The Delta Mind
The Snake and the Rope: Why Money Is Not Value
The Gap
In 2017, a group of Google researchers published a paper describing the transformer, the architecture beneath nearly every large language model since. The idea was given away. Much of the first great wave of money from it landed elsewhere, at companies that turned it into products people could use, and at the chip maker that sold the hardware to run it.
Now turn the picture around. For the better part of two decades, Nvidia built software and hardware for a kind of computing that most of the market did not yet need. The value was being created steadily, year after year, and the price caught up only when demand arrived all at once.
These two stories look like opposites: an idea given away that others turned into money, and a builder who got paid late. They are the same story. In both, money and value ran on different clocks, and the gap between the clocks was wide enough to hand a fortune to someone other than the creator, or to make one years after the work was done.
Most of us carry a quiet assumption that money measures value, that price is the world’s accumulated judgment written down as a number. Economists gave the assumption a formal name, the efficient market hypothesis, and for a long time it was taught as close to fact. This essay begins from the opposite observation. Money and value are not one quantity measured imperfectly. They are different things that meet only occasionally, and often by accident.
To see why, I want to go down layer by layer to the root. We will pass from the money in an account, through the stories that make things look valuable, through demand that is real and demand that is manufactured, to whatever remains when all of that is stripped away. It is a descent, and at the bottom there is something unexpectedly freeing.
Three Gaps
The first gap is value that is never priced. Linux runs a large share of the world’s servers and was given away. Wikipedia is the reference work of a civilisation and carries no price. The web, GPS, and the unpaid care that raises children and tends the old all create value on a scale that dwarfs many listed companies, yet almost none of it shows up as money.
A famous card-company advertisement made a version of this point by listing things money cannot buy. It is worth being careful here, because unbuyable and free are different conditions. A mother’s attention cannot be bought at any price. Wikipedia is free only because nobody built a toll gate in front of it. The second kind is the one that matters for this argument, because it shows that value without capture is often a fact of design, not a law of nature.
That points to the rule beneath the first gap: money follows control, not value. You get paid where you can exclude someone, own a bottleneck, or sit on the road everything must travel. A toll booth earns whether or not the road is any good.
The second gap is value priced on the wrong signal, and the present day offers plain examples. Safe Superintelligence, the lab founded by Ilya Sutskever, was last priced at about thirty-two billion dollars in 2025. As of August 2026, by public reports, it had shipped no product and reported no revenue. Thinking Machines Lab, founded by Mira Murati, raised about two billion dollars at a valuation near twelve billion before it had revealed what it was building. Neither is a fraud, and either may yet deliver something extraordinary. The point is narrower: today’s price rests on the founders’ names and on belief, not on anything a buyer has been able to use. The price is a measurement of belief about the thing.
Belief can also simply be wrong. Builder.ai, a London startup once valued at about 1.5 billion dollars and backed by Microsoft, marketed an AI that would build software as easily as ordering a pizza. It entered insolvency in 2025, after reports that much of what it presented as AI was the work of hundreds of human engineers.
The third gap is value priced late. Nvidia’s long wait belongs here, and so does the transformer paper, whose money arrived years later and mostly elsewhere. The efficient market hypothesis says prices reflect available information. Even if we grant that, it says nothing about whether prices equal value, and nothing about when they will. In 1980 the economists Sanford Grossman and Joseph Stiglitz showed that a perfectly efficient market is impossible: if prices already held all the information, no one would be paid to gather it, and so no one would. Some inefficiency is the price of anyone doing the work of pricing at all.
Add the old trader’s warning that the market can stay irrational longer than you can stay solvent, and the lag stops being a defect awaiting correction. It becomes a structural feature of how prices are made.
Three gaps, then: money absent where value exists, money present where value is only claimed, and money arriving on a schedule of its own. They look like three problems. I think they are one problem seen at three depths, and the way to see that is to go down.
Down the Stack
Start at the top, with money itself. Money is a claim on other people’s effort, and it is paid out when someone is willing to part with it. Willingness depends on what they believe they are getting. So beneath money sits perceived value, the buyer’s sense of worth at the moment of paying.
Perceived value does not form in a vacuum. It is shaped, and often shaped deliberately. Beneath it sits manufactured value: branding, narrative, status, scarcity staged for effect, the whole craft of making a thing look worth more than a plain inspection would suggest. Manufactured value is not always dishonest. A well-told story about a well-made thing is an ordinary part of commerce. But it is a layer apart from the thing itself, and it is the layer where most of the money-making effort in a modern economy is spent.
Under that sits something harder to see, which I will call engineered demand. Persuasion works on what people believe. Engineering works on what people want, and it can do so by building a need rather than meeting one. A product designed around variable rewards, endless scrolling, or the small intermittent jolt that keeps a hand returning to a screen produces demand that looks identical, on a sales chart, to demand for something people truly need. The number is the same. The thing behind the number is not.
Beneath engineered demand lies reflective demand: what people would choose if they were calm, informed and free of the pull. It is slower to form and harder to measure, but it is the part of demand that survives contact with time.
At the bottom is real value, what the thing actually does for the person who has it once the story, the pull and the pressure are removed. I will come back to how to define that without sliding into metaphysics. For now, notice the shape of the whole descent. Each layer stands on the one below it, and each layer can drift away from the one below it. Money can float free of perceived value, perception can float free of the thing, and demand can float free of need.
The three gaps from the previous section are simply places where that floating happened. Unpriced value is real value with no mechanism at the top to turn it into money. Wrongly priced value lives in the layers of perception and manufacture. Late-priced value is reality arriving slowly through the layers of demand. The gaps are the symptoms. The descent is the anatomy.
Where Money Detaches
Two things can corrupt the step from the thing to the money, and they are different enough to deserve separate names.
The first is deception. It acts on belief. A buyer is led to think the thing is better, safer, rarer or more necessary than it is. The remedy is information, and information eventually arrives, which is why deception tends to be priced out in the end. The end may be years away, however, and the fortunes made in the interval are real.
The second is compulsion. It acts not on what the buyer believes but on what the buyer can choose. A craving, once installed, does not need the buyer to believe anything false. The buyer may know perfectly well that the thing does not serve them and return to it anyway. This is what makes compulsion commercially stronger than deception: information does not dislodge it. A person who has read every study on the harm of a habit is not thereby released from it.
Notice what this does to the money signal. Of all the forms of demand, compulsive demand is the most dependable. It returns on schedule, it does not respond to price the way ordinary demand does, and it scales with exposure rather than with need. So the places where money-making is loudest are, uncomfortably often, the places where the connection to real value is weakest. This is not an accusation against any one industry. It is a structural observation: the mechanism that produces dependable revenue is the same mechanism that detaches revenue from benefit.
It also explains why the usual defence, that the market gave people what they wanted, says less than it seems to. Wanting is the very thing being worked on. If the want was produced by the process that profits from it, then the sale confirms the process, not the value.
What Real Value Is
Here is the difficulty a careful reader will raise. If real value is what survives once persuasion and compulsion are removed, someone must say what counts as good, and the right is exactly what people disagree about. Cultures, traditions and individuals differ, sometimes deeply, on what is worth having.
I do not think the argument needs to settle that, and the way around it is worth noticing. Instead of defining what is valuable, define what corrupts the one who judges. Real value, on this view, is value the person would endorse on calm reflection, with the deception and the compulsion removed. We do not need to agree about what is good. We only need to agree that a choice made under manipulation or craving is not a clean signal of what the chooser values. A follower of any tradition can accept that much, because every tradition has some word for the state in which judgment is clouded.
Seen this way, integrity gets a modest and workable meaning: acting so that the other party’s judgment stays intact. The word shares a root with integer, which means whole. An offer is whole when what is claimed matches what is made, and what is made matches what is needed. Integrity is not an ornament added to value. It is the condition under which value can be told apart from its imitations. Anyone who sets out to do the right thing, however they understand the right, is in effect trying to create value that will survive reflection, because they are not trying to win by clouding the judge.
The definition has one honest limit. It does not make real value easy to measure. Reflective endorsement can only be observed over time, and in the meantime the louder signals are the ones that are easiest to manufacture. But a hard measurement is still a measurement, and it comes with a useful consequence: it shows where to look.
The Lag Is an Audit
If real value sits at the bottom and money at the top, the long delay between them has a plain meaning: it is the time reality takes to audit perception. Every manufactured layer is a loan against the future. The product is bought, used and judged, and the judgment travels back up the stack, belief by belief, until the price corrects. Builder.ai was audited first by a lender that seized funds and then by the truth about who did its work. Thinking Machines Lab reportedly saw talks at a far higher valuation fall through early in 2026, and is reportedly in talks again at a lower one: the market is still auditing the name against the work. The lag is not the market being foolish. It is the audit still in progress.
Integrity, on this reading, is what makes the audit unnecessary. A thing that is what it says it is, for the person who buys it, has nothing to be found out. There is no later correction, because there is no gap to correct.
It would be dishonest to promise that integrity pays. The foundation of value is not a guarantee of capture. Manufactured value can be richly rewarded for years, and those who profit from it are not always caught. Real value with no chokepoint, like the open-source software that quietly runs much of the world, can go unpaid indefinitely. What integrity buys is not a faster payout. It buys durability, and freedom from the anxiety of waiting to be audited. It wins on the long clock, not on every clock, and a person who plans around the short clock alone is betting on not being looked at.
The present moment makes the point sharp. Artificial intelligence is producing real capability at remarkable speed, while a good part of the money that follows it so far settles at chokepoints such as chips, cloud capacity and distribution, rather than with the people the capability serves. Whether that capture proves durable depends on whether the value beneath it survives the audit. It is the same question, asked of a new object.
The Rope
There is an old image in the Indian philosophical tradition. A traveller walking at dusk sees a snake on the path and freezes. His heart races, his body reacts, and the fear is entirely real. Then someone brings a lamp, and the snake turns out to be a rope. The rope was always a rope. Nothing in the world changed when the light came. What changed was the seeing.
Money behaves a little like that snake. We treat price as a verdict: this is worth what it sells for, this person is worth what they earn, this idea is worth what it fetches. The pull of the market, the feeling of falling behind, the sense that a number on a screen is a judgment on us, all have real effects on the body and the day. But look at the stack. The price is a belief, the belief was shaped by a story, the story was carried by a craving that was engineered, and none of it touches the question of what the thing actually does. A person who sees this does not stop living in a world of prices. The rope is still there, and ropes have uses. What they stop doing is jumping.
I want to guard this against a misreading. Seeing the stack is not an argument for ignoring money. Rent still has to be paid, and a business still has to earn. The mature position holds two things together: money matters, and money does not define value. Neither cancels the other. This is the discipline of holding paradox that I have elsewhere called the Delta Split, and it is the only stance that avoids both naive idealism and quiet surrender.
Nor should the ending be too glib. Seeing a craving does not dissolve it. Someone who understands compulsion can still feel its pull long afterwards. What changes is smaller and more important: the belief that the pull is a verdict on what is valuable. Once that belief loosens, the grip loosens with it, slowly, and the person begins to choose from a place they recognise as their own.
For anyone trying to untangle money from real value, this is the whole gain. It is not a better strategy for chasing the price, and not a retreat from it, but the quiet recognition that the thing they feared was never holding them. They were free all along. They only needed the light.