downgrade your priors

why do we need to know things?

I’ll never forget that fourth Knicks game. I was sitting in a bar with a group of quantitative finance traders, and the mood was suboptimal. With less than ten minutes on the clock, the Spurs were 29 points ahead.

A few of us were jokingly monitoring the situation on Polymarket. The prediction market gave the Knicks a 99.6 percent chance of losing. One of my friends decided he would have higher expected value in going home, so he left the watch party.

Then, miraculously, the tide shifted. The Knicks came back, possessed with electric ferocity, to win the game by a single point. At the buzzer, the entire bar was roaring at the top of its lungs, jumping in ecstasy, drunkenly high-fiving complete strangers. I can’t describe the primordial passion I felt in that moment. It was a barbaric, religious fervor I’ve only experienced a few times in my life.

Walking home through the revelry, my thoughts drifted back to the prediction market. What was the point? The main reason people watch sports is because they don’t know what’s going to happen. If the game were completely predictable, we would all follow my friend’s lead and skip the fourth quarter. Instead, we stay watching for that one-in-a-hundred hope of things going unbelievably right.

In his book The Uncontrollability of the World, the sociologist Hartmut Rosa connects sports-watching to the giddiness of a first snowfall. No matter how advanced our weather forecasts get, there’s still a delightful magic in letting that initial snowflake land on your tongue. It is the unexpected “intrusion of a new reality,” reawakening your childish astonishment in the surrounding world.

The inexorable project of modernity is to make itself legible, manageable, and profitable. With each advance in categorization, however, we make ourselves a little more boring. Ski resorts create artificial snow before the first flurries arrive; Polymarket tells you when to leave the basketball game.

Without coincidences, we numb ourselves into mathematical indifference—until a sufficiently uncontrollable event wakes us up from our sluggish subroutines. In these moments, we have the chance to realize how little we actually knew, and look around with renewed wonder.

But what if this could happen more often? I love faceplanting into the perfect sentence. I love when my friend invites me to a random movie and I have no idea what’s about to happen. That’s how you write poetry and fall in love with a cult classic. And yet ChatGPT always “knows” the next token, while TikTok always “knows” the next video I want to see. The prediction becomes reality, and I forget to notice the snowflakes.

What good is the “wisdom of the crowds” when I want to feel stupid? What good is predictive text when I want to write unpredictably?

The quant traders talk a lot about “updating their priors”—adding new information to create a more accurate prediction of the world. I find it interesting that their priors are never complete. You always have to collect new data, change more parameters, slowly refine the model as the world evolves.

Maybe there’s something to be said for downgrading our priors as well. So many of our best moments come out of inherently incomplete knowledge. It might be better to choose what to know, and what not to know—and I personally don’t want my basketball games spoiled for me.