Surprise

Information is one over probability, so the rarer something is, the more informative it is. Claude Shannon got there first and put a logarithm on it.² Something you were certain of tells you nothing; something you thought impossible tells you a lot. The moment a thing clicks is exactly this: something rare, true, and outside the picture of the world you already had. Surprise also lives in the gap between the thing and what you already know, so the same introductory video is an evening well spent for a beginner and a wasted one for the person who works in the field.

Fidelity

Surprise alone would rank static as the best content on earth, since nothing is less predictable than noise. So the second term asks whether what you took out is true, whether updating on it leaves you better at predicting the world. Fiction only has to be faithful to itself: a film can have magic in it, as long as its characters still make sense.³ The useful part is that fidelity can go negative. A confident, coherent falsehood is worse than nothing, because you learn it. So the most damaging thing you can consume is the one that is gripping, hands you a tidy framework and is wrong, and that is a fair description of a good conspiracy theory.

Extraction

How much you can get out. Hold the work fixed and the reader decides: a good enough reader gets something from a soap opera, and someone who stops at the first level gets a list of circles out of the Divine Comedy. It is the one term that belongs to you rather than to the work.

Cost

Time is the obvious half. The other half is pleasure, and it is the scarcer of the two: for practical purposes you can always earn more money, but you cannot earn more pleasure.

And the price compounds. A three-minute short charges you three minutes, then charges you again by raising the bar for whatever comes next, so the book you meant to open afterwards now feels like nothing. That is the answer to the TikTok puzzle. The people who built it are excellent at one term, and it is the cost, which they have made enormous and almost invisible.

The equation has one blind spot. A comfort rewatch or ten minutes of small talk carries almost no information and is still worth doing, because it belongs to rest and to connection, which run on a different scale. Score them with this one and you end up asking a painting to justify itself. The equation is for the hour you were going to spend on information anyway.

There is a catch inside it, too. You can only know how surprising something was after you have consumed it, because knowing in advance would mean you had already been told. So every choice of what to watch next is a guess, and whoever makes the guess decides most of the outcome. Usually that is an algorithm. Deleting the app does less than you would hope: the half hour you used to spend in it is still there, and something fills it. What seems to work is a queue, long enough and made only of things you already judged worth it, so that something better is always one tap away and the algorithm never gets a turn.

So before asking whether the next thing is worth your hour, ask who picked it.