logarithmic scale. On that scale, humanity today sits at approximately K=0.73. Type I is K=1.0.
The gap doesn’t sound large. In energy terms, it’s roughly 500 times more power than every power plant, oil well, solar farm, and wind turbine on Earth running simultaneously. At sustained growth, Michio Kaku estimates 100-200 years to get there, assuming we sustain growth and assuming we don’t destroy ourselves first.
You can’t reach Type I with 195 nation-states in uncoordinated competition. Planetary-scale energy infrastructure requires planetary-scale coordination: climate management, the suppression of self-destructive technologies, and the unified allocation of resources across borders.
The Paris Agreement represents the most serious attempt at planetary coordination in history. It’s voluntary, non-binding, and insufficient. Getting from K=0.73 to K=1.0 is a governance problem masquerading as an engineering problem.
And governance problems run on the adaptation clock.
Now compress the full arc into the Kardashev frame.
The 12,000-year story of technological civilization is a project of increasing energy capture and coordination complexity. Agriculture raised the floor. The Bronze Age built the first states. The classical empires scaled them until they couldn’t. The Islamic Golden Age preserved and extended what the collapses didn’t destroy. Print and gunpowder ended the old power monopolies. The Enlightenment built the legitimating framework for what came next. Industrialization multiplied energy capture by an order of magnitude. Electricity restructured time. Nuclear crossed the first threshold into non-solar energy. Digital made every piece of human knowledge simultaneously accessible and simultaneously contested. AI is beginning to restructure how that knowledge is generated and applied.
Each transition arrived faster than the last. Each one demanded institutional and philosophical adaptation on a timeline that never compressed to match. Each was harder than the previous, with less time to land.
Agriculture spread over five millennia and its philosophical response took three thousand years. The printing press spread across Europe in fifty years and its resolution took two centuries. The internet reached a billion people in seven years and we’re still mid-adaptation. One AI company reached a billion users in two years.
The institutions are slower than the technology, and that’s the problem we don’t have an answer to.
If the Fermi Paradox is real, if the universe is genuinely 13.8 billion years old with billions of potentially habitable planets and we’re surrounded by silence, one answer is that the K=0.73 to K=1.0 corridor is where civilizations typically end.
Civilizations more often end through coordination failure than catastrophe. The gap between what a civilization can build and what it can govern is the filter that catches them.
We’re at K=0.73. The technologies that could kill us or carry us through are being built simultaneously, by the same people, in the same decade. The philosophical frameworks for managing them run 50 to 150 years behind. The institutions governing them were designed for a previous regime, and in our case maybe before that one.
But the loop is running ever faster.
The question is whether this iteration completes the project.
Or confirms what the silence of the universe already tells us.