The Good Winter

There is a particular feeling in the air right now, and most people I talk to have it whether or not they have a word for it. The sense that the floor is being replaced while we are still standing on it. Institutions we assumed were load-bearing turn out to be decorative. A pandemic, a land war in Europe, prices that lurched, an artificial intelligence that went from party trick to co-worker in about eighteen months. It reads as chaos. I want to argue it is something more specific, and much more hopeful: a season.

The frame is the Fourth Turning, from the historians William Strauss and Neil Howe. Their idea is that history in the modern West moves in a long cycle — a saeculum, roughly the length of one human life, eighty to a hundred years — and that this cycle has four moods, like seasons. A confident High, a rebellious Awakening, a fraying Unraveling, and then a Crisis: a winter, when the old order breaks down and a new one is forged in a hurry. We are, on their reading, deep in the winter.

Before I go further, the honest asterisk, because this whole essay is allergic to hype and I am not going to make an exception for the idea it is built on. The Fourth Turning is a lens, not a law of physics. It is unfalsifiable in the way horoscopes are unfalsifiable — flexible enough to fit almost any decade if you squint. Strauss and Howe drew a clean pattern through American history, and clean patterns drawn through messy history should always be held at arm's length. So I am not asking you to believe the calendar. I am asking you to keep the one genuinely useful thing inside it: crises are generative. Winters end in construction. And for that claim, there is a second, more rigorous body of work that says the same thing with actual data.

The rhyme that has receipts

The economist Carlota Perez spent her career studying how transformative technologies actually roll out, and she found a pattern so regular it is almost embarrassing. Every great technology — canals, railways, steel, oil and the motor car, and now information and AI — arrives in two acts. First an installation phase: finance rushes in, a bubble inflates, everyone gets giddy, and then it crashes. Then, after the crash, a deployment phase: the technology stops being a casino chip and quietly seeps into ordinary life. Roads, factories, homes, hospitals. The deployment phase is the real prize. Perez's name for it is the golden age.

Railways had their mania and their 1840s crash, then built the Victorian economy. Electricity and the car had their roaring twenties and 1929, then — after the worst winter the last saeculum had — became the post-war boom. Notice the shape: the golden age comes after the crisis, and is built by the people who lived through it. If Perez is right, the dot-com bubble was our installation mania, 2008 was the crash, and we are now early in the deployment of computing and AI. The fireworks are behind us. The building is ahead.

This also dissolves the most tiresome objection to new technology — "if it is so revolutionary, where is the payoff?" In 1987 the economist Robert Solow quipped that "you can see the computer age everywhere but in the productivity statistics." He was right, and then he was wrong. The historian Paul David explained why: when electric motors arrived, factories kept their old layouts — one giant central engine, belts everywhere — and saw almost nothing. The gains came decades later, once a generation of managers rebuilt the factory around the new thing. Technology waits for us to reorganise around it. AI is at exactly that awkward stage now: everywhere, and not yet in the numbers. That gap is not failure. It is the pause before deployment.

We are not our great-grandparents

Here is where I part company with the doom merchants, and I want to do it with data rather than a motivational poster. If we are heading into a hard, generative decade, the single most important fact is that we are entering it with tools the last winter could not have dreamed of. The generation that faced the 1930s did it without antibiotics at scale, without the internet, with most of the planet unable to read. Compare the starting line.

None of that says the world is fine. Plenty is broken, and some of it is getting worse. It says the trajectory of human capability has been relentlessly, measurably upward — the point Hans Rosling spent his life making in Factfulness, armed with the World Bank and Our World in Data rather than vibes. And the deepest advantage is one that does not fit on a chart: the marginal cost of knowledge has collapsed to roughly zero. A curious teenager in a small town — I was one, in Jhansi, with a borrowed machine and a single black-covered BASIC book — now has the same lectures, the same open-source tools, the same documentation, and now the same AI tutor as a graduate student at a famous university. That has never been true before in the history of our species. Not once.

What a winter is actually good for

Downturns feel like the wrong time to start things. The data says the opposite, almost rudely. General Electric, IBM, Disney, HP, Microsoft — all born in slumps. The last real winter's aftermath gave the United States the GI Bill and a mass middle class. And the most recent crash is the tidiest example of all: Airbnb, Uber, WhatsApp, Square, Stripe, Slack — a remarkable share of the last decade's defining companies were founded in or right after 2008–09. Recessions are cruel to incumbents and kind to builders, because the incumbents freeze and the talent shakes loose and the rent goes down.

The winter also quietly rewrites what a skill is. When facts were scarce and expensive, being a walking encyclopaedia was a career. Now facts are free and a model will recite them on demand, so the value moves up a level — to judgment, taste, the ability to frame a problem, and the plain willingness to keep learning. Erik Brynjolfsson calls the trap of pointing AI at replacing people rather than augmenting them the "Turing Trap," and it is the whole ballgame. The studies that have come in are encouraging on exactly this point: give a support agent an AI assistant and the least experienced workers improve the most; give developers or writers a good model and the floor rises faster than the ceiling. The tool is a leveller when we let it be one. Which of those two stories we get is, as ever, a choice.

Where it is being built

The winter is global, but the answers are wonderfully local. I have worked across enough of these places — India, the Gulf, the United States, and now the Netherlands — to have lost the habit of thinking any one of them has the whole recipe. They each hold a different piece.

You cannot talk about any of this without Robert Shiller's narrative economics: the idea that the stories an economy tells itself spread like contagions and then move real money, real capital, real careers. (I wrote about it at length in Narrative Economics.) Silicon Valley's genius is partly a story — that failure is honourable and the swing is worth taking — and the story summons the capital that makes it true.

The Netherlands, where I now live, runs two complementary stories a train ride apart. Eindhoven's is the least glamorous and most underrated in Europe: that a modest region can out-cooperate far bigger rivals, with industry, universities and government sharing a single roadmap — the machine that prints the world's most advanced chips comes from there. Amsterdam supplies the other half, the capital-and-talent city where European scale-ups actually scale. Deep tech in one town, growth capital in the other. The wider European backdrop is Mario Draghi's blunt 2024 report: a continent magnificent at inventing and too timid at scaling, sitting on world-class science it keeps letting slip abroad.

Germany is the most instructive story on the continent right now, because it is a superb engine being asked to rebuild itself while still running. The car is the economy — Volkswagen, Mercedes, BMW, Bosch — and it is in a white-knuckle pivot to electric under real pressure from China's BYD, on top of the energy shock that followed the war in Ukraine. Munich holds the deep-tech and industrial crown (BMW, Siemens, the defence-AI upstart Helsing); Berlin is the scrappier startup capital (N26, Trade Republic, Zalando); and Aleph Alpha and DeepL fly a European flag in AI. The famed Mittelstand — thousands of quiet "hidden champion" firms — is being tested on whether its patience can become reinvention. It is this whole essay compressed into one country: winter forcing an incumbent to become a builder again.

India is running the story I find most quietly thrilling. Bengaluru's is the public-rails bet: build an open identity and payments layer once — India Stack, and UPI now moving more than ten billion payments a month — and let a million businesses bloom on infrastructure nobody has to rebuild. But the part that moves me most is that the story no longer stops at the metros. Pune — the "Detroit of the East", where Tata and Bajaj and Mercedes build cars, wrapped around a university town and an IT belt — shows the physical and digital economies reinforcing each other. Behind it, a wave of tier-2 and tier-3 cities — Indore, Coimbatore, Jaipur, Nagpur, Kochi — are leapfrogging on absurdly cheap mobile data and a decade of new highways, airports and freight corridors. Mumbai holds the money, and an informal, improvisational resilience the spreadsheets never fully capture. India's real advantage is not one city; it is depth — a young population and public infrastructure carrying opportunity into places the last century's growth never reached.

And the board is geopolitical now, which is itself a winter symptom. The old story — frictionless globalisation, build anywhere, ship everywhere — is fracturing into something warier. Chips are fought over between the United States, China, Taiwan and the Netherlands, an export-control war waged largely through that one company near Eindhoven. Energy was re-routed overnight when the war in Ukraine turned pipelines into leverage. Supply chains are being "re-shored" and "friend-shored" closer to home. Economists gave the slowdown a clumsy name — slowbalisation — but the human meaning is simple: where things get built is becoming a question of trust again, and that reshuffles the entire map. The winter is a contest of narratives as much as of technologies — and, the encouraging part, narratives can be changed on purpose.

Work came unbundled from place

There is a quieter revolution threaded through all of this, and it is the one most likely to touch your own life: work has come unbundled from place. For most of history, opportunity was geography — you got the jobs your city happened to have. The pandemic broke that assumption in about a fortnight, and it has not fully snapped back. A developer in a tier-3 Indian town, an illustrator in Lisbon, a data engineer in Nairobi can now sell into the richest markets on Earth without moving to them. Cheap data and public digital rails mean the small-town builder I once was would today have not only the same learning as a graduate student at a famous university, but the same market. That is genuinely new, and it is enormous.

The gig and creator economies are the same coin. On the bright side, a real lowering of the gate: anyone can start a business of one, monetise a narrow skill, and find the hundred true customers who care. On the shadow side — and I want to be honest about it — precarity. Platform work can offer the flexibility of being your own boss with none of the safety net that used to come attached: no sick pay, no floor, an algorithm for a capricious manager. This is exactly where the next question stops being abstract. As more of us live gig to gig, the case for a modest, unconditional floor under everyone stops sounding utopian and starts sounding like plumbing — which is the honest way into the two ideas that get argued about most.

Breaking two spells

If we are going to be visionary, we have to be ruthless about the hype, because hype is how good ideas get discredited. Two spells need breaking.

The first is the Singularity — Ray Kurzweil's forecast of a moment, around 2045, when machine intelligence runs away from us into something unrecognisable. It is a genuinely serious idea surrounded by an unserious cult. The corrective is Roy Amara's law, the wisest sentence in technology: we overestimate the effect of a technology in the short run and underestimate it in the long run. Both halves matter. The people promising artificial general intelligence by next Tuesday and the people insisting it is all a stochastic parrot are usually making the same mistake from opposite ends — mistaking the steep part of a curve for the whole curve. The grounded position is the boring one: transformative, yes; on a human timescale, with human institutions dragging on the wheels, and no fixed date. Skeptics like Rodney Brooks and Gary Marcus are not spoilsports here; they are the friction that keeps the wheel from spinning off.

The second spell is universal basic income, which gets discussed as either the utopia that saves us from the robots or the handout that ruins us, and is neither. We now have real trials instead of arguments. Finland gave 2,000 people a basic income for two years: wellbeing clearly improved, employment barely moved — neither the promised surge nor the feared collapse. Stockton, California handed 125 residents $500 a month; full-time employment among them actually rose, and the money went to the dull, load-bearing stuff — bills, repairs, groceries. Sam Altman's OpenResearch ran the largest version, $1,000 a month to a few thousand Americans, and found recipients worked slightly less but chose more deliberately, spent more on essentials and on helping others. The honest reading is neither salvation nor apocalypse. It is a policy tool with evidence — modest, humane, no miracle. That is not a disappointing answer. In a field this overheated, an evidence-shaped answer is the exciting one.

The green thread runs through all of it

The same winter that is remaking work is forcing the largest re-engineering project in history — the energy transition — and here the news is genuinely, structurally good. Solar has followed its own quiet Moore's law (Swanson's law: costs fall by roughly a fifth with every doubling of production), and in 2020 the International Energy Agency called it flatly "the cheapest source of electricity in history." Batteries are tracing the same curve. For the first time, the clean choice and the cheap choice are converging, which means decarbonisation stops depending on virtue and starts depending on arithmetic — a far more reliable engine.

The catch is real and I have written about it honestly elsewhere — AI itself has a growing physical footprint, and pretending otherwise is its own kind of hype (see The weight of a thought). But the frame that says technology versus the planet is a false binary. The economist Elinor Ostrom won a Nobel for demonstrating, with decades of fieldwork, that communities can and do govern shared resources — forests, fisheries, water — without either privatising them or wrecking them, as long as they build the right rules together. The "tragedy of the commons" is not a law of nature. It is a failure of cooperation, and cooperation is a technology we happen to be very good at when a crisis makes us remember how.

The oldest technology is each other

Which brings me to the part the spreadsheets always miss. Every Fourth Turning in the record ends the same way — not with a gadget, but with people rebuilding the institutions and the solidarity the previous era let rot. The New Deal, the post-war international order, the mass expansion of education: winters are when societies remember they are societies. You can already see the small version of it — the open-source movement, which is quietly the largest act of mutual aid in economic history; the mutual-aid networks that self-assembled overnight in 2020; the plain human instinct, under pressure, to turn toward each other rather than away.

I find that steadying, because it means the most futuristic thing you can do in this decade is also the oldest: be useful to the people around you. The winter does not reward the person with the cleverest take about it. It rewards the person who plants something.

So plant something

If you have read this far, you did not come for doom, and I have none to sell. Here is what I actually believe, stripped of hedging:

This is not a bad time to be alive and building. It is arguably the best one there has ever been to start — the tools are free, the incumbents are distracted, the capital is looking for conviction, and the problems are large enough to be worth a life. Learn in public; the compounding is absurd. Pick a real problem over a hyped one, and pick deployment over fireworks. Find your version of a Brainport — a room where industry and curiosity and generosity sit at the same table — and if there isn't one, that is a thing you are now allowed to build. Bet on the clean option because it is becoming the cheap option. And help the people near you, not as charity but as strategy, because that is what actually holds when the floor is being replaced.

Winter is not the end of the story. It is the part where you decide what the spring is made of. That decision is not made by history, or by the machines, or by whichever narrative shouts loudest this quarter. It is made by what we choose to plant, now, in the cold — and I cannot think of better weather for it.

Sources & further reading: William Strauss & Neil Howe, The Fourth Turning (1997) and Neil Howe, The Fourth Turning Is Here (2023); Carlota Perez, Technological Revolutions and Financial Capital (2002); Robert Shiller, Narrative Economics (2019); Erik Brynjolfsson, "The Turing Trap" (2022) and productivity studies with Li & Raymond (2023); Hans Rosling, Factfulness (2018), and Our World in Data (Max Roser and colleagues) for the human-progress figures; World Bank poverty data; International Energy Agency, World Energy Outlook 2020 (solar as "the cheapest electricity in history"); Ray Kurzweil, The Singularity Is Nearer (2024), with Roy Amara's law and the skepticism of Rodney Brooks and Gary Marcus; the Finnish basic-income experiment (Kela, 2020), the Stockton SEED study (West & Castro Baker, 2021), and OpenResearch's unconditional cash study (2024); Elinor Ostrom, Governing the Commons (1990); Paul David, "The Dynamo and the Computer" (1990); and Mario Draghi's report on European competitiveness (2024). Figures are rounded, best-available estimates and vary by source and method.