On the Garden (against Citrini)

In 1661, André Le Nôtre completed the gardens at Vaux-le-Vicomte for Nicolas Fouquet, the French finance minister. The gardens were so spectacular that Louis XIV, upon visiting, had Fouquet arrested on embezzlement charges that historians now consider largely fabricated, and hired Le Nôtre to build something even bigger at Versailles.

Versailles is a garden designed from above. The Grand Canal extends nearly a mile along the central axis, aligned with Louis’ bedroom. The parterres de broderie are geometrically precise, their boxwood hedges trimmed into scrollwork patterns that can only be appreciated from the upper windows and most interior rooms of the palace. The allées radiate outward in perfect symmetry. The orange trees are placed in silver tubs such that they can be moved indoors in winter, because the garden’s geometry is not built to accommodate seasonal variance.

All images throughout from “Plans raisonnés de toutes les espèces de jardins” 1838 by Thouin, Gabriel, 1747-1829. Some scans from the Internet Archive, some from my own copy.

You could draw Versailles on paper and execute it to specification without ever visiting the site. In fact, I hear many billionaires across the gulf have done so, and this is essentially what Le Nôtre did. The plan precedes the place. The geometry is imposed on the land. The land’s existing contours — its hills, its drainage, its mature trees — were obstacles to be flattened. Le Nôtre moved tens of thousands of cubic meters of earth to level the terrain. He diverted rivers. Where the land resisted his plan, the land lost.

The French formal garden starts in the endgame — the perfect geometry — and works backward. The garden, once built, is meant to look as though it has always existed exactly as it does. Time is the enemy. Overgrowth is the enemy. Nature is the enemy. The gardener’s job in the French tradition is to arrest all three. Versailles requires at any given time hundreds of gardeners whose sole purpose is to prevent nature from reasserting itself and to restore it to its original plan. The geometry must be held.

The English, over centuries of thinking about gardens, developed a relationship with the natural world that has no real equivalent in any other culture.

Lancelot Brown was born in 1716 in Kirkharle, Northumberland. He was not an architect, not a painter, not a theorist, but a gardener. He began his career tending the gardens at Stowe for Lord Cobham, and rose through his ability and an extraordinary eye for what the land could become.

The story is that Brown would arrive at an estate and spend months walking the grounds with the owner, studying the contours of the land, the fall of light across a hillside, the path of an existing stream, the position of mature trees that had been growing for centuries before anyone thought to design around them, and pronounce after much consideration that the place had “great capabilities.”

Not that he could dominate the land like the French. Great capabilities — as though the land already knew what it wanted to be, and his job was to figure it out.Brown designed more than one hundred and seventy landscapes over thirty years. Blenheim. Chatsworth. Warwick Castle. Croome Court. The method was always the same. He walked the land. He worked with the existing contours. He planted trees not in rows or geometric patterns but in clumps and belts that mimicked the way trees naturally colonize a hillside. He created lakes by damming existing streams, allowing the water to find its own level and shape. He smoothed the transitions between the garden and the surrounding countryside until it wasn’t clear where the estate ended and nature began.

It’s worth pausing here on Central Park. Not only because it is probably the most famous landscape in America short of the national parks, but because it’s widely celebrated as a particularly American example of an English garden. I don’t believe it is one, and it’s worth explaining why.

Frederick Law Olmsted and Calvert Vaux won the commission to design Central Park in 1858. Olmsted had visited England, had walked the great estates, had seen Brown’s work and Repton’s work, and understood the aesthetic. The sunken transverse roads that cross the park invisibly below grade are ha-has, directly borrowed. The Ramble is meant to feel like an English woodland walk. The Great Lawn is the rolling parkland of a Brownian estate, scaled for a city far larger than the English countryside.

But the thing Olmsted built is fundamentally different from what the English built, and the difference matters. Central Park does not look like New York. This is obvious once you notice it and invisible until you do.

Manhattan is a granite island. The native landscape is rocky, vertical, and harsh. The Manhattan schist, some of the oldest rock on the eastern seaboard, breaks the surface everywhere. Before the city, the island was a tangle of salt marshes and tidal creeks in the lowlands, dense hardwood forest on the ridges, and massive glacial boulders deposited ten thousand years ago scattered across the terrain. It was dramatic, strange, and wild in a way that looks nothing like the English countryside.

Olmsted buried most of it. He imported hundreds of thousands of cubic yards of topsoil. He planted nearly five million trees, shrubs, and vines to create rolling meadows, gentle woodland walks, and pastoral vistas that belong in Oxfordshire, not on a North Atlantic granite ridge. The Sheep Meadow is an English lawn. The Ramble is an English woodland. The whole composition is a fantasy of the Home Counties dropped onto Manhattan.

This may seem like a pedantic distinction. It is not, and it is not merely aesthetic.

This is Le Nôtre with English aesthetics. The geometry is naturalistic rather than formal, but the method is French: impose a vision on the land by eliminating what’s already there. The curves are designed on paper. The wild areas are planted to specification. The meadows are manufactured. The whole thing is a representation of nature — an extraordinarily beautiful one — but it is not nature tended. It is nature performed.

We can only imagine what Brown would have done with this site. He would have arrived and walked the land and seen the schist and the glacial erratics and the drainage patterns and the salt marsh. And he would have said: this has great capabilities. He would have made a fundamentally New York park.I’m writing about gardens today because I work in technology, and technology is almost exclusively in the business of building new Versailles.

The pattern is so consistent it is almost impossible to see until you write it all out. A new system arrives. It surveys the landscape of whatever came before — the existing tools, the inherited flesh-and-blood workflows, the accumulated habits of millions of people and processes — and it levels the terrain. It imports its own topsoil. It plants its own geometry. The previous system is not incorporated, adapted, or even respectfully buried. It is flattened, because the new plan has no room for it. It is easier to build a beautiful geometric formation on top of it.

And then the maintenance begins. Hundreds of engineers — our gardeners — are deployed to hold the geometry. They prevent the natural entropy of real life from reasserting itself. They trim the hedges and patch the cracks and seed over the footpaths that emerge from the natural use of the system. They keep the parterres de broderie crisp and legible from the upper windows of the executive suites and boardrooms where the plan was drawn. The system, once shipped, is meant to look as though it has always existed exactly as it does. Time and user behavior are the enemy, and they must be eliminated. The mess of real life — flesh and blood pressing against the borders of an imagined, idealized, perfect system.

This is expensive. It is extraordinarily, ruinously expensive. And we keep doing it because the French garden is beautiful the day it opens. But there is a much worse thing that happens. When the geometry becomes too expensive to hold — when the hedges grow faster than the gardeners can trim them, when the system calcifies into something no one can move through — we do not attempt to work the land of the existing Versailles. We do not call in Brown. We build a new Versailles next door.

Healthcare is the clearest case. The existing system, for all its faults, is something much closer to an English garden than anyone in technology wants to admit. It is not ordered. It is not geometric. It is a strange, sprawling, deeply human landscape that grew over decades through the accumulated decisions of millions of diffuse people — doctors, patients, insurers, regulators — each responding to the contours of the terrain as they found it. It is inefficient in ways that are maddening and functional in ways that are invisible and hard to understand. I am not claiming it is elegant. I am not claiming it works for everyone. But the way a primary care doctor coordinates with a specialist and a pharmacist is neither elegant nor efficient, but load-bearing and functional, and all those pieces grew there for a reason, the same way a tree grows on a hillside for a reason even if no one planted it there and it interrupts our hedge line.

So what did technology do? It looked at the landscape and saw ugliness. It saw the inefficiency and the wait times and the paperwork and the experience that degrades year after year, and it did what Le Nôtre would have done: it leveled the ground and built something clean. Telemedicine pill mills. Cash-pay clinics. Lifestyle prescriptions delivered to your door with the frictionless ease of Uber Eats. The new garden was undeniably prettier and nice to walk through. The hedges were low, the paths were wide, the geometry was modern and inviting. But it had no relationship to or learning from the land it was built on.A patient who once had seventy-five percent of their care subsidized — via commercial or federal insurers — now paid for everything out of pocket. The coordination that the old system provided, imperfect and frustrating as it may have been, vanished. Drugs conflicted with each other. Medical supervision thinned to the point of performance. The consumer was invited into a beautiful new garden and discovered that the messy old landscape, the one that looked so ugly from above, had at least been keeping them from eating the dangerous plants that filled this new Versailles.

And the uncomfortable part: the palace itself is not paying for the maintenance of those beautiful French gardens. Versailles was maintained by the accumulated tax revenue of a million erstwhile and quite angry Frenchmen who never once set foot in the garden. The geometry was for Louis alone, but the bill was not his. The technology version of this is no different in structure, only in who receives the invoice. The consumer who fled the old system because it was slow and ugly and expensive arrives in the new one and discovers that every cost the old system absorbed — the coordination, the subsidy, the regulatory overhead that was maddening but also protective — has been transferred directly onto them. The garden is free to enter. The maintenance is yours.

This pattern is everywhere once you see it. Defense procurement. Financial markets. Crypto. The existing landscape is human and strange and disorderly in ways that are costly, but it is also adapted to its terrain in ways that are genuinely functional. And rather than studying it — rather than walking the land and asking what it’s already trying to become — we flatten it and plant something French. Something clean and geometric and spectacular on the day it opens. Something that requires an army of gardeners you pay for from that day forward. Something that has severed every root system that was holding the invisible weight of the hillside together.

I call this process parallel construction. A new Versailles built on cleared ground, next to a landscape that needed tending, not replacing. The old system left to decay. The new one uncontained, growing, passing more and more cost to the consumer. Neither one serving the people who actually live on the land.

What we almost never do is send someone to walk the land first. To spend months studying the contours of how people already work, the streams of information that already flow somewhere, the mature trees — the legacy systems, the inherited practices, the things that have been growing for decades — and to ask what this place is already trying to become. To approach a problem and say: this has great capabilities.

But there is a subtler case for parallel construction than mere caution. When the new system runs alongside the old rather than replacing it, the returns from technological progress can diffuse back upstream — slowly, imperfectly, but without requiring the destruction of everything load-bearing in the original landscape. We have seen this in healthcare already. The best products for years have been cash-pay, outside the normal system entirely — faster, cleaner, more responsive. And yet the innovations that proved themselves there have begun to push back into the mainstream: pricing transparency, patient-facing data, direct communication between doctor and patient that the old system made nearly impossible. The geometry of the new garden, tested in parallel, slowly reshaping the old one without having to raze it first.

Almost all of the language model discourse in recent days has imagined AI like French gardeners — or rather like viruses from outer space, inflicting themselves on society with no concern for what came before. It pretends we have no immune system to radical societal change.

This is a Citrini piece published this week as a particularly damning example of this. It participates in a form of Randian genre fiction in which markets are unceasing and rational beings that sit outside of human creation and are thrust onto us without choice. The fear of mass job displacement is real, but it rests on a flawed premise — that what we currently sit atop are radical, infallible systems of pure market competition. Capitalism has never actually been this. Global markets are, at most, a few hundred people coordinating with each other to make difficult trade-offs, organizing trillions in capital and billions of jobs. That is not a natural force thrust onto us.

We do not have to be French about this. We can look at what came before, learn from it, and usher in a societal reorganization closer to the English model — a more cultivated, intentional garden. It is foolish to pretend that anything in the history of modern financialized techno-capital is a raw and unfeeling system. This is a fiction we tell ourselves when market participants make trade-offs that hurt people, cause societal damage, and are not ones we would make in retrospect. The markets we participate in are intensely gardened. The people doing the gardening have names. It is a set of choices that people with names and addresses and human souls in the process of making. The Citrini piece is written as if the gardeners do not exist — as if the spiral arrives, as if unemployment happens, as if the daisy chain unravels the way a storm unravels, impersonally, without a hand on any of it. This is the most important fiction in the piece. Not the SaaS claims. The bloodlessness.

As we usher in our new age, we can choose to not be the French gardener. We can choose to be English, if not American about it.

The land almost always knows what it wants to be. The question is whether we are willing to listen.