I spent much of last week with people who farm. Farmers, landowners, land agents, and others whose working lives turn on the health of the ground they live on. Somewhere over those days an idea I had half held for years came clear, the way they sometimes do when what has been in front of you all along finally gets your attention
You cannot structure a culture. You can only grow one, and it takes its own time. We can catalyse changes, we cannot manage them. The insight arrived through how they talked to each other, and the land they were responsible for.
Farming is close to the economist’s picture of a perfect market. Thousands of producers, a largely undifferentiated product, and no single farm able to move the price. On paper they are each other’s competitors. In practice they lend machinery, share labour at harvest, compare notes on what is working for them, and sit on the same committees.
The competition that bites sits downstream. Four supermarket groups sell around two thirds of Britain’s groceries, and Tesco alone sells more than a quarter. Whilst the supermarkets are not especially profitable by the standards of other industries, their concentration gives them leverage over everyone who supplies them, serious enough for Parliament to create a Groceries Code Adjudicator in 2013 to police it. A farmer is a price-taker selling into a handful of buyers, with the margin squeezed where the power is weakest. It is the story of so many of our markets, from utilities to healthcare. When we apply business ethics to the commons, it is the commons that pays the price.
It is a familiar story. What struck me was what the farmers do with it. They treat competition as something that happens on top of the conditions they create for it, and they protect the conditions first. The farms will still be around when our current batch of supermarkets are history.
Nobody benefits from a contest that wrecks the ground it is played on. A farmer who mines the soil for three good years has borrowed from a system that does not renegotiate. Mariana Mazzucato draws the line between value creation and value extraction; the people I was with live on the creation side by necessity, because the land holds all the cards, no matter who has the power.
The word culture was a farming word before it was anything else. The Latin cultura meant the tending of land. Cicero borrowed it for the mind, cultura animi, the cultivation of a person. The metaphor has gone dead in the way we use the word, and so organisations speak of culture as something to be designed, launched, restructured and exploited.
A farmer, teacher or doctor would see the error at once. You do not build soil, or people, or health. You create the conditions, you wait, you pay attention, and you accept that most of what matters happens below the surface and on its own timetable. You can destroy soil, health and personal confidence quickly, but only repair it slowly.
Culture in an organisation works the same way. It is organic, a function of accumulated habit, trust and tacit knowledge of the people in it, and it grows, stagnates or erodes depending on how we care for it. A restructure can strip it out in a quarter, and no restructure can ever repair it.
This week also brought Grand Theft Auto VI, thirteen years after its predecessor, and by most reports the largest entertainment launch there has ever been. Rockstar has made a point of the fact that people made it: hundreds of thousands of animations, each art-directed by hand, and no generative AI. Azeem Azhar picked this up this morning as a story about craft, while the Economist suggests the game may mark the end of an era, for reasons that are as much economic as technical. Very few companies can fund a decade of work on one product, and the budget, which Take-Two has never disclosed, is estimated at well over a billion dollars.
It is easy to read this as a story about technology. Read alongside the farm, it is a story about culture. The craft in that game was grown over decades, in people, in a studio that kept and trained them long enough for the skill to compound.
The same company dismissed more than thirty staff in October last year. The union involved calls it union-busting; Take-Two says the dismissals were for gross misconduct, and the matter has been before an employment tribunal. The wider industry has a well-documented habit of releasing teams once a game ships. Whether Rockstar follows that habit in the coming months will tell us which it believes it owns: the game, or the ground the game grew in.
If the answer is the game, the next one will be made from whatever is left, with AI filling the spaces where people used to be. That would be a choice about culture, made in the language of efficiency.
Agriculture is not alone in this. Education, healthcare and public utilities are closer to the farm than to the factory. Their real output is slow, cumulative and mostly invisible in any single year: an educated population, a healthy one, infrastructure that holds.
We have pressed business structures onto them regardless, with business metrics, short-term targets and performance ratings. We did it because the business lens is the one most of us were trained to use, and it makes complex living systems look manageable. Pupils become attainment scores. Patients become episodes of care, and both become costs to be weighed against outputs, and evaluated in the language of returns.
James Scott tells the story of eighteenth-century German foresters who replaced mixed woodland with neat rows of a single profitable spruce. The first rotation was a triumph. The second and third declined, because the forest had been reduced to the one thing the forester could count, and everything that had kept the soil alive was gone. The Germans came to have a word for it: Waldsterben, forest death. It took a full generation to show.
Which brings me to AI, and to why this matters now.
AI is a stunning technical achievement, and it can do a great deal more than add a few points of margin. But it arrives in organisations through whatever lens already governs them. In a business that sees itself as a structure, AI will be used as a structuring tool: to make the work more legible, to reduce the people to what can be measured, and to remove the ones who cost more than they appear to return. It will produce a first rotation that looks like a triumph.
The other path is the farmer’s. Use AI to strengthen what people can do, so that skill compounds rather than being stripped out. Treat the people and what they know as the soil the organisation grows in, and judge each use of the technology by the question a farmer asks of every season: does this leave the ground better or worse for the next one?
I think we are at the point where that choice gets made, mostly without anyone noticing that it is a choice. One path enhances people. The other applies a sterile kind of AI to buy a few years of efficiency for organisations whose structure is already obsolescent.
The farmers I spent the week with already know how that second story ends. They have watched what happens to land that is worked for the yield and not for the soil, and they have decided what matters.
So should the rest of us.
sources
Mariana Mazzucato, The Value of Everything (2018). Value creation and value extraction.
James C. Scott, Seeing Like a State (1998). Scientific forestry and the costs of legibility.
Cicero, Tusculan Disputations, II.13, for cultura animi.
Richard Sennett, The Craftsman (2008). Skill as something that compounds in people over time.
The Economist: https://www.economist.com/culture/2026/11/18/grand-theft-auto-vi-may-mark-the-end-of-an-era-in-gaming
Azeem Azhar:


