The Conservation of Surprise

This series runs on one quantity. You carry an Expectation; the Actual arrives; the gap between them is surprise, and surprise is the only currency attention is ever paid in. The Reality Equation reads that gap as a ratio on a log scale — S = ln(A/E) — and the last piece showed the log was never a choice: any readout of a ratio that adds must be a logarithm. That was an argument about lawfulness. I want to pull one more consequence out of it, because it is sitting in plain sight and almost nobody in the business of expectations behaves as if they know it.

Logs telescope. If your expectation of some event starts at E₀ and reality finally arrives at A, it does not matter how many times your expectation gets revised along the way — a teaser here, a leak there, a rumor, a walk-down, a spoiler. Each revision pays you a piece of surprise, the log of the new expectation over the old. Add the installments up and the middle terms cancel in pairs:

That is the whole theorem, and it is not deep — it is the same additivity that made the log lawful, read in the other direction. But say it in the series’ vocabulary and it becomes a strong claim about experience:

The total surprise an event can pay you is fixed by where your expectation began and what actually arrived. Everything in between only decides the schedule of payment.

Surprise, in other words, is path-independent — conserved between endpoints the way height is conserved between the base and the summit of a hill. You can take the stairs or the elevator or a decade of switchbacks; the climb is the climb. A trailer does not add surprise to a film and a spoiler does not subtract it. Both move a payment from one date to another. The anticipation industry — an industry that spans movie studios, product launches, investor relations, and every countdown clock on the internet — is not in the business of manufacturing surprise. It is in the business of scheduling it.

Once you see the ledger, you see the ledger everywhere.

The teaser campaign is an installment plan. Your expectation of the film starts at another sequel. The teaser lifts it; that lift is surprise, paid now, months before the premiere. The trailer lifts it again — a second payment. By opening night your expectation stands so close to the film itself that the theater can only pay out what the campaign has not already spent. The studio has not changed what the film is worth to you. It has moved most of the payout to dates it controls, on screens it owns, surrounded by ticket links.

The guidance call is the same instrument run in reverse. Earnings season is scored as beat-or-miss, and the beats are remarkably reliable: by FactSet’s running tally, 86 percent of S&P 500 companies beat their consensus earnings estimate in the second quarter of 2026, against a five-year average of 78 — quarter after quarter, on numbers that are supposed to measure surprise. The engineering is no secret. Expectations are guided down between quarters precisely so the announcement lands as a small positive payment in public. The telescoping identity says what that beat is made of: surprise collected early, analyst by analyst, in the walk-down — with the sign flipped at the end so the last installment reads as good news. The total, from what the market believed at the start of the quarter to what the quarter actually was, is unchanged. Only the schedule moved, and the schedule is the product.

And the spoiler — the villain of the whole folk theory — turns out to be the most honest test of the claim. If spoilers destroyed surprise, spoiled stories should be worth less. In 2011, Leavitt and Christenfeld ran the experiment at UC San Diego: hundreds of readers, a dozen short stories — twist endings, mysteries, literary pieces — with and without the ending disclosed in advance. The spoiled versions were not rated worse. On average they were rated slightly better. The folk theory says the spoiler burned the story’s one asset. The ledger says the spoiler rescheduled a payment — you were surprised at the water cooler instead of in the theater — and the story kept every other way it pays. The data sides with the ledger.

The second campaign in the figure is the interesting one, because it names a feeling.

A campaign can lift expectation past anything the Actual will be able to fund. Every installment it pays on the way up is real — the thrill of the teaser is not counterfeit surprise; it is surprise, felt in full. But the identity is merciless about where that surprise came from. If E has been driven above A, the early payments were an advance on value the event does not contain, and the final installment arrives with a minus sign. Disappointment is not a separate emotion that follows hype the way a hangover follows wine. In this accounting it is the accounting: negative surprise, the ledger balancing, the repayment of exactly what was borrowed. ln(A/E) goes negative by precisely the amount the campaign overshot.

This is worth saying to my own industry with the bark still on. The launch culture of AI — demo reels cut for maximum lift, capabilities teased quarters ahead, every release “a new era” — is running the second ledger in the figure. The surprise those campaigns pay out is real, which is why they work. And the repayment is real, which is why the same crowd that gasped in March is contemptuous in June, and why each cycle of it spends something that does not renew: the audience’s willingness to let its expectation be lifted at all. A borrower who never repays does not get to keep borrowing. There is a reason the products that changed people most — the ones this series keeps returning to — arrived under-announced, with E still sitting at another chatbot.

Hype does not create surprise. It borrows surprise from the event, at interest, and disappointment is the collection.

If surprise were perfectly conserved, the scheduling industry would be a zero-margin business — every dollar of attention it collected early would be a dollar the event lost, and the customer would have no reason to prefer any schedule over any other. That is not the world. The trade persists because the identity holds exactly only for a bookkeeper who never forgets, never feels direction, and never changes size. You are not that bookkeeper, and the previous essays in this series have already named the three leaks.

1. Forgetting re-inflates the total

   The telescoping sum assumes each installment stays absorbed. It does not. Expectations depreciate — steeply at first, then slowly — so the lift a teaser buys in January has partly drained away by the June premiere, and the gap it closed can be sold again. The trailer re-pays a surprise the teaser already paid. That refund is the marketer’s margin, and it is why campaigns are drumbeats rather than single announcements: the schedule is tuned to the decay curve.

2. Installments carry an angle

   Surprise has an imaginary part — the direction you were facing when it landed. A spoiler pays the twist’s magnitude at the wrong angle: alone, out of context, aimed at nothing. The premiere would have paid the same magnitude perpendicular to two hours of built expectation. Equal payments, on the real axis; nothing alike as experiences. The ledger conserves the amount. It says nothing about what the amount does to you.

3. Small payments never cross the threshold

   Noise and Awe drew the line between surprise that spends you and surprise that resizes you. Awe needs a payment too large for the model that receives it. Ten installments of 0.14 nats sum to the same total as one payment of 1.39 — and change nothing, because none of them ever exceeded what your expectation could file. Conservation of amount is not conservation of effect. This is why fleeing spoilers is not superstition: you are not protecting the quantity of your surprise. You are protecting its concentration.

One more honest crack in the frame: the identity takes the Actual as fixed, and the world does not always oblige. Test screenings re-cut endings; roadmaps bend toward the demo. When A itself is edited to meet the expectations the campaign created, the endpoints move and the conservation is being gamed from the other side. That case deserves its own essay.

Here is the practical residue. For any event you are walking toward, the total surprise on offer was set the moment your expectation and that event’s reality diverged. You cannot raise the total by consuming more coverage of it — every preview, every leak, every speculative thread is an early withdrawal from a fixed account, taken at the wrong angle, in denominations too small to resize you, with a middleman collecting margin on your forgetting curve.

What you can choose is the schedule. The discipline the ledger recommends is almost embarrassingly old-fashioned: let expectation lie. Skip the trailer. Read the book cold. Let the quarter report itself. Not because anticipation is sinful, but because you are the only party in the transaction with an interest in taking your surprise in one payment, at full angle, large enough to clear the threshold — and every other party is paid to break it into installments.

The anticipation industry knows the total is conserved. That is exactly why it sells the schedule.