Psychology
On 29 November 1962, Britain and France signed a treaty to build a supersonic airliner together. The document carried a defect that mattered more than anything in the engineering. It had no clean way out. Either government could come to want the programme stopped. Neither could stop it alone without breaking an international agreement. The exit was designed out of the project before the first rivet went in.
The exit was not missing by accident. It was removed so that nobody would ever have to make the decision twice.
Concorde first flew in 1969 and entered commercial service in January 1976. It was a magnificent machine. It was also, by the middle of the sixties, understood inside both governments to be a commercial loser. Development estimates that started near seventy million pounds finished above a billion.
Fourteen aircraft ever entered airline service. The programme ran until 2003 and never returned what it cost to build. In 1976 two biologists, Richard Dawkins and T. R. Carlisle, gave the behaviour its name in a paper in Nature. They called it the Concorde fallacy: continuing to put money into something because of the money already in it.
Every founder knows the arithmetic. Money already spent is gone. It is not an input to whether the next dollar is worth spending. The only live question points forward: from here, knowing what I know now, is this the best use of the next month and the next hire.
Everybody knows that. Almost nobody does it. The distance between knowing the rule and following it is not an intelligence problem. It is an identity problem.
The real mechanism
Sunk cost gets taught as a math error. It is not one. A math error gets corrected the moment you show somebody the math. Show a founder the numbers on a dying service line and watch what he does. He will not dispute the numbers. He will defend the decision.
Because the spend is not sitting in a spreadsheet. It is sitting inside his account of himself. Three years, a rebuild, and two hires is not a line item. It is a chapter. Killing it does not cost him the money, since the money is already gone. It costs him the meaning of the three years.
You are not loyal to the project. You are loyal to what it cost you.
Nietzsche wrote that a man who has a why to live for can bear almost any how. True, and it runs the other direction too. Give a person enough suffering and he will manufacture a why to make it bearable. The harder the road was, the more necessary it becomes that the road was going somewhere.
Which is why the projects that are hardest to kill are never the cheap ones. They are the ones that hurt.
How it shows up
It rarely announces itself. From the operator’s seat it wears ordinary clothes. A service line that eats half your delivery hours and returns a fraction of your margin, kept because you built the entire process around it. A client who stopped being profitable a year ago, kept because you reshaped the business to serve them, which is the quiet cousin of customer concentration risk. A unit that has swallowed every renovation decision you made and still trails the comps. A hire you have spent nine months training who is still not the person the seat needs.
Notice what they share. In every one of them, the case for continuing is written in the past tense.
The test
One question kills it, and it has to be asked out loud, ideally to somebody who does not work for you.
If I did not already own this, would I buy it today at today’s price?
If the answer is no, you are not holding an asset. You are holding a receipt.
The question works because it strips out the history. It hands the decision to the version of you who walks in this morning with no chapter to protect. That person has no trouble walking away, because as far as he is concerned nothing was lost. Nothing was lost for you either. It was lost already, on the day the money left.
I quit a chair I was good at. Years of skill, a book of repeat clients in Canada’s largest barbershop, in a trade I had actually mastered. All of it was real, and none of it was a reason to stay once I could see the ceiling on it. The skill was not wasted. Speed, precision, and performing in front of a line of people waiting all came with me. What did not come with me was the chair, so I left the chair.
Later I cut services out of the agency that I had built the whole delivery system around. Same move, less romance about it. Narrowing the menu cost me a system I was proud of. The system was not the business. It was overhead wearing a costume.
A sunk cost only becomes a real loss when you spend the future defending it.
The exception worth naming
The principle gets abused in the other direction too, so it is worth marking the line. Not every long, expensive, unrewarding stretch is a sunk cost. Some things compound slowly and pay late, and quitting them early is its own failure mode.
The difference is evidence. If the thesis is intact and the results are lagging, that is patience. If the thesis has quietly been replaced by the investment itself, that is sunk cost. Patience has a reason that survives being stated to a stranger. Sunk cost has a reason that only makes sense to the person who paid.
What you spent is a fact. What you do next is the only part still yours.
