The Exchange Rate on Bad News

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 the gap as a log — S = ln(A/E) — and the recent essays pulled three consequences out of that reading: the log was never a choice, the total is conserved between endpoints, and the Actual is not always deaf to the Expectation. Today I want the fourth consequence, the one carried by the smallest mark on the page: the sign.

The log is antisymmetric. Swap the endpoints and the magnitude survives while the sign flips:

The equation, in other words, certifies parity. A beat and a miss of equal ratio are the same size of event; the windfall and the shortfall in this morning’s scene are the same distance from what you expected. If attention paid for surprise at face value, the two notifications would have had an equal claim on your thumb.

They did not, and this is not a flaw in you personally. The teller at the attention window honors the magnitude and then applies an exchange rate — and the rate depends on the sign.

The premium on the minus sign is one of the most replicated findings in the behavioral sciences, measured by four different instruments over four decades.

The oldest instrument is choice. When Tversky and Kahneman fit their cumulative prospect theory to gambling decisions in 1992, the median subject weighed a loss about 2.25 times as heavily as an equal gain — the loss-aversion coefficient, and the closest thing the literature has to a posted price. The broadest instrument is the literature itself: Baumeister and colleagues’ 2001 review, titled with unusual candor Bad Is Stronger Than Good, walked domain by domain — feedback, memory, relationships, first impressions, learning — and found the asymmetry in nearly all of them. The most physical instrument is skin: in 2019, Soroka, Fournier, and Nir wired up 1,156 people across seventeen countries on six continents and played them television news; on average, skin conductance rose more for the negative stories. And the largest instrument is the click. In 2023, Robertson and colleagues analyzed the Upworthy archive — 22,743 randomized headline experiments, about 105,000 headlines, roughly 370 million impressions — and found that each additional negative word in a headline raised its click-through rate by about 2.3 percent, while each positive word lowered it by about 1 percent. Relative changes, small per word, measured with the cruelest precision the internet allows: the A/B test.

The math prices the two signs at par. The nervous system quotes a spread — and everything downstream of the nervous system trades on it.

Here is the thing about a mispriced currency: nobody who profits from it needs to understand it. Anyone paid in attention will find the premium by feel, the way water finds a crack.

Television found it first and said it out loud. If it bleeds, it leads was a local-news assignment policy before it was a proverb — a Boston news director was quoted saying it as early as 1982, and Eric Pooley’s 1989 New York magazine piece made it famous as an indictment. Notice what the phrase actually is: a trading rule. A newsroom does not need the loss-aversion coefficient; it has the overnight ratings, and the ratings had already run the experiment. The doom-forward front page is not an editorial failure, whatever the editors’ souls are doing. It is a market clearing at the posted rate.

Then the feed automated the discovery. A recommender system trained on engagement is a machine for finding whatever your attention overpays for, and it learns your personal exchange rate without anyone writing it down — tirelessly, at the resolution of the individual scroll. Nobody instructed the algorithms of the last fifteen years to prefer alarm. They were told to maximize a number, and the number was denominated in a currency that pays double for minus signs. The feed is the loss-aversion coefficient, industrialized.

It is tempting to file the asymmetry as a bug, and that would be a misreading. The premium was correct where it was set.

The rate is not really a price on badness. It is a price on irreversibility. In the world that tuned your nervous system, the two signs were not mirror twins in their consequences, whatever the log said about their magnitudes: you could survive missing a windfall every day of your life, and you could survive missing a predator roughly once. Losses cliffed; gains crawled. And the news that reached you was actionable almost by definition, because it arrived from within earshot — what you could hear about, you could usually still run from, fight, or fix. Under those two conditions, paying double per negative nat is not a bias. It is actuarially sound.

The physiology paper carries a second finding that matters here: the average conceals the person. Country explained almost none of the variance in Soroka’s data, and a substantial minority of individuals reacted more strongly to the positive stories. The rate is not a constant of the species; it is a personal setting. It was just never, for anyone, an adjustable one.

Now run the two conditions against the world the feed actually delivers. The negative surprise arriving on your screen this morning is overwhelmingly not yours: not your fire, not your collapse, not your war. It is not actionable by you, and it is not reversible by you, and it is frequently not even proximate to anything you will ever touch. The ancestral link between hearing about a loss and being able to do something about it — the link that made the premium sound — has been severed by the same technology that made the delivery free. You are paying a survival premium, in doubled attention, on losses you cannot suffer and cannot prevent. The spread between the evolutionary rate and the actionable rate is the arbitrageur’s entire margin. Doomscrolling is what it feels like to be the counterparty.

Yesterday’s essay argued that as delegation approaches everything, one sentence remains yours to author: the condition under which the machine may spend your attention, priced in surprise. Today’s essay adds an amendment. Denominate the wake condition in surprise — but set the exchange rate yourself, because the default is the ancestral one, and every system trained on your engagement will inherit it. Left alone, your stack learns your λ and wakes you for minus signs, twice as eagerly, forever. Instructed, it can quote the rate the old world never let you post: wake me for what I can act on, whichever sign it carries. Batch the rest — both signs — for morning.

This is, as far as I can tell, new. Your subconscious filter was tuned by selection and does not take requests. The synthetic one — the layer of machines now standing between the world and your attention — takes dictation. For the first time in the species’ run, the exchange rate on bad news is something a person can author rather than inherit.

The premium may still be right. Not every modern loss has lost its cliff: reputation, capital, and trust are still asymmetric — gains crawl and losses cliff, exactly as they did on the savanna — and a person whose stack discounts minus signs is a person building the padded room the placebo essay warned about, an expectation protected from the world at the price of ever learning from it.

So state the failure condition plainly. If, five years from now, the people who set their own exchange rate are systematically blindsided — miss the layoff forming two levels up, the client quietly churning, the symptom that was a sentence in a feed they filtered — then the old rate was the right rate, and this essay mistook a working alarm for a stuck one. I don’t believe that is how it goes, because the instruction was never less bad news; it was price by what I can act on. But if that distinction cannot survive contact with real filters — if actionability turns out to be unjudgeable in advance — then the safe setting is the ancient one, and the arbitrageurs keep their margin honestly.

Tomorrow at 6:12 the notifications will arrive again, and something will decide which one you see first. The question this series keeps arriving at is no longer what your attention costs. It is who sets the price — now that, for the first time, the pen is in your hand.