Lesson · Economy Fundamentals
Zero Surprise
Why the whole economy, from the earnings call to the gold vault, is built to do one thing: meet expectation.
The granary
Slide the Actual. The Expectation stays at 2.13. Watch what happens to the store when the ratio leaves one, in either direction.
Actual 2.13 ÷ Expectation 2.13 = R 1.000 · S = ln R = 0.000 · The store holds still.
An illustration, not a model. How far the ratio sits from one sets how hard the store is shaken. A miss is weighted about 2.25 times a beat of the same size: the exchange rate on bad news.
Contents
The ratio on the call
Every quarter, a public company stands up and reports what actually happened. The people listening already had a number in their heads. In my framework, that is the whole game. Reality is a ratio, Actual over Expectation: R = A / E. How strongly a reality is felt is its natural log: S = ln R.
Put the call through it. The Actual is the reported earnings per share. The Expectation is what the people who hold the stock carried into the call, which the Street summarizes as the consensus. Two-thirteen over two-thirteen is one. The natural log of one is zero. Zero surprise.
A note for students, because the words get tangled. The number on that call is earnings per share, not the P/E ratio. The P/E is what the market is willing to pay for each of those dollars, and it is itself a wager that future expectations will be met. It is expectation all the way down.
Two more things before going on. First, name whose reality this is. The company is not the one doing the comparing; the holders of its shares are. They are the Actualizer on the call. Second, Wall Street already computes this. When an analyst says a company beat by three percent, that percentage is, for small beats and misses, almost exactly ln R. The Street has been measuring surprise for decades. It just never called it that.
The granary
Why does anyone care? Start before money.
The first hard problem in economics was never making a surplus. It was keeping one. Grain rots if it gets wet. It feeds rats. In the Joseph story in Genesis, the plan for seven fat years followed by seven lean ones was simple: put away a fifth of every harvest, and guard it. A surplus that cannot be stored is not wealth. It is a feast.
Today the harvest is enormous. Pick any figure with fourteen zeros and you are in the neighborhood of the world’s stored value. I do not want to call it currency or capital; both words are too small. It is just the store. And the store has the same enemies grain always had. It can rot. It can evaporate. It can be taken. The textbooks have a name for each leak. Leave the names aside for now and keep the picture: a granary, and a great many people worried about moisture.
Now ask why someone with a hundred billion dollars buys shares in a company. Not because they believe in it. Not because they want it to succeed. They need somewhere to put the grain. Every asset is a granary, and the holder has one real question: will it keep?
Predictability is the product
Here is the principle, stated plainly. Everything about storage concerns the future, and the future cannot be inspected. It can only be expected. So the holder of the store does not need a granary that grows. They need one that does what they expected it to do. Quarter after quarter, they are measuring one thing: how close the ratio stays to one.
That is why the safest bond in the world pays the least. It surprises the least. Return is not a reward for virtue. It is the price someone must be paid to carry the surprise you would rather not carry yourself.
Predictability is the product. Return is the price of its absence.
The same reading works for every store there is:
- Bonds. The purest version. A bond promises an exact Actual on an exact date. Its only surprise is not getting paid.
- Equities. Earnings arrive every quarter and are held against what was carried in. The call is the moisture check.
- Real estate. Rent and resale, measured against what the buyer assumed when they signed.
- Art and collectibles. No earnings at all. The Actual is what the next holder will pay, against what this holder expected.
- Gold. Thousands of years of meeting the one expectation that matters to it: that it will still be wanted.
- Bitcoin. The same wager as gold on a far shorter record, which is why its ratio swings so far from one.
Even the language of markets is the language of expectation. A bear expects prices to fall. A bull expects them to rise. Neither word describes what happened. Both describe what someone expects.
Why a big beat is not a gift
Now the counterintuitive part. Suppose the CFO reports three dollars a share against two-thirteen. The ratio is about 1.41; its log is about 0.34. The trader’s phone lights up. Is it good news?
To the trader, perhaps. To the holder of the store, it is a crack in the granary wall. Three reasons.
- A beat is a forecasting failure. If the people closest to the company could not see a forty percent swing coming, the next swing can go the other way just as easily. A granary that surprises you with extra grain can surprise you with less.
- A beat moves the bar. The ratio cannot stay at 1.41. The holders’ expectations retrain on the new Actual, and next quarter has to clear it. In The Conservation of Surprise I showed that surprise is a log, so it telescopes. A big beat now is often a disappointment borrowed from later.
- The minus sign costs more. The math is symmetric: a doubling and a halving are mirror twins. People are not. As I wrote in The Exchange Rate on Bad News, attention pays roughly twice as much for a miss as for a beat of the same size. A store that swings both ways loses more on the way down than it gains on the way up.
The executives already know this. When three finance professors surveyed hundreds of chief financial officers, most said they would give up real economic value to report smoother earnings. Their reason: missing a target, or swinging wildly, makes a company less predictable, and predictability is what the share price is paying for. They were describing ln R = 0 without the symbols.
No one can touch the denominator
It is tempting to read all of this as manipulation: companies talk the number down so they can step over it. Some try. Researchers have found reported earnings bunching just above the thresholds that matter, the analysts’ consensus among them.
But notice what is actually possible. No company can reach into the minds of its holders and set their Expectation. The denominator cannot be touched, by anyone, by any means. What a company can do is act. A guidance call is an action. It leaves an artifact: a transcript, a number in the record. That artifact becomes history, history becomes Actual, and only Actuals retrain what people predict. That is why walking expectations down takes a quarter of patient, consistent artifacts, not one phone call. The road to the denominator always runs through the numerator.
What AI changes
So far, this is old. Here is what is new.
AI reaches both inputs at once without touching either term. On the expectation side, prediction is becoming cheap, constant and dense: every holder will carry a model trained on every filing, shipment and receipt. On the actual side, operations run by AI land closer to plan. Fewer things happen by accident.
The obvious conclusion is that surprise is going away. It is not. Surprise is conserved; only its payment schedule and its address change. As AI wrings it out of everything that can be predicted, what remains collects in the one place no model can reach in advance: the decision someone owns. The new idea. The bet a founder makes. The outcome someone is liable for.
The economy will pay more and more for a ratio of one. And it will pay the humans who stand where the ratio is not one.
Five predictions
The quarterly jolt dissolves into a daily drip.
When actuals stream into models that never sleep, expectations update so often that the quarterly report stops being an event. The total surprise is the same. It is paid in pennies instead of lumps.
Landing replaces beating.
The proudest word on a future call will not be “beat.” It will be “landed”: we said 2.13 and delivered 2.13. A large beat will be read as a forecasting failure, and boards will ask why their own systems did not see it coming.
Boring gets expensive.
Companies whose AI-run operations hit plan quarter after quarter will be priced more like bonds than like wagers. Predictability, manufactured at scale, will command the premium that growth used to.
The store-holders multiply.
As I argued in Trillions, AI delegates are becoming economic actors. Each one that saves on someone’s behalf is a new holder of a small store, and each wants what the hundred-billion-dollar fund wants: a ratio of one. Demand for stores that keep will grow faster than the stores themselves.
Human pay moves to where the ratio is not one.
The work that pays will cluster around carrying surprise: founding, underwriting, guaranteeing, being the name on the outcome. I made the liability case in Paid to Be Liable. This is the same claim from the other side. When predictability is cheap, the person who absorbs the surprise is the scarce one.
Where I stand in this
I spent the E2open years inside supply chains, which are expectation machines: a forecast on one side, a shipment on the other, and an industry paid to close the gap. Now I watch the same principle from a smaller room. What a restaurant owner buys from us at Charleston AI is not a technology. It is a ratio of one: every call answered, every booking kept, no surprises on a Friday night.
It is also the first thing I tell my students. Your first employer is a store-holder too. In First Customer I tell them to keep their needs low so they can treat that employer as a customer. The other half is this: become the person whose Actual reliably equals what was expected of them. In any company, that person is the safest granary in the building.
The claim ledger
- Already true
- Executives say plainly that they manage for predictability and would give up value to get it. Analysts measure surprise as a percentage that, for small surprises, is the log of the ratio. Reported earnings bunch just above the consensus.
- What has to happen
- Actuals must stream continuously enough, and holders’ models must update fast enough, that a quarterly report carries little left to learn. AI-run operations must actually land closer to plan, which is an engineering claim, not a philosophical one.
- Where I am probably wrong
- Today, markets usually reward a beat. If they still reward big beats in 2031, the store-holder’s flinch is real but outvoted by the trader’s cheer, and my second prediction fails. The log also breaks where the ratio does: when a company loses money, the ratio can go negative and ln R stops meaning anything, which is why the Street switches to dollar differences there. And not every holder wants zero surprise. Venture capital is the part of the store paid to want the opposite, and this essay does not describe it.
Go back to the call. Two-thirteen against two-thirteen, a few seconds of silence, a stock that barely moves. A student hearing it for the first time would think nothing happened.
Nothing happened. That was the product.
Background
- Genesis 41: Joseph and the granaries of Egypt.
- Degeorge, Patel and Zeckhauser, “Earnings Management to Exceed Thresholds,” Journal of Business, 1999.
- Graham, Harvey and Rajgopal, “The Economic Implications of Corporate Financial Reporting,” Journal of Accounting and Economics, 2005.
- The Reality Equation
- The Conservation of Surprise
- The Exchange Rate on Bad News