In 1666, the Great Fire of London consumed 13,200 houses in four days. What followed is the part the history books tend to rush past -- was not just a city rebuilt but a set of human behaviours invented from scratch. Before the fire, there were no fixed-price building contracts. A gentleman told a mason what he wanted, the mason built it, and both parties settled accounts when the dust cleared. After the fire, Nicholas Barbon, a speculator, physician, and -- depending on which historian you ask -- the most creative con artist of the seventeenth century, saw an opportunity in the rubble. Not just a commercial opportunity. A political one. The city needed to be rebuilt quickly. Parliament needed it rebuilt cheaply. Nobody was asking too many questions about quality.
Barbon did something nobody had done at scale before. He offered to rebuild your house for a fixed price. You told him what you wanted, he told you what it would cost, you paid him, and he delivered it. The genius of the arrangement was that it transferred all the risk from you to him. If the timber cost more than expected, his problem. If the labour took longer, his problem. If the foundations turned out to be unstable, his problem. You got certainty. He got the contract.
What Barbon understood -- and what every builder, contractor, and technology consultancy has understood ever since -- is that human beings will pay a premium for the feeling of control over an uncertain process. The fixed-price contract is not a tool for managing construction. It is a tool for managing anxiety.
There was, of course, a catch. Barbon's buildings were notorious. Thin walls, cheap materials, corners cut with the casual indifference of a man who had already been paid. He is remembered today not as the father of modern construction contracts but as the father of jerry-building. The London Building Act of 1667 -- one of the first building codes in the world -- was written in direct response to the quality of his work.
That was 1666. I want to tell you about something that happened to me about nine years ago, the uncomfortable truth at the centre of this story is not that I hired bad builders. It is that I was a bad client.
The Slow Accumulation
If you ever want a recommendation for a bad builder, I would say me first. Not the builder -- me. I chose them. I managed the project. I failed to do the due diligence that was available to me. I did not hire a quantity surveyor, which the Federation of Master Builders will tell you costs 1-3% of the project budget and has the highest return on investment of any fee you will ever pay on a renovation. I did not use a formal contract. I did not check whether my builders had, individually or collectively, ever executed a job this big.
They had not.
This is the part that most renovation horror stories leave out, and it is the part I need to name before I blame anyone else. The builders were not malicious from the start. They were in over their heads. They had not individually executed a project of this complexity before, and they were naive about their own capabilities in the way that every craftsman who has ever taken on a stretch assignment has been naive. They believed they could do it. I believed they could do it. We were both wrong.
And they were not making this choice in a vacuum. In an industry where 82% of contractors bid below cost to win work, the alternative to overbidding is often no work at all. The honest builder who prices a job at what it will actually cost loses the job to the one who prices it at what the client wants to hear. My builders were not outliers. They were the system working as designed.
The realisation of betrayal was slow, not sudden. It was not a single moment of discovery but a gradual accumulation -- a drip, drip, drip of small things that were not quite right. The bathroom tap that ran a thin bead of water down the wall behind it whenever you turned it on, because it had not been seated properly -- something I would not discover for three years, by which time the plaster behind the tiles was rotten. A conversation where the answer to my question was just a little too quick, a little too smooth, delivered with the confidence of someone who is performing knowledge rather than possessing it.
Then the splashback glass.
I had paid for a single, large piece of glass as a kitchen splashback. One piece. No joins. The builders could not be bothered to remove the taps. Rather than take the taps off, disconnect them, fit the glass, and reconnect -- which is, to be clear, the job -- they insisted the glass cutters take the glass away and cut it in half so it could be fitted from both sides. Two pieces where there should have been one. A join where there should have been none.
I see that join every single day.
That is not the worst of it. Under a floorboard, during a later repair that I should not have needed to make, I found a Dominos pizza box. With scraps of pizza still in it. Stuffed under the floor for no apparent reason by people I was paying to build me a home.
You might think I am telling you this because it is dramatic, because pizza boxes under floorboards make good stories at dinner parties. I am telling you this because the pizza box is the Ronan Point of domestic renovation. In 1968, when investigators pulled apart that collapsed tower block in east London, they found newspapers stuffed in panel joints. Not malice -- indifference. The kind of indifference that settles in when a builder is in over his head, behind schedule, out of his depth, and has stopped caring about the work because he has started caring about survival. Newspapers in 1968. Pizza boxes nine years ago. Technologies change. Human nature does not.
Samuel Morton Peto and the Descent
Samuel Morton Peto was, by the 1840s, the most respected railway contractor in Victorian Britain. He built Nelson's Column. He built the Reform Club. He built railways across three continents. He was a Baptist, a philanthropist, a Member of Parliament, and -- by all accounts -- a man of genuine integrity. He paid his workers fairly. He completed contracts on time. He was, in the language of the construction industry, a safe pair of hands.
Then he took on too much.
The railway boom of the 1840s and 1850s created extraordinary demand. Peto, like every successful contractor before and since, expanded. He took on contracts that were larger, more complex, and more geographically dispersed than anything he had managed before. The investors demanded returns. Parliament granted railway concessions with enthusiasm and oversight in roughly inverse proportion. The fixed-price model that had made his reputation became the mechanism of his destruction. When costs exceeded estimates -- and they always did, because railway construction through unfamiliar terrain is genuinely uncertain work -- Peto absorbed the losses. When the losses mounted, he borrowed. When the borrowing became unsustainable, he began to cut corners.
This is the arc I want you to see, because it is the arc of my builders.
Peto did not wake up one morning and decide to become dishonest. The dishonesty developed incrementally, under pressure, over time. A cheaper material here. A thinner rail there. A foundation that was adequate rather than excellent. Each individual compromise was small. Each was rationalised. Each was the product not of villainy but of a man who had got in over his head and could not find a way back.
When the Overend, Gurney banking crisis hit in 1866, Peto's entire empire collapsed. He was bankrupted. The man who had built Nelson's Column died in relative obscurity, his reputation destroyed not by a single act of fraud but by a long, slow descent from good faith to corner-cutting that was -- and this is the point -- structurally inevitable given the incentives he faced.
His workers, for the record, fared worse. The navvies who had dug the cuttings and laid the rails had no pensions, no redundancy, no recourse. They went to the next job or they went to the workhouse. The system that broke Peto ground them to dust.
That was 1866. My kitchen was 2017. The distance is cosmetic.
The Arc You Recognise
My builders were not Samuel Morton Peto. They did not build Nelson's Column. They were not Members of Parliament. But the arc was identical. They started with good intentions and genuine, if naive, confidence. They took on a job that exceeded their capabilities. When things went wrong -- and in renovation, things always go wrong, because you do not know what is behind that wall until you open it -- they did not stop and say "we are out of our depth." They pushed forward. The corners they cut got bigger. The quality dropped. And somewhere in that process, the relationship between us shifted from collaboration to something adversarial.
I watched YouTube videos at midnight to understand earth continuity issues. I watched YouTube videos to learn how to apply silicone around a bath fitting -- not how to do it well, mind, but how to tell whether someone else had done it at all. These were things the experts were supposed to handle. The fact that I was teaching myself their trade on a screen at half past eleven, still in my coat because I had just come from work, whilst paying them to do it -- that should have been the moment I fired them. It was not. Because by that point, the sunk cost fallacy had me as firmly as it had them. I had invested too much money, too much time, too much emotional energy. They had invested too much labour, too much reputation, too much identity. Neither of us could walk away.
The economists have a name for this kind of market. Construction is a credence good -- a product or service whose quality cannot be evaluated even after consumption. You can taste a meal and know if it was good. You can drive a car and know if it handles well. But a re-wired house? A re-plumbed bathroom? The defects may not reveal themselves for a decade. The tap that weeps behind the tiles. The earth bonding that was never tested. The joist that was notched too deep to take the pipe. My house has been revealing its defects for nine years, and I am still finding them.
I should be precise here, because the blanket condemnation of fixed-price contracts is too easy. Replacing a boiler is not the same as renovating a Victorian terrace. A boiler replacement is commodity work -- known scope, known components, known installation patterns. You can price it accurately. A fixed-price contract works fine. But structural renovation of a house built 130 years ago, where you genuinely do not know what is behind that wall until you open it? That is uncertain work. The fixed-price contract turns that uncertainty into a bet, and the builder absorbs every surprise as a loss. The question is not whether fixed-price contracts are bad. The question is whether you are using a commodity contract for custom work. Most domestic renovations are custom work priced as though they were commodity. That is where the trouble starts.
The UK has no mandatory licensing for general builders despite 81% of the public supporting one. Stop and think about that for a moment. Medicine is a credence good -- you cannot evaluate your surgeon's work while you are under anaesthetic. Law is a credence good -- you cannot assess your solicitor's competence until the verdict. Financial advice is a credence good -- you will not know if your pension was well managed until you retire. All three are licensed professions. Construction, where the consequences of failure include structural collapse, electrical fire, and gas explosion, is not. Why? Because the skilled builders who would benefit from licensing are too fragmented and too busy surviving to lobby for it, while the system as it stands suits the large developers and volume housebuilders who run their own quality assurance and profit from a domestic market where small builders absorb all the risk and homeowners absorb all the consequences. Apprenticeship completion rates fell to 41% in 2024. The skills pipeline is collapsing. Nobody with power is doing a thing about it.
The uncomfortable truth I keep returning to, the one that Nicholas Barbon's clients in the 1670s could have told you, and that Peto's railway investors learnt to their cost in 1866: an uninformed client with strong opinions is the worst thing that can happen to a build.
I was that client.
The Tour of Shame
I do the tour of shame now. Visitors come to the house and I walk them through it and I narrate, internally, every hidden defect behind every wall. The bedroom door that does not sit square in its frame -- you can see daylight through the gap at the top if you look at the right angle, and I always look at the right angle. The plaster in the hallway that is slightly uneven because they skimmed it too fast, and every time the evening light hits it from the west I can see the ripple. The electrical work that I had to check myself because I could not trust that they had tested it properly. Nine years of walking through my own home and seeing not the home but the archaeology of a failed relationship.
The IKEA effect -- Norton, Mochon and Ariely showed that labour leads to love -- works only when the outcome is successful. Build a bookshelf and it wobbles slightly and you love it more because you built it. But a failed renovation inverts this entirely. The labour creates not love but hyper-vigilance. You notice every imperfection because you paid for it, you fought for it, you watched it go wrong, and you could not stop it. Nine years of noticing is the IKEA effect in reverse.
The builders, to their credit -- or perhaps to nobody's credit -- did not set out to do this to me. They set out to do a good job and failed. I set out to be a good client and failed. We were both trapped by a market structure that Akerlof described in 1970 as the market for lemons: when the buyer cannot distinguish quality from junk before purchase, the market degrades until only junk is left. That is domestic construction in Britain. The FMB reports that fear of cowboy builders costs the UK economy ten billion pounds a year in forgone work. 40-78% of renovations exceed budget. But they also report that 75% of builders say cowboy clients are a serious problem. The dysfunction runs both ways.
What Barbon Knew
Nicholas Barbon, standing in the ashes of London in 1666, understood something that we have been trying to unlearn for 360 years. He understood that the desire for certainty in an uncertain process creates a contract that is adversarial by design. The fixed-price contract says: I will tell you what it costs, and you will pay it, and whatever happens behind the walls is my problem. It is a bet disguised as an agreement. The builder bets he can do it for less than the price. The client bets the price covers everything. Both are usually wrong.
Barbon also knew that an unregulated market with no quality signals will always drive out the honest operators. His jerry-built houses were cheaper than the good ones. His clients chose price over quality because they had no mechanism to distinguish the two. Three hundred and sixty years later, the mechanism still does not exist.
If you are renovating a house, the advice is brief and specific. Hire a quantity surveyor. Use a JCT Minor Works contract. Ask yourself whether the work is commodity or custom -- and if it is custom, stop pretending a fixed price means anything. Use time-and-materials with stage gates, independent inspection, and the right to walk away.
But I did not write three thousand words about Nicholas Barbon and a pizza box to give you renovation tips.
If you commission digital transformation -- or, as is now unavoidable, AI transformation -- you are standing in Barbon's London right now. The ashes are different. The dynamics are identical.
You are buying a credence good. You cannot evaluate the quality of an AI model's training data any more than I could evaluate the wiring behind my walls. You cannot tell whether the retrieval-augmented generation pipeline your vendor built is robust or brittle -- not at delivery, not six months later, possibly not ever -- until it hallucinates in production at the worst possible moment and you discover that nobody tested the edge cases because nobody scoped them and the contract did not require it. The defects are hidden by design. The plaster goes up. The demo looks clean. And behind the wall, there is a pizza box.
You are hiring vendors who are in over their heads and will not admit it. The AI consultancy market in 2026 is my builders in 2017 at industrial scale. Teams that have never shipped a production AI system are bidding for contracts that require it. They believe they can do it. They are not lying -- not yet. They are naive about their own capabilities in exactly the way that Peto was naive when he took on one railway too many. The overconfidence comes first. The corner-cutting follows. The dishonesty arrives last, incrementally, structurally, and by then the sunk cost fallacy has you both locked in.
You are signing fixed-price contracts for work that is genuinely uncertain. Building an AI system on your proprietary data, integrated with your legacy architecture, subject to regulatory constraints that are changing quarterly -- that is not a boiler replacement. That is opening a wall in a Victorian terrace and discovering that the previous owner ran the plumbing through a load-bearing beam. The honest answer is "we do not know what this will cost until we understand what we are dealing with." The winning bid is the one that pretends otherwise. Barbon would recognise the arrangement instantly.
And you have no feedback loop. In construction, I could have hired a quantity surveyor -- an independent expert whose job is to see what I cannot see and say "stop" before the damage is done. In AI transformation, the equivalent barely exists. The people evaluating the vendor's work are often the vendor's own team. The client's internal reviewers -- if they exist at all -- lack the specialist knowledge to assess model performance, data quality, or architectural decisions that will compound for years. You are the homeowner who does not know what good wiring looks like, paying someone who may not know either, with no one in the room whose job it is to tell the truth.
The sunk cost trap closes around AI programmes even faster than it closed around my kitchen. The budgets are larger. The switching costs are higher. The executive who championed the initiative has staked reputation on it. And the vendor -- behind schedule, under-resourced, performing confidence rather than possessing competence -- cannot afford to say "we are out of our depth" any more than my builders could. The incentives are aligned for silence. The silence is aligned for failure.
None of this is inevitable. Use contracts that acknowledge uncertainty rather than wishing it away. Insist on independent technical review at every stage gate -- someone who works for you, not for the vendor. Demand that the commodity-versus-custom question is answered honestly before a price is agreed. Build the feedback loop that the market will not build for you.
I have spent nine years living with the consequences of a market that Nicholas Barbon invented in the 1670s, that Samuel Morton Peto navigated in the 1850s, and that every organisation commissioning an AI transformation is navigating right now. The technology changes. The materials change. But the human dynamics -- the overconfidence, the sunk costs, the credence good problem, the slow accumulation of small compromises until the whole thing is rotten -- those do not change.
I know that every morning. The join in my splashback glass tells me.
(Views in this article are my own.)