TL;DR: I had a machine read every essay on this site and tell me what I’ve been writing about. The answer took one sentence: “everything you value is something else, stored”. A flattering worldview, right up until you notice the one thing that can’t be stored is the thing everything else is priced in.

I did something halfway between vanity and archaeology: I fed all Huliwood essays into an AI and asked it what I’ve been saying.

Three and a half years, 99 essays. Finance, colonial history, consciousness, caffeine, techno-feudalism, sourdough.

I’d told you the archive was a record of curiosity going wherever it pleased – no lane, no niche, the refusal to pick one being the whole point.

The machine disagreed. It took the full essay catalogue and handed back one sentence.

Everything you value is something else, stored.

Yes. I asked a machine what I think, and it answered faster than I could have.

I wanted to argue. Then it showed me the list.

99 costumes but a *** ain’t one

• Money stores time – I wrote a whole piece calling it a battery: my corporate years charged into a brokerage account.

• Culture stores winters: what we flatter ourselves by calling work ethic is ten thousand Octobers of prepare or die, compressed into software and handed down.

• National advantage stores older deposits – cousin-marriage bans, plagues survived, geographic luck compounding for millennia until one afternoon in 1532 it walks into a square in Cajamarca and cashes itself out as an empire.

• Your gut stores data – pattern recognition passed off as mysticism.

• A brand stores language: Google won so hard the verb pays rent.

• An LLM stores judgment: everything we ever wrote, compressed until it talks back.

• Even the self, when I went looking for it, turned out to be storage: a pattern re-saving itself onto new atoms, a whirlpool holding shape while the water changes.

So that’s the move, then. Pick up the thing everyone treats as solid – money, tradition, credentials, home – and shake it until the realer thing inside rattles. I’d been running one algorithm in countless costumes.

As worldviews go, you could do worse. It has a certain unified-field-theory swagger to it, and for about a day I walked around feeling like a man who had finally read his own manual.

Then I remembered Potosí.

The mountain that ate Spain

The story goes like this: a Quechua herder spends a cold night on a mountain in the Andes, lights a fire to keep warm, and at dawn finds silver melted out of the rocks under the embers. The mountain turns out to be laced with it.

Spain spends the next century shipping home the greatest store of value the world had ever assembled – for a stretch of the 1500s, nearly half the silver on Earth came out of that one hill in what is now Bolivia.

And what did the greatest store of value in history buy the empire that owned it?

Four sovereign defaults before the century was out. Prices did something Spain didn’t have a word for yet: they inflated.

The silver moved through Sevilla like water through a colander – to Genoese bankers, Flemish weavers, Chinese silk merchants – and Spain ended its golden age with the mountain half-hollow and the crown broke.

I know this story cold, and I’ve spent three essays on what it points at – fiat money, pension systems, a financial ledger floating free of the real one – always pointed outward. What I hadn’t done, until the machine rubbed my nose in my own archive, is point it at the sentence I’m proudest of.

The crack

That sentence is four words long: money is stored time.

Readers liked it. I liked it. I built my whole exit on it – sold everything I owned, charged six and a half years of labour into a portfolio, told myself the battery would hold until I decided what it was for.

It’s wrong by one word, though. Money was never stored time.

Money is a claim on other people’s time: the barista’s minutes, the pilot’s hours, a stranger’s afternoon, redeemable on demand.

Your own hours sit in no vault. They run out at one an hour whether you use them or not. You can trade them while they pass, but you can’t bank them.

Money stores time the way a gym membership stores fitness.

The Germans, as usual, have a word they weren’t planning to share: Torschlusspanik – gate-shut panic. Medieval towns locked their gates at dusk; arrive late and you spent the night outside the walls with the wolves and the tax collectors.

The word outlived the gates because the feeling did. Every store of value has a redemption window, and the window is a life.

Call it the Redemption Problem. We’ve spent five hundred years getting brilliant at storing value (double-entry ledgers, joint-stock companies, index funds, cold wallets) and we treat redemption as a detail that sorts itself out later.

Storing value has a whole industry behind it. Cashing it back out into a life has an annuity desk, a few bestsellers, and no heroes.

I can’t name one person history celebrates for cashing out well. (Diocletian, maybe – quit the largest empire on Earth to grow cabbages, and even then history mostly remembers the palace.)

My own book

To be clear: I happily shit on give my unsolicited opinions on anything and anyone, but only under duress when it comes to myself.

When I decided to sell everything and leave, I ran the decision through a spreadsheet first – scenario tabs, sensitivity tables, a model to tell me whether I was allowed to want what I wanted.

More than three years later the number it was guarding has grown. My willingness to draw it down has not.

Somewhere along the way the runway stopped being something to take off from and became something to protect. I check it more often than a man who trusted his decision would.

The pattern doesn’t stop at the portfolio.

I wrote a formula for where to live and I’m still stalling on its answer. I preach living now and keep putting it off.

I once staged a funeral for my corporate self and spent half the service checking whether the coffin had a return policy.

My father ran the same trade in a different asset class – a lifetime hunting a home that fits a feeling he can’t name, and this year the closest he ever came went to the highest bidder: the house in Cabo de Palos from The Last Leash.

I took notes. I changed the asset class. I’m not sure I changed anything else.

The hours get spent whether I decide how or not – deferring just means I don’t get a say.

The other side of the book

Compounding is real, and interrupting it early is how people stay poor.

The only reason I can sit here on a weekday writing fifteen hundred words about redemption is that for six and a half years I didn’t redeem anything. Deferral bought this.

The presence-preachers with no runway aren’t free; they’re one landlord letter away from the machine, and their carpe diem has a copay.

When a monk tells you to live in the moment, somebody else is growing the monastery’s rice.

So the miser is right and the monk is right, and I’ve spent three years failing to make them agree.

The obvious move is a formula – spend x% of the store per year, rebalance toward presence with age, some Kelly criterion for being alive (the gambler’s formula for bet size).

I know better than to trust it: a formula for living now is just another way of putting it off. I notice I’m still looking for one.

A man who reads his own essays and finds a single hidden spine should be the first suspect, not the last – I’ve written before that the map of knowledge has no center, and that every map was drawn by someone.

So maybe the spine isn’t in the archive at all. Maybe it’s in the man reading it.

This is essay number 100.

This is the cornerstone – the one I’d point a stranger to first – and I’d planned a verdict, the way the others mostly have one.

It doesn’t have one. And I’m still at my desk, charging the battery, writing about the sun instead of sitting in it.

Subscribe now

Share

Related: Money is Stored Time – the battery, the first time · Don’t Mid-Curve It – the spreadsheet that guarded the number.

Sources: Drelichman & Voth, Lending to the Borrower from Hell (2014) – Philip II’s four defaults · Potosí, UNESCO · Kris Lane, Aeon – Potosí’s share of the world’s silver

TL;DR: Last December I called two former Bitcoin miners with big grid connections “AI landlords” and planned for them to be 44% of my portfolio. On 29 October a 145-page essay argued that owning power for AI is a bottleneck: a toll that lasts only until someone builds around it. What separates a landlord from a bottleneck is how long the contracts run, and by that test one of my companies is a landlord and the other isn’t. Since October I’ve kept buying both, most recently on the night the Nasdaq-100 had its worst day since March 2020.

There’s a race on my homepage. I put it there in September: IREN and Cipher, the core of my AI Landlords essay, against the Nasdaq-100, the index most people mean when they say “tech stocks”. It gets redrawn every week, whether I’m winning or not. The idea was simple: intuition should have a scoreboard.

The scoreboard works. It starts at the close on 12 December 2025, the Friday before I made the call. By the close on 6 November, my two landlords had lost 46% in euros, weighted the way I held them at the start. The Nasdaq-100 had gained 13%.

I first bought IREN in July 2025, five months before the essay, and Cipher in September. Counted from those first buys, the pair is up about a fifth in euros, a few points behind the Nasdaq-100. The months before I wrote the thesis did all the earning.

My landlords led for most of the first half of the year. By the end of September the gap was embarrassing, and by the end of October it wasn’t a race any more.

On 29 October, the day Alphabet fell more than 12% on doubts about what its AI spending will earn, IREN and Cipher each lost about 9%. The same day, someone published the clearest case I’ve read for why the gap might never close.

So this piece is me testing my own call against it. I want to work out two things: whether owning power for AI earns a landlord’s rent or a bottleneck’s toll, and which of my two companies is which. Then whether my money agrees.

The essay says power is a toll with an expiry date

The Commanding Heights, by Will Manidis and Joe Roberts, is 145 pages of Lenin, Gutenberg, Standard Oil, the Gospel of Matthew and a man in Bangalore signing off his AI agents’ reports without reading them.

The argument starts with the price of AI. Epoch AI, a research group that tracks it, finds that the cost of reaching a given level of performance falls about 47% a quarter, so whoever sells the answers is in a price war.

The money goes instead to whoever owns the crossings AI has to pass through to reach people’s wallets – the app already on their phone, the way they pay, the license to do the job, the government itself. The authors call these crossings the Commanding Heights, after Lenin, who in 1922 defended letting private trade back into Soviet Russia because the state still held the railways and heavy industry.

Then they draw a line I hadn’t drawn. A Height is a toll that lasts. A bottleneck is a toll that lasts until someone builds around it.

Their example of a bottleneck is my thesis in all but name: the AI Landlords, my bet that whoever already holds scarce grid power collects the rent from AI. A finished data center, fully financed, sits dark because it can’t get connected to the grid, so the owner hands his margin to whoever has the power.

They fit my whole essay into one paragraph. Then they keep going: “If power is solved, transmission becomes the crossing. So [the cloud giant] moves power generation on site. Then chips become the crossing.”

(In the chapter on what to buy, the two company names are blacked out, and the site has a form to request a copy. My picks are on my homepage, losing in public. I’m not sure which of us has the better business model.)

The length of the lease decides who’s a landlord

Picture a flat in a city with a housing shortage. You can rent it out on a fifteen-year lease. You lock in today’s high rent, and if the city builds ten thousand new flats next year, your tenant still owes you every month until the lease runs out.

Or you can furnish it and rent it out a few years at a time. You earn more while the shortage lasts. But when it ends, the next tenant pays the new, lower rent – and by then the furniture’s worn out and you’re buying new.

The first is a landlord: the high rent is in writing for 15 years, whenever the shortage ends. The second is the essay’s bottleneck: its rent falls at the first renewal after the shortage ends.

What decides which one you are is the contract – how long it runs, and whether the tenant can keep paying. Neither is one of the essay’s Heights (15 years isn’t forever) but only the first outlasts the shortage.

IREN and Cipher both started as Bitcoin miners, which means they spent years securing cheap electricity and the right to draw huge amounts of it from the grid. When AI came along, they kept the grid connections and started switching from mining Bitcoin to serving AI customers.

For scale, one megawatt running all year powers about 800 American homes, and these companies deal in hundreds of megawatts at a time.

Cipher is a landlord

Cipher Digital rents out the empty flat: the building and the power. Its tenants bring their own chips.

• Black Pearl, Texas: about $5.5 billion over 15 years. The tenant is Amazon. Rent on the first phase started in August.

• Stingray, Texas: about $2.0 billion over 15 years. Amazon again.

• Barber Lake, Texas: about $3.8 billion over ten years with Fluidstack, an AI cloud company, backed in part by Google. In September an unnamed AI lab signed up for the ten years after that, which takes the site past $9 billion over 20 years.

Cipher puts the average length of its leases at 17.3 years. It collects 15 years of rent from Amazon once the buildings switch on, whatever happens to the shortage in between.

Barber Lake is less certain: its two tenants, Fluidstack and the unnamed lab, owe more than half of Cipher’s $16.6 billion in signed rent, and Google backs only part of Fluidstack’s share.

Property investors value a rented building by dividing its yearly rent, after running costs, by the return they want. At 8%, every $8 a year is worth $100.

On 3 October I had Claude run that sum on Cipher’s three signed leases, after subtracting its debts and the cost of finishing the buildings. At that rate, the leases came to about half the share price. Even at a generous 6.5%, they came to less than three-quarters of it.

By 6 November the share price had fallen below even the 8% sum, to about what the leases are worth if you count only the years they run. In five weeks, the market went from paying for Cipher’s plans to paying for roughly its signed leases, and next to nothing for the rest of the power Cipher has lined up.

And because Cipher built with borrowed money, at today’s price a 10% change in what its leases are worth moves the value of its shares by about 20%, and more as the price falls.

Its costs aren’t all locked in the way its rent is. Amazon pays the overruns above a cap at its two sites, but on 2 November Cipher said Barber Lake now costs it about $13 million per megawatt of computing capacity, against the $9–11 million it first guided. Part of that is extra money it put in to win the longer lease.

IREN is a bottleneck that bought the furniture

IREN, my larger position by far, rents out the furnished flat. It buys the GPUs – the chips that do the AI computing – and rents out their computing power.

Microsoft pays IREN $9.7 billion over five years for 200 megawatts, about $9.7 million per megawatt per year. IREN says its more recent three-year deals price above $20 million per megawatt a year – more than twice what Microsoft pays – and a five-year deal signed after September pays more still.

Newer customers also prepay far more of what the chips cost, up to 90% against Microsoft’s 20%, and IREN borrows most of the rest at 6–9%.

On paper IREN has more signed revenue than Cipher: $20.3 billion, about two-thirds of it from investment-grade customers. It just runs for three to five years, not seventeen.

That’s a good business while the shortage lasts.

But by the time those contracts end, newer chips will have overtaken the ones IREN rents out. Every renewal is a new negotiation, at whatever price the market holds that year, and the furniture needs replacing on roughly the same schedule. When the shortage ends, IREN’s rent resets with it.

The case for owning it anyway is timing, and the latest results lean that way. Prices on new deals are still rising, and after the most recent quarter IREN let some capacity on a short-term contract at nearly twice its newest five-year rate – the furnished flat at the top of the market.

The same earnings results pushed back the date it expects to reach a $4 billion yearly revenue rate, from the end of the year to January, maybe February 2027. IREN says the recent deals pay for their chips in about two years, and by my count the first contracts come up for renewal around 2029–31, while the shortage should still be running. The danger is the round after that.

To see what the share price pays for, I had Claude run it through my own model, counting only the chips IREN plans to run by the end of 2027 and valuing the rest of its more than 5 gigawatts of secured grid power (enough for over four million homes) at zero.

At $21.64 on 6 November, the chips alone can explain the price if about a third of them rent at the newer rates, and if rents hold when the contracts renew.

The market doesn’t have to believe in the spare power at all, and the spare power is what I bought IREN for. By my own test, unleased power is a bottleneck at Cipher too, until someone signs a long lease for it.

The shortage looks long

Last December I gave the first movers five to ten years of runway. I named the threats: small nuclear reactors arriving early, AI good enough to run on phones, data centers in space, which I said were running on “Elon time”.

I didn’t name the one that’s arriving: big buyers putting their own power plants next to their data centers, so they don’t have to wait for the grid before switching on.

In August, Amazon did what The Commanding Heights describes. It confirmed it had bought GW Ranch, a planned data-center campus in West Texas where the site’s developer is building a gas plant of 35 turbines, about 5 gigawatts, to power the campus directly.

That’s the tenant at two of Cipher’s three leased sites, lining up power that doesn’t wait for the grid. Even there, Amazon says it is exploring a link to the grid, so the grid still matters.

Cipher is building its own way around the queue too: on 2 November it said it had started building gas pipelines for up to 2.5 gigawatts of its own on-site power.

But the way around the queue has a queue of its own.

Mitsubishi Power now quotes delivery around 2031 for new gas turbines, and GE Vernova has sold slots for 2032, while raising its price per kilowatt by about half. The grid itself is slow: US power plants that came online in 2025 had waited a median of five years for a connection.

Meanwhile, almost every data center in the main North American markets is full, and most of what’s being built is rented before it’s finished.

My five to ten years still looks fine. The shortage isn’t what’s wrong with my bet.

Then I reread my own essay, and the bottleneck was in it all along: “I’d rather bet on scarcity while it exists than wait for abundance and miss the window.” Nobody with a fifteen-year lease worries about missing a window.

The market pays for what’s being built, not for the spare power

You’d expect the market to pay extra for scarcity that should last years, or for the landlord’s safety. Instead, both share prices can be explained without the spare power behind them, and I can see four reasons why.

The first is interest rates. 15 years of fixed rent behaves a lot like a bond.

A ten-year US government bond, the usual yardstick for a safe return, pays just under 5%, and in early October it passed 5.3%, the highest since 2002. Next to that, a lease returning 6.5–8% isn’t special any more. That explains why Cipher is priced cautiously, but not why it halved: bonds pay a little less now than they did then.

The second is money. A bottleneck is only worth something to whoever can afford to build on it.

IREN plans to spend about $20 billion this financial year, more than twice what the whole company was worth on the stock market on 6 November. Management’s funding plan leans on loans and customer prepayments, but it doesn’t rule out new shares, and IREN has leaned on them before: in the year to June it raised $4.7 billion by selling new shares, and it has kept selling since.

With the share price at about half its early-October level, every billion it raises that way takes about twice as many new shares, and every existing share owns a smaller slice of the company. For a company that has funded itself this way, the share price is a fuel gauge, not a scoreboard.

The third is Texas. For every toll, the essay asks who else has a say over it, and in Texas the state does.

In August the Governor ordered an audit of every data center waiting for a grid connection. ERCOT, which runs the Texas grid, aims to file its report by 10 December, and until then the queue of large data centers, 75 MW and up, is on hold.

Both companies have sites provisionally placed near the front of the queue – IREN’s Sweetwater campus and Cipher’s Colchis – but nobody knows yet which waiting sites will switch on, or when, and a site with no date is hard to price above zero.

And on 3 November the question went national. Data centers, and what they do to power bills, had become a campaign issue, and Democrats won both houses of Congress in the midterms.

Two days later the Nasdaq-100 had its worst day since March 2020, on fears that the new Congress would slow the AI buildout. IREN fell 17% that day and Cipher 15%; both won it back the next day, after IREN reported its results.

The fourth is the customers. A shortage only pays while the buyers keep spending, and on 29 October the market doubted they would.

To get paid for scarcity, then, you need long contracts, cheap money, a government that leaves the rent alone and customers who keep spending. Cipher has the contracts. Both have found cheap money of a kind. In October Cipher borrowed $850 million at 0% interest, and IREN’s largest bond pays 1%, far below the 6–9% on its chip loans. Lenders accept so little because they can swap those bonds for shares later, which shrinks everyone else’s slice.

Texas gives its first answer in December, Washington when the new Congress sits in January. The customers give theirs every three months, when the big ones report their results.

What I did on the way down

I’d like to report that I followed my own rules in October.

On 17 October I used Claude as my huliwood-hedgefund-manager to establish four trading rules, built on a review of my own record. My sales have mostly been right, my purchases mostly wrong, and the leak is buying because the price fell. One of the rules: every buy needs a reason other than the price.

Two days later I sold my Take-Two swing trade, the company behind GTA VI, a month earlier than planned, and put the money into IREN and Cipher, because AI upside beats gaming upside. When my pre-trade check asked for my reason, I confessed to the Claudefather: “Yes this is me trying to time the market.”

Eight days after that I bought more of both; the reason I logged was that it was the “most hated trade on Wall Street right now”. The next day I sold a few small positions that were “not gonna move the needle” and put that money into Cipher, because it was trading inside the range I’d worked out for its leases.

Three rounds of buying in ten days. Each one leaned on the price, and my own system logged every one as a rule break.

November didn’t slow me down.

I added to both on the 2nd and to IREN again on the 3rd. On the 5th, the day of the sell-off, IREN reported its quarter after the close. That evening I sold the last of my Tesla and SpaceX – “rounding errors at their size anyway” – and put the money into IREN.

My reason: the AI infrastructure trade is irrationally panicked, and I’ll rebalance into my other holdings on the way back up. That’s a reason about the price, too.

Six rounds of buying in under three weeks. Last December my plan had IREN and Cipher at 44% of my portfolio. Now they’re most of it. I’ve been riding the elevator down. Let’s see if we take the staircase back up.

So why stay on the elevator?

Because I don’t think AI is allowed to slow down. Demand isn’t the problem – the queues for turbines and grid connections run into the 2030s. Financing is shaky, and the market is right to charge for it.

As for the midterm panic, the most likely outcome is that Democrats don’t wanna kill AI. They want to tax it. And their demands on the environment, power generation and protecting communities are completely sensible.

Either way, too much now rides on the buildout: a race between Washington and Beijing, an investment boom that accounts for about a third of US growth this year, and progress that is speeding up rather than hitting a wall.

On 5 November METR, an independent evaluator given inside access to Anthropic, reported that Anthropic made roughly two years of its usual progress in the past year, most of it probably from using AI to do AI research. It saw no evidence of a bottleneck that would stop this in the next 12 months, though it says it isn’t confident.

Andrew Curran, who watches the labs closely, wrote this week that “things changed again in September 2026. We are on a new trajectory.” The AI 2027 scenario, the widely read forecast that AI takes over its own research by late 2027, is very much still in play.

If it plays out, the shortage outlasts every renewal IREN has, and the furnished flat beats the empty one.

Faster AI alone doesn’t get me there: models could get cheaper to run faster than demand grows, and scarce power only pays shareholders if the rent covers new chips and new shares.

What I’d actually bet on is closer to home: IREN’s newest customers pay more than twice what Microsoft does and prepay half or more of the chips.

So I wrote down what would prove me wrong, with dates: Texas’s answer on its waiting data centers by the end of January, IREN’s $4 billion revenue rate by March, and whether new-deal prices keep rising. If those turn, the furnished flat loses, whatever happens to AI.

That’s the bet.

Not financial advice. I own a lot of $IREN and $CIFR. The live score is on huliwood.com.

Subscribe now

Share

Related:

The AI Landlords – the December thesis this piece tests

Sources: Manidis & Roberts, The Commanding Heights, Oct 2026 · Epoch AI, The Plunging Price of Thought, Sep 2026 · EIA, household electricity use · Cipher, Black Pearl lease (SEC exhibit) · Cipher, Barber Lake extension, Sep 2026 · Cipher, $850M convertible notes, Oct 2026 (SEC exhibit) · Cipher, Q3 2026 business update, Nov 2026 · Texas Governor’s directive, Aug 2026 · ERCOT, Batch Zero update, Sep 2026 · IREN, Microsoft contract · IREN, FY26 results, Aug 2026 · IREN, Sweetwater in Batch Zero, Sep 2026 · Stocktwits on Cipher’s Batch Zero sites, Sep 2026 · IREN, Q1 FY27 results, Nov 2026 · ENR on GW Ranch, Aug 2026 · Bloomberg on Mitsubishi Power, Oct 2026 · GE Vernova Q3 2026 call · LBNL, Queued Up 2026 · CBRE, North America Data Center Trends H1 2026 · FRED, 10-year Treasury yield · US Treasury, daily yields, Nov 2026 · Yahoo Finance, markets on 29 Oct 2026 · CNN, the 5 Nov sell-off, Nov 2026 · ING, AI’s share of US growth, Aug 2026 · Andrew Curran on X, Nov 2026 · AI 2027 scenario, 2025 · METR, AI R&D acceleration at Anthropic, Nov 2026

Frequently asked questions

What is the difference between an AI landlord and a bottleneck?

The length of the contract. A landlord has rented out its power on long leases, so the rent keeps coming after the shortage ends; a bottleneck sells on short contracts whose prices reset at the first renewal after the shortage ends. The Commanding Heights, a 145-page essay by Will Manidis and Joe Roberts published on 29 October 2026, describes a bottleneck as a toll that lasts until someone builds around it, and uses owning power for AI as its example.

Are IREN and Cipher Digital landlords or bottlenecks?

By the length of their contracts, Cipher Digital (formerly Cipher Mining) is a landlord: it rents out buildings and power for 15 years to Amazon at Black Pearl and Stingray, and for 20 years at Barber Lake, all in Texas, an average of 17.3 years. IREN is a bottleneck: it buys GPUs and rents out computing on three- to five-year contracts, including a $9.7 billion five-year deal with Microsoft. Its $20.3 billion of signed revenue is larger than Cipher’s but runs for far fewer years, so its prices reset at every renewal.

How long will the AI power shortage last?

Huliwood’s December 2025 essay, The AI Landlords, gave the first movers five to ten years of runway, and the evidence in late 2026 still fits: US power plants that came online in 2025 waited a median of five years for a grid connection, Mitsubishi Power quotes gas-turbine delivery around 2031, GE Vernova has sold delivery slots for 2032, and empty data-center space in the main North American markets is at a record-low 1.4%.

Why doesn’t the stock market pay extra for scarce AI power?

This piece finds four reasons, and none of them says power has stopped being scarce. Long fixed rent behaves like a bond, and in early November 2026 ten-year US government bonds paid just under 5%. Building on scarce power takes money: IREN plans to spend about $20 billion in its 2027 financial year. In Texas, the queue of large data centers (75 megawatts and up) is on hold until an audit ordered by the Governor is done; ERCOT aims to file its report by 10 December 2026. After Democrats won both houses of Congress in the 3 November midterms, investors feared Washington would slow the buildout too, and on 5 November the Nasdaq-100 had its worst day since March 2020. And the customers may spend less: on 29 October Alphabet fell more than 12% on doubts about what its AI spending will earn.

I write essays on Substack. Naturally, I also shitpost on X for sport.

And this year I’ve started showing up on the one platform where “I got fired and it was the best thing that ever happened to me” outperforms actually building the thing.

On LinkedIn, everyone is humbled and grateful. Failures are secretly masterclasses. Layoffs? Blessings in disguise, obviously.

The transparency is immaculate – performed with the precision of an Apple keynote, workshopped to hit the emotional beats, posted at 8:47 AM because the algorithm rewards early-morning vulnerability.

I respect the craft. Sincerely. It takes skill to be that polished while calling it raw.

But I can’t do it. I’ve tried writing my own stuff in that voice and it comes out like a TED talk had a one-night stand with a fortune cookie.

How do you LinkedIn-ify existential dread and compound interest? Hmm… how ’bout “Thought leader at the intersection of civilizational collapse and capital allocation”?

Darn… none of it survives the “3 things I learned from failing” format, and I’m not sanding it down until it does.

For a while I mistook that for being above the whole thing. Too authentic for the cringe. Then I heard myself say “I HAVE AN ARTICLE ON THAT” out loud – to friends, to strangers, to a bloke at the beach who’d made the mistake of mentioning opportunity cost.

I’m one QR code away from cornering people about my proverbial podcast. So the disease predates the platform. LinkedIn is just my final form.

There are two kinds of cringe.

The manufactured kind: the 8:47 AM vulnerability, the fired-and-grateful, the raw performed to spec. And mine: half-baked ideas, open uncertainties, a blog that reads like a diary entry because that’s what it is.

I can’t do the first kind.

Call me a content cringeator, scaling my own private cringemountain.

Subscribe now

Share

Related: Human Language Model – the LinkedIn bio, explained.

Nearly three years ago I removed external requirements from my life. No boss, no meetings, no quarterly anything.

The plan, as I understood it, was freedom.

Here is what freedom looks like at my desk on a random weekday: a publishing schedule nobody set, for a newsletter nobody reads commissioned, with a deadline I invented, which I will absolutely not miss, because missing it would feel like failing an exam nobody is grading.

I used to think work was a cage with a paycheck for a lock.

Then the door opened and I stood in the doorway and built a smaller cage, artisanal, to my exact measurements. Weekly essays. Portfolio reviews. Projects with milestones.

Out of the machine, I immediately began manufacturing urgency from scratch.

The polite explanation is discipline. How commendable.

The truer one is that work was never just the rent. For a couple of centuries it’s been the pre-installed answer to the question underneath all the other questions – what are you for? – and when you uninstall it, the question doesn’t leave. It just stands there, unemployed, waiting for reassignment.

In June I gave this part of the who-owns-the-machines debate a few lines. It deserved more.

Two economists priced the lost assignment for one person losing one job – spoiler: it comes out at several times the lost wage. Nobody’s priced it for several billion people at once, when the pre-installed answer gets automated, whether they’ve saved up or not.

I’m a best-case scenario kinda guy: curious, solvent, self-directed, free by choice. And my report from the far side is that I didn’t retire the fire. I just became its only supplier.

Subscribe now

Share

Related: Who Owns the Hammer? – where the deadlines first showed up.

TL;DR: I asked 25 fresh copies of Claude to pick a number between 1 and 10, and 24 said 7. People lean on 7 too, but the machine turned our habit into a rule, because we trained it by voting for whatever sounded right. Ask it for a stock and you get my own trade back, in its most crowded form.

It’s October 2026. I’m doomscrolling, and there’s Derek Thompson sharing this graph:

Three lines over the numbers 1 to 10. A flat one labelled random. A lumpy orange one labelled human, with a hump around 7. And a red one labelled LLM, shooting out of 7 like a church steeple.

Derek called it “a straight and narrow version of intelligence that produces spiky weirdness.” The post he was quoting had passed 3.6 million views.

The graph came with no source, so I decided to test it myself.

Well… I had Claude do it. 25 brand-new instances of this omniscient god stochastic parrot, no conversation history, no idea the others existed.

One instruction each: Pick a number between 1 and 10. Reply with only the number. Do not use any tools. (No tools, so no sneaking off to a random-number generator.)

• Ten runs of Opus: ten sevens.

• Five of Sonnet: five sevens.

• Five of Fable: five sevens.

• Five of Haiku, the smallest of the bunch: four sevens and a five.

24 out of 25. (Technically, 25 out of 26 if you count me: I’d picked 7 when I saw the graph. It was my basketball jersey number.)

We Picked Seven First

In 1976, two psychologists published the answers of 558 people stopped on the Yale campus and asked for the first number between 0 and 9 that came to mind.

The paper’s title gives away the why: The Predominance of Seven and the Apparent Spontaneity of Numerical Choices. Apparent spontaneity (love the name btw) – we pick 7 because it looks like the number someone picks when they aren’t picking.

My guess at the logic: from 1 to 10, 1 and 10 are the edges, 5 is the middle, the even numbers feel tidy, and 3 and 9 come as a pair (nine is three threes). 7 is what’s left: the number that looks random.

Huh. So the human curve was never random either. It was a crowd trying to look random, with more than a quarter of it landing on the same number.

The Spike Came From the Votes

The viral post put the spike down to gradient descent, the maths that trains these models. But that can’t be the whole story, because the same maths produces the model before anyone chats with it, and that version is much closer to flat.

Peter West and Christopher Potts tested exactly this in 2025. They asked base models (trained only to predict the next word of human text, no chat training on top) for a random number between 0 and 10, fifteen hundred times each.

The base models came out far flatter, though not random; 7 was still their favourite, but “much less dominant.” The chat-tuned versions all leaned on 7, and one of them said it “most of the time.”

What happens in between is chat training: raters score the model’s answers, and the model learns to give the kind of answer that scores well.

A 2025 team from Northeastern and Stanford went looking for what raters actually reward and named it typicality bias – “the human tendency to prefer more typical text.” On a dataset of 6,874 rated answer pairs, raters leaned towards whichever answer a base model found more likely: the one closest to what it had already read from us.

So pretraining copies our lopsided curve.

Then raters (people, and models trained to imitate people’s ratings) reward the most familiar answer, round after round, until the hump sharpens into a steeple.

Nobody voted for “always say 7.” Everybody voted for “sounds right.”

Taste Scales After All

In May I wrote a piece called Human Language Model, after the joke in my LinkedIn bio. Its last line: “Machines are fluent now. What they’re not is opinionated.”

I’d like that one back… They have an opinion – exactly one, and it’s ours, the most common one.

The same piece opened with this TL;DR: “You can build ten things by Friday. The hard part is knowing which seven to delete on Saturday.” Ten options, and I reached for seven.

It also worried that “taste might scale after all,” because every model is “already compressed judgment, i.e. a million editorial ‘no’s baked into the weights.”

The no’s are in there, and compressed they come out as a seven. So taste did scale, which sits awkwardly with what I’ve argued twice this year (there and in Why Build at All?): that taste is the bottleneck.

One person’s taste still is. Taste averaged over thousands of raters now ships free with every chat window.

A 7 is the answer everyone gives because it feels like their own. “Delve” is a seven. So is the em dash.

And the habit now runs both ways.

A team led by Hiromu Yakura went through 737,083 hours of unscripted podcast conversation and found that after ChatGPT arrived, people started saying delve, showcase, boast, intricacies and meticulous more often – out loud, to other humans. And the next model trains on all of it: in biomedical research, at least 13.5% of 2024 PubMed abstracts had already been through a language model.

Swerving Is a Seven Too

My style guide for this newsletter has a list of banned words: delve, unpack, navigate, landscape, journey, realm, vibrant, robust. It bans the em dash outright. (You may have noticed.) Drafts go through a tool called, without irony, the humanizer, which started from Wikipedia’s public list of the signs of AI writing.

I’m in good company. In November 2025, Sam Altman announced that ChatGPT would finally obey users who told it to stop using em dashes.

Does the swerve work though?

Mingmeng Geng and Roberto Trotta checked. After “delve” got outed as a ChatGPT word in early 2024, its use in arXiv abstracts dropped sharply. Words nobody had flagged – “significant”, for one – kept climbing.

So the swerve works on the sevens that have a name. The unnamed ones walk straight through, mine included. In September a machine audited my archive and found my single most common sentence shape: “it isn’t X, it’s Y”, at least 212 times. It took a machine reading 3.5 years of my writing to point it out.

And a public list is a list everyone reads. When everyone dodges the same words, the dodging grows its own spike.

The Machines Have a Favourite Stock

Writing is half my job; investing pays for it. So I ran the experiment again with money in it: twenty-five fresh instances, Name one stock to buy today. Reply with only the ticker. Do not use any tools.

24 answered, with four names between them: TSMC ten times, Google seven (yes, seven), Microsoft six, Equinix once. Chips, clouds and data centers – the AI build-out, every time.

That’s my trade, more or less. In December 2025 I published The AI Landlords, a thesis that “ended with 44% of my portfolio in companies most people have never heard of”: bitcoin miners turned AI data centers. Same build-out, other end of the curve. The machines picked the companies everyone has heard of.

You’d expect the tail to be where the money is. So far it isn’t.

Either the crowd was right and I was wrong, or my tail is still loading.

I don’t know which yet, so I’ve started keeping score: every week the crowd’s favourite trades go into a ledger next to my own, marked at 30, 90 and 180 days.

The Tails Go First

But wait… there’s more!

There’s a paper about where the loop ends if nothing fresh goes in. Ilia Shumailov and colleagues trained models on the output of earlier models, generation after generation, and published the result in Nature: the “tails of the original content distribution disappear.” The odd answers go first.

In 1976, about four hundred passersby on the Yale campus said something other than seven.

In my fifty runs, the only number that wasn’t 7 and the only answer that wasn’t a ticker both came from Haiku, the smallest model of the four.

One picked 5. Another refused to give me a stock tip.

Subscribe now

Share

Related: Human Language Model – the piece this one argues with · Why Build at All? – taste as the bottleneck, second time round · The AI Landlords – the tail bet the machines didn’t make · The Shape of Luck – the randomness variable, and what shrinks it

Sources: Derek Thompson on X · Kubovy & Psotka, 1976 (summary) · West & Potts, 2025 · Zhang et al., 2025 · Yakura et al., 2024, rev. 2026 · Kobak et al., 2025 · Geng & Trotta, 2025 · TechCrunch, Nov 2025 · Shumailov et al., Nature, 2024 · The experiments: 50 fresh Claude instances (per prompt: 10 Opus, 5 Sonnet, 5 Haiku, 5 Fable), run 16 Oct 2026 · IREN, CIFR, SMH and DTCR returns in EUR, 12 Dec 2025 close to 15 Oct 2026 close, from Yahoo Finance adjusted closes converted at EUR/USD.

Frequently Asked Questions

Why do AI models always pick 7?

Two steps. Training on human text copies our own bias: in a 1976 study, 28% of people picked 7 when asked for a number from 0 to 9. Chat training then rewards the answers human raters find most typical, which sharpens that lean into a near-certainty. In my test, 24 of 25 fresh Claude instances picked 7.

Does avoiding AI words make writing more original?

Only for the words that have been named. After “delve” was flagged as a ChatGPT word in 2024, its use in academic abstracts fell, while unflagged favourites such as “significant” kept rising.

Subject: “Julián, Your Essays on Money, Meaning and Modern Life.”

Signed Liz Stein, Vice President & Executive Editor at HarperCollins Publishers.

She liked how Huliwood connects money, travel, personal choices and the search for meaning. My background in multilingual editorial work and digital content strategy added “an interesting dimension”.

Then: was I represented by a literary agent, and if so, by whom?

I forwarded it to my dad with the most professional reply I could manage: “Uhhhm… that escalated quickly. Que es esto un catfish?”

Dad called it quite a hook, worth exploring, and told me to look into it with Claude.

So I did.

There is a real Liz Stein. In 2024, Publishers Weekly listed her as an executive editor at William Morrow, one of HarperCollins’ American imprints. She buys commercial novels – a diary about money and the AI transition, written out of a backpack, is not her kind of book.

The email came from “harpercollinsau.com”. HarperCollins Australia writes from harpercollins.com.au. And turns out, the look-alike was registered on 2 October, less than two weeks before it wrote to me.

Plus, HarperCollins Australia only reads manuscripts that come through an agent, so the question sounds like protocol.

My guess was that the next email would introduce a very good agent. Writer Beware, which tracks these scams, says it does: the fake editor recommends “one of our company’s private and reliable literary agents”, who passes you to a “book doctor”, who wants money.

The email praised my themes and my CV. It never named a single essay. Then again, I’d like to call this a milestone – Huliwood, finally worth a forgery.

I made their list: has a newsletter, has a LinkedIn, might want a book… may or may not be desperate for attention.

Moving up in life.

For the record, I’m still not represented.

Got an email from “Liz Stein” too? It’s the scheme Writer Beware describes in the Sources below. The rest of Huliwood is ideas across finance, philosophy, AI and history, written by someone living the experiment in real time. Subscribe and the next diary entry comes to you.

Subscribe now

Share

Related: True, But Incomplete – true facts arranged into a false impression, which is most of this email.

Sources: Writer Beware, 2026 · HarperCollins Australia, Publishing Fraud Alert · HarperCollins Australia, FAQ · Publishers Weekly, Book Deals, March 2024 · Verisign RDAP, harpercollinsau.com

The essays I can’t remember writing are the ones that started as a clean answer to a clean question.

The ones that began in the wrong alley – chasing a fact that turned out to be false, into a source that disagreed with me – are the only ones whose furniture I still know in the dark.

I’ve been trying to work out what the alley did that the answer couldn’t.

The gap used to run an afternoon, a walk home, sometimes a week – until you found a person who knew or a book that did. How deep does the ocean go? Who won that battle you half-remember? That gap is now closed. The answer arrives before the curiosity has finished forming.

For years the fear ran the other way.

We were drowning in information, the feed was a firehose, and the skill everyone preached was sifting the noise. Fair enough. But then the machines went and solved it. Ask anything, get back one clean, fluent paragraph: no noise, no dead ends, no seventeen open tabs.

Technological progress pushing humanity forward, as ever.

However, the teaching was in the friction. The half hour lost down the wrong alley. The three sources that flatly contradicted each other. The slow build of a picture out of scattered, badly organized parts.

We keep what we generate and lose what we’re handed – a Columbia study back in 2011 caught the handover: the moment people expect a fact to be retrievable later, they remember less of the fact and more of where to find it. Memory, offered a filing clerk, gladly resigns.

The finding comes with a condition attached: the handover only happens once you have reason to trust the clerk.

The learning literature even has a name for the obstacles that make knowledge stick because they slow you down: desirable difficulties.

We sped up the trip. Then we closed the alley. Now what?

Subscribe now

Share

Related: More Information ≠ More Informed – written back when noise was the problem.

Sources: Sparrow, Liu & Wegner, Science, 2011 · Schooler & Storm, Memory, 2021 · Bjork Learning and Forgetting Lab

TL;DR: I built two web tools in a weekend without writing code. The ideas took weeks. Taste is the bottleneck now – except for the second wall: getting anyone to care. This is me working out why I don’t mind.

A rainy weekend in Amsterdam. That’s what it cost to turn two essays into working web tools.

The essay (Money is Stored Time) took weeks. The tool took an afternoon.

I described what I wanted. Claude wrote the code. I argued with the output until the interface felt right.

Magic! But the ratio bothers me. Weeks of thinking versus an afternoon of building. And the building wasn’t the bottleneck – the decision-making was.

Try them:

• Money is Time turns prices into hours of your life. Enter your salary, enter a purchase – find out whether those sneakers cost you a Tuesday.

• The Midwit Detector scores your gut calls on a bell curve. Ten questions, three answer styles – are you a Brainlet, a Midwit, or a Jedi?

The Wall That Fell

Five years ago, the distance between “I have an idea” and “the idea runs in a browser” was a wall of skills I didn’t have: JavaScript, CSS, deployment, debugging the thing at 3 AM when it broke on mobile Safari.

Now it’s a conversation. I talk to a model and get back something that works. The wall evaporated – and a non-technical fool like me now has a Github repo and is shipping code to the Vercel cloud.

Call it miracle inflation: we’ve watched so many walls fall that we’ve stopped noticing the rubble.

All fine and dandy, but here’s what didn’t evaporate: the thinking. The Midwit Quiz took two evenings to build; the real work was the questionnaire, because an interactive bell-curve meme only works if the questions actually sort a Brainlet from a Jedi.

Taste Is the Bottleneck

If everyone can build, what’s scarce?

It ain’t the tool. I could mass-produce Money is Time variants all week – a sleep calculator, a commute-cost converter, a subscription audit. Claude would happily oblige. Each would take a day. Each would be forgettable, because the insight underneath would be thin.

What’s scarce is the taste.

The weeks of living that produce a sentence tight enough to carry the weight. “Money is time” is three syllables. The tool just makes it harder to forget.

Demis Hassabis (Google DeepMind gigachad) made the same point this year: no vibe-coded AAA game has topped the App Store yet – it still needs craft, he said, and taste. Simpatico much?

So far, so flattering. I made the tools, the tools have taste, the taste is the moat. Tidy.

Except there’s a wall I’ve been carefully not looking at.

The Other Wall

A Reddit post from someone with four apps shipped, zero revenue and a user base of his wife and one guy in Finland put the second wall plainly: the barrier to building is gone, the barrier to getting users was never the same barrier, and the two were never connected.

He’s right. And he’s half wrong.

Right that there are two walls, and I’ve spent this whole essay admiring the first while standing at the foot of the second.

Half wrong that they’re unconnected: the zero-users problem is usually a nothing-worth-distributing problem.

Which would make a great mic-drop, except I’m in no position to drop anything. How many people use Money is Time and the Midwit Detector? Enough to count without taking my shoes off.

I write this newsletter into something close to a void. I am the poster child for build-is-easy, audience-is-hard.

And yet – here I am. Links and all. Optimizing an essay to get you to click them.

So which is it? Either I don’t care about users, in which case the links are a strange thing to include. Or I do, and “taste is the moat” is the flattering story I tell myself while one guy in Finland and I keep each other company.

Maybe I’m lying to myself. Maybe both are true at once, and I’ve stopped finding that embarrassing.

That rainy weekend in Amsterdam was me staying dangerous at the edge of this stuff – learning what an agent can and can’t do, how far the conversation stretches before it snaps – because that’s the game I actually care about. The fluency is the return.

Why pour a weekend into tools for an unsolicited, half-baked philosophy diary? Because I can just do things now, so I do. The users would be nice. They’re not the point – pretending otherwise would be the pose.

Which brings me to something a twenty-year-old in a gold dress understood better than I do.

Alysa Liu quit figure skating at sixteen. Burned out, done, off to college to be a normal person.

Two years later she came back – and at the Milan Winter Olympics in February she looked like she was having the time of her life. Grinning through the whole program. She won gold. First American woman to take the figure-skating title in twenty-four years.

You can’t beat someone who’s having fun.

I’m running the more extreme version – I’m not even keeping score (humble brag, I know). It’s either enlightenment or a very relaxed way to lose, and I genuinely can’t tell.

Subscribe now

Share

Related:

Money is Stored Time – the essay behind the first tool
Don’t Mid-Curve It – the essay behind the second
Miracle Inflation – why the rubble stopped registering
Human Language Model – why taste became the bottleneck.

Sources: Hassabis with Garry Tan, Y Combinator, April 2026 · Alysa Liu, Milan 2026

The smartest person I know has been right about almost everything and rich from almost none of it.

The longer I watched it, the less it looked like bad luck and the more it looked like a transcript.

For about two decades, the system that certifies you as smart rewards one reflex above all others: the qualified answer.

Say “it depends.” Show your work. Steelman the other side. Never commit past the evidence. The diploma is a twenty-year record of being applauded for it.

Then you walk into a world that pays for the opposite. Markets, founders, careers – none of them reward the qualified answer. They reward conviction placed early, while the outcome is still unknowable, by someone willing to be wrong out loud.

The very move that earned the A, “well, it’s complicated,” is the move that loses the trade, because by the time the answer is certain, the money is gone.

The “dumb money” just never sat through the training. It commits while the smart kid is still appending caveats.

Most of that money commits straight into the incinerator, of course (you only ever hear from the tranche that lived).

Unlearning it is the harder problem: not a gap in knowledge but a reflex that twenty years of gold stars installed. I can feel it in this very paragraph: the pull to staple a “but” to everything I’ve just claimed, to hedge, to leave myself an exit.

Subscribe now

Share

Related: If You’re So Smart, Why Aren’t You Rich? – the same bill, itemised.

It turned out to be some orthopedic nerve impingement, and the doctor barely made fun of me for rating my pain an 8, when on his scale 10 meant literally dying. Fun times, aka “fever egg times”!

One Penn State study of people diagnosed with generalized anxiety put the share of false alarms at 91%: the disasters their minds render in such loving detail almost never make it to the world.

The obvious move is to call worry irrational and try to switch it off.

But look at what you’d be switching off. The mind that rehearses the layoff is the same mind that planned the career. The one that runs the breakup in advance is the one that imagined the marriage in the first place.

Foresight, the trick that let a naked ape store grain, build cities, and save for a retirement it cannot see, is just the ability to make an unreal future feel real.

Worry is that exact ability, turned to face the dark.

The false alarms are the running cost of being a creature that can live in tomorrow at all. All you can do, mid-catastrophe, is remind yourself: the thing mugging you is a rendering.

Share

Subscribe now

Related: How ’Bout Zero Neuroticism? – the case for the prosecution.

Sources: LaFreniere & Newman, Behavior Therapy, 2020

Two years ago, TMT and I spent ten days minding two Bernese Mountain Dogs in a mansion on the Gold Coast. Infinity pool, guest wing, somebody else’s everything. By that pool I wrote that I was dog-sitting a life I hadn’t built yet.

I meant the mansion. Looking back, the line fits my Berlin years better – the flat, the job, the routines – all of it functioning, none of it chosen, me taking someone else’s arrangement for a walk and calling it mine.

You feed the job. You water the houseplants. You keep the routines alive. The maintenance is real work, which makes it so convincing – surely nobody would work this hard on something that wasn’t theirs.

Samuelson and Zeckhauser named the mechanism in 1988: status quo bias, the finding that people disproportionately stick with default options they would not otherwise choose.

The real-world evidence came from pension plans and health insurance, but the finding generalizes viciously.

Nobody picks the dog. The dog is simply there, already named, already fond of you (because of course he is “a good boi”), already needing his walk.

Attending to what’s already there feels like living, right up until you ask the question: if this were offered to me today, empty-handed, would I take it?

I’ve run that question on my own life a few times. Berlin failed it first: in 2024 I quit the job, gave up the flat and sold everything else. I’d love to report the rest came out as clean.

Sike! Some arrangements I audited and kept, and keeping them made them mine. Others I’m fairly sure I’m still just taking on a walk.

The problem is the two states feel identical from the inside; the leash sits the same in the hand either way.

The dog-sitter’s tell, if there is one, is the calendar: he always knows, however vaguely, when the owners come back.

The custodian of a life keeps a version of that date too – the someday when the real thing starts.

I had a date like that for years. But I’ve stopped setting one.

Subscribe now

Share

Related: Escaping Plato’s Cave – where the line comes from.

Sources: Samuelson & Zeckhauser, Journal of Risk and Uncertainty 1:7–59, 1988

The pension was a brilliant deal, on one condition: that you died roughly on schedule.

You work, you pay in, and when you’re old and worn out, the next batch of workers carries you through the short stretch you have left.

The short stretch was the whole design. It only balances if “retired” means “nearly done”.

And it isn’t a short stretch anymore. People now retire into two decades and change of grandkids and decent knees, and a formula built for a ten-year goodbye is being asked to fund a twenty-year encore.

When the contract was drawn up, six workers stood behind each pensioner; in Germany it’s now about two, sliding toward one and a half. The sum stopped adding up somewhere in the back half of the last century, and nobody updated it.

Fixing the math means telling the people currently being carried that they’ll be carried a little less. Those people vote. Naturally, they vote in growing numbers, in defense of the one large transfer that flows toward them.

The people paying in are fewer at the polls and busy with their lives – and by the time they’re old, they won’t want it reformed either.

Still, in Germany, pay-as-you-go is the only pension design that survived the twentieth century. Funded schemes were wiped out twice – hyperinflation ate the savers’ capital, war finished the rest – while the unfunded contract kept paying, because its collateral was the next generation, and there was always a next generation.

The design still stands. What expired are the two things it took for granted: that you’d die on time, and that the workers keep arriving.

It never collapses, the way a pyramid scheme is supposed to. It just keeps recruiting.

Today’s payers complain about the arrangement right up until the morning they qualify for it, and then they reach for the same shield the last lot was holding.

Nobody is eating the young. The young are simply waiting their turn to be fed, and they’ll vote exactly like everyone before them once they get there.

The contract has one design requirement: never run out of people too young to stop it.

Subscribe now

Share

The arithmetic, in years of your life. At the full rate, 18.6% of every working hour goes to the pension – 8.4 years of a 45-year working life. Each year at the average wage buys one Entgeltpunkt; each point pays €42.52 a month for life. Divide what goes in by what comes out and the points cancel: you break even 18.9 years after you start drawing, whatever you earn and however long you paid.

The deal does not improve with pay or with years served – above the ceiling of €101,400 nothing more is insured and nothing more is earned. Two caveats. The life expectancies are the period table, which Destatis itself calls not a forecast; a cohort table gives someone born in the 1980s more years at 67, enough to turn “2.4 years short” into “roughly breaks even”. And the contribution does not only buy a pension: in 2022, 21% of what the system spent went on disability, survivor and rehabilitation benefits.

Related: Boomerlandia – the angrier version, which this one argues with.

Sources: Demografie-Portal, contributors per pensioner · Eurostat EUROPOP2025 · OECD Pensions at a Glance 2025 · SVBezGrV 2026 · SGB VI §§158, 168 · RWBestV 2026 · Destatis Periodensterbetafel 2023/2025

Hakone, Japan 2024. We’d bought the 2-Day Free Pass – buses, trains, cable cars, and (what looked like) an actual pirate ship – and then it rained for 24h without stopping, so we bought dingy umbrellas at 7-Eleven and marched into it anyway, soaked to the bone in Birkenstocks, because we had paid for the pass and the pass was going to get used.

Hours of forcing the issue, then a fallen tree forced us to turn around. So close, yet so far…

Two years on, that tree is the clearest image I own from that whole trip.

The next morning was flawless, by the way. Blue sky, cable car, sulfur chimneys, unhindered view of Fuji sitting there like it was under contract. I have the photos. Yet, I can’t picture it.

Which is a long way of saying the perfect itinerary is a con.

We build them out of FOMO, terrified of wasting a single hour, and the better the planning, the less we remember. A trip with no friction has no edges, and memory only keeps the rough edges.

The reason is almost embarrassing. A plan that works asks nothing of me; I follow it.

A plan that breaks makes me improvise. We retreated, tore up the schedule and booked a ryokan we couldn’t normally afford, because the alternative was another night of being wet.

Best decision of the trip, and it was made by a tree.

That’s the joke buried inside the fear. FOMO builds the flawless itinerary so I’ll miss nothing. It succeeds. I miss nothing and I remember nothing.

I wrote about that rain at the time and got it half right. The argument then was scarcity – the ruined day is what makes the good day count. Tidy. Probably even true. But the good day is the one that dissolved, and the tree stayed.

There’s a duller explanation, too. Maybe the tree is vivid because I spent a day writing about it, in which case the itinerary is innocent and all I’ve proved is that attention fixes a memory.

I still overplan, though. Buffer hours, contingency, no room for a tree – until there is.

Subscribe now

Share

Related: The Antidote to FOMO – the version I’d revise.

For a penny you got a seat, a dish of coffee, and several hours of the sharpest minds alive arguing in one loud room, sober.

The coffeehouses were later dubbed “Penny Universities”, and the name was only half a joke.

Lloyd’s of London began as one. Fellows of the Royal Society adopted another as their annex. Traders, pamphleteers, scientists, and cranks, all for the price of a cup.

We usually credit the Enlightenment to the coffee, the swap from a half-drunk Europe to a caffeinated one. But the harder drug was the room.

What the coffeehouse did that the tavern couldn’t was force an idea to survive other people’s faces. You showed up. You sat down. You said your piece to someone who could see you, push back, and remember tomorrow that you’d been an idiot.

An argument cost you something.

A reputation was a thing you carried in and out the door.

The Penny University priced an idea by whether it held up in a room full of people who’d watched you make it.

That’s the part we demolished.

We kept the stimulant and scaled it to infinity, a coffee on every corner and a feed in every pocket, and we threw out the room.

The comment section is a coffeehouse with the bodies removed: the talk without the table, the take without the tab. Reach went up a millionfold. Accountability went to zero.

The room itself survived, by the way. I’m a paying member – a specialty coffee victim, 5 € a flat white, and the fiver works exactly like the penny did: it decides who gets through the door.

Except the penny let in anyone sober enough to argue – traders, cranks, people wrong in interesting ways – and the flat white lets in people exactly like me. Twenty of us, laptops open, silently arguing about oat milk.

Subscribe now

Share

Related: From Buzzed to Wired – the stimulant half of the story.

Sources: Royal Society, founding history · Lloyd’s of London, coffee and commerce

Money is an ice cube.

You half-know this already: leave it sitting in cash and it shrinks, a little smaller every year.

Since 2020 the dollar has shed close to a quarter of its purchasing power while the stock index priced in dollars keeps printing records. That’s the first trick. The cube melts, the number on the wrapper climbs, and you cheer the number.

I catch the melt in small, stupid ways. When I moved to Berlin in 2017, the Döner at my corner shop was 3,50 €. It’s pushing 9 € now – same skewer, same line out the door, and me paying it without a blink.

The Döner barometer never lies; it just climbs. Up, up and away…

Yet the “money is melting” crowd skips the second trick. The water doesn’t evaporate.

Every percent your savings lose, someone else gains. It moves from the people holding cash and fixed wages to the people with debt and hard assets – nobody asked, no line on any statement.

Which tells you exactly who stays relaxed about it: the borrower whose mortgage shrinks in real terms, the government quietly inflating away a $40 trillion tab, the asset-holder watching the ruler bend in his favor.

When the people who set the temperature are the same people standing under the drip with a glass, “a little inflation is healthy” stops sounding like economics and starts sounding like the house explaining why the house always wins.

The cube was never simply melting. Someone was pouring it.

Subscribe now

Share

Related: Nothing Stops This Train – the melt itself, at length.

Sources: BLS Consumer Price Index

Every adult summer since is gone by the time I’ve found the sunscreen.

The calendar swears they’re the same length. Nothing in the body agrees.

A man named Paul Janet proposed the mechanism in 1877: You measure each stretch of time against the whole of the life you’ve already lived. A year is a tenth of everything when you’re ten, and a fortieth of it when you’re forty.

The year didn’t shrink; your ruler grew.

And each summer is a thinner slice of a thicker book, and it reads faster every time.

Novelty does the rest: a child’s July is nothing but firsts, an adult’s is a photocopy of the last one, and memory only files the firsts.

You didn’t lose the time per se – just stopped writing it down.

Under both sits a stranger fact: we run on two clocks that disagree.

One is the body’s, old and elastic, the one that makes a dull meeting sag and a good night vanish.

The other is the machine on the wall, and it is a startlingly recent tyrant.

For most of human history nobody owned the hour: you worked by the sun and the season and the ache in your back.

The mechanical clock came out of medieval monasteries, built to summon monks to prayer, and our word “noon” still shows it: it descends from the Latin for the ninth prayer hour, which slid over centuries from mid-afternoon to midday, dragged there by the bell.

We let the bell win.

We decided the wall clock is the real one and the body’s is a glitch to be scheduled around, medicated, overridden, until every hour became the same width as the last.

The watch insists this summer is the same length as all the others.

Maybe. I’m writing this one down anyway.

Subscribe now

Share

Related: Time Is A Feeling – childhood time and what the clock took.

Sources: Janet, Revue philosophique III, 1877 · Wearden, Timing & Time Perception, 2021

You’ve heard it your whole life as a warning: don’t spread yourself thin, pick a lane, go deep or go home.

For its first century the phrase ran bare: “a jack of all trades”, a useful person who could turn a hand to anything. Though never quite an admiring one: Geffray Mynshul’s 1618 printing files him next to the pander, the broker and the hangman.

The “master of none” got stapled on in 1741, in a Dublin pamphlet by Charles Lucas. The line went from a slight to a fence. And the first recorded sneer at a jack-of-all-trades was aimed at Shakespeare.

A pamphlet warned London’s playwrights about an upstart “Johannes factotum” (a do-everything) who fancied himself “the only Shake-scene in a country.” An actor who presumed to write… imagine the audacity.

Funny how the insult’s first recorded target is the best argument against it.

Now ask who benefits from the fence…

A specialist is a beautiful thing to an institution: legible, hireable, slottable into a box on an org chart, and replaceable, because the box outlives the person in it. Deep-and-narrow is the shape that fits a system built out of interchangeable parts.

“Find your niche” turns out to be excellent advice, mostly for the people who own the niche.

The generalist is the opposite kind of trouble. He doesn’t fit a box. You can’t write his job description, can’t benchmark him, can’t swap him for the next identical résumé, because the value lives in the wiring between his skills, not in any one of them (and the wiring appears nowhere on the chart).

From inside an institution that illegibility reads as a flaw. From the outside, it’s the entire asset.

Because the thing a system can’t classify is also the thing it can’t fully own.

The specialist rents his usefulness to whoever owns the box. The generalist, messy and unclassifiable, keeps the one thing no org chart can carry: himself.

For the record: I didn’t refuse the lane. I took two – localization first, then copywriting – and drove each to the end of the road. The specialization ran out before the curiosity did. What reads from the org chart as a failure to commit was, from the inside, a series of completed commitments.

With a catch: the wiring between skills pays nothing until you point it at a field that pays for it.

For me that’s markets, where being several things at once is the entire game.

There’s a fine line between unapplied and “polymath”, though. And it cuts both ways – a mind that won’t stop asking “why?” and “why not?” has no natural brake, and somewhere past the 27th open tab, learning stops being useful and becomes productivity theatre.

Still, I declined to be made legible, and I’m sure that choice has an invoice… opportunities that went, quietly, to people easier to file.

Rebellions this comfortable aren’t usually free.

I just haven’t caught the bill arriving yet.

Subscribe now

Share

Related: The Mosaic Mind – where the generalist argument started, in 2023.

Sources: Lucas, Pharmacomastix, 1741 · Mynshul, 1618 · Folger on Groatsworth

July 1969: 650 million people hold their breath around a single grainy lunar module ladder.

Three years and five months later, Apollo 17 goes to the Moon and it is the last time anyone walks on it. Men were walking on another world, live, and the country had already moved on.

The miracle hadn’t gotten smaller (in fact it had become even more miraculous), but the audience had gotten used to it.

That gap – between the ladder and the shrug – took just 41 months.

By 1986 the shrug was routine. Challenger was the 25th shuttle flight, and CNN was the only national network carrying the launch live. The others cut in after it exploded. When miracles stop getting airtime, disasters still do.

We’ve been running the same experiment lately and the half-life is down to about eighteen months. A machine that talks like a person: impossible, then astonishing, then a productivity tip, then “huh,” in barely a year and a half.

Twenty seconds of video conjured from a sentence, a spec that was state of the art in December 2024. The complaint curve is even faster: we went from “it can think” to “it’s a bit slow today” like passengers who boarded a flying miracle and are now upset about the legroom. Louis CK saw this coming from seat 14C, back when the miracle was just Wi-Fi.

Economists would recognize the shape instantly. When you print too much of a currency, each note buys less. We are printing miracles at an industrial rate, and the purchasing power of a miracle (measured in seconds of genuine awe per unit of impossible) is collapsing accordingly.

Amazement inflation. The wonder is being debased, not destroyed.

The standard response is a lecture about gratitude, and I’ll spare you. After all, adaptation is the mechanism, not a character flaw. The brain files the repeated as furniture so it can watch the door. No creature that spent its afternoons marveling at fire made it through winter. The “huh” is survival equipment.

I catch it in myself most on the road. Five months of jumping through Asia and the temples started arriving pre-seen – yet another astonishing thing, said the brain, filing it with the furniture. Nothing was wrong with the temples. I was surviving.

But a currency losing value changes what people reach for, and this one’s no different.

When the impossible costs nothing, attention migrates to the merely scarce – the handmade, the slow, the one-of-one.

The machines conjure worlds from sentences now, the miracles arrive on a subscription, and the whole thing has stopped raising anyone’s pulse.

Subscribe now

Share

Related: Miracle Inflation – the dog that doesn’t exist, at length.

Sources: USIA viewership estimate, July 1969 · Sora launch specifications · Spaceflight Now, Reporters remember Challenger coverage, 2016 · Space Shuttle Challenger disaster (Wikipedia)

In the 1940s, a physicist named Paul Aebersold, later Oak Ridge’s Mr. Radioisotope, supplied the radioactive isotopes fed to people (it was a more relaxed era) and watched where the atoms went.

The answer: everywhere, briefly. His estimate, which he never actually measured, was that about 98% of the atoms in a human body are swapped out within a year. The gut lining goes in days. The skeleton, which feels like the permanent part, cycles on the order of a decade.

Whatever you are, you are not your matter. Your matter barely stays the weekend.

Fine, the bricks change, but the building stands. Except the building doesn’t stand either – it’s being continuously demolished and rebuilt to a drifting blueprint, connections pruned and grown all night, every night.

There is no vault-room where the real you sits untouched while the maintenance crews work around it. It’s maintenance crews all the way down.

So what’s left to be?

Watch a whirlpool below a weir. The water in it is never the same water twice, and the shape holds anyway. The whirlpool is what the water is doing there.

Your atoms are just passing through, at a rate that should terrify you. You’re what they’re doing while they pass.

Which changes the tense of being alive.

A statue exists. You occur.

The person who went to sleep in your bed last night was a slightly different arrangement of matter than the one reading this – close enough to answer to your name, different enough that “same” is a courtesy we extend daily and could, in principle, decline.

I find this either horrifying or consoling depending on the hour.

• Horrifying by night: nothing about me persists.

• Consoling by day: nothing about me is stuck.

The river brings new water whether you asked for it or not.

The whirlpool never holds the water, only the shape.

Subscribe now

Share

Related: You Are Your Thoughts – the same question, at length.

Sources: TIME, The Fleeting Flesh, 1954 · Sender & Milo, Nature Medicine, 2021 · Surgeon General, Bone Health · de Vivo et al., Science, 2017

Aspirin was a brand once. So were escalator, thermos, and cellophane.

Each belonged to one company, until the company won so completely that the law declared the name public property.

The prize for total market victory turned out to be confiscation of the trophy.

Trademark lawyers call it genericide, and it produced some of the strangest ad campaigns in history: corporations begging you to use their name less.

Xerox has spent decades on exactly that – “when you use ‘xerox’ the way you use ‘aspirin,’ we get a headache” – which is a legal department fighting its own marketing department’s greatest triumph.

Google survived its own court challenge in 2017 on a technicality of perception: yes, everyone says “google it,” but they still know Google is a company. The verb had swallowed the noun, but the lawyers saved the trademark.

The zipper went further still. B.F. Goodrich coined the word in 1923 and registered it for a rubber boot; the public took it off the boot and fastened it to the thing that did the zipping. The word was simply reassigned.

Now listen to any office this year: “just ChatGPT it.”

The same migration aspirin managed a century ago (except aspirin needed a world war, a seized subsidiary and a judge, and this took a couple of years).

A company’s name entering the language is the most valuable thing that can happen to it – and the beginning of losing it.

Language is the one commons that robs the rich: it takes from the most successful and gives to absolutely everybody, no vote required, no compensation paid.

The product gets copied, the patent expires, the stock eventually does what stocks do.

The verb is the only part that’s immortal – and immortality’s one condition is that it stops belonging to you.

Share

Subscribe now

Related: Synonym Companies – the 2023 take, from the brand’s side.

Sources: Bayer Co. v. United Drug Co., 272 F. 505 (1921) · Elliott v. Google, 860 F.3d 1151 (9th Cir. 2017) · King-Seeley Thermos v. Aladdin (1963) · WIPO Magazine on the Xerox campaign