The Long Winter, and Why We Stop Watching series.
Winter in Washington this year was the kind that settles in and refuses to leave. Grey mornings, frozen sidewalks, the light gone by five. On nights like that you don’t go out. You make coffee, put something on, and let a series carry you through the dark. So I watched a lot. Crime, mystery, suspense: detectives on submarines, thieves in Paris, four investigators chasing the same body across four different centuries.
Somewhere in that binge I noticed something strange about myself. I abandoned series I knew were excellent — three episodes in, gone — and I finished others I would never call good. It wasn’t about quality. I was leaving good stories and staying with weak ones, and I couldn’t explain the pattern. Something other than taste was deciding whether I kept going.
The question followed me out the door. One cold Saturday morning I went walking on the trails near Bethesda, half to clear my head and half because I had been sitting too long. Somewhere along the path, the shape of an answer arrived. What if leaving a series is not a single verdict — “I liked it” or “I didn’t” — but a decision you make again and again, episode after episode? Each time the credits roll, you either continue or you don’t. That is not a preference. That is a survival process: a probability of walking away, given that you had made it that far. And the moment I saw it that way, I realized this was not a detour from my day job at all. It was my day job in a different costume.
The paper that came out of that walk is now published in Poetics, a journal that studies culture — art, media, literature, taste — with the tools of social science. Most people have never heard of it, but it is exactly the right home for a question like this one: how a story is built, and how that structure meets the way we actually watch.
The paper treats a season of television as what it really is: an object that releases information over time. Suspense, after all, is just the careful control of what you are allowed to know, and when. So I built a small set of tools to describe how a story does that, using four measurable dimensions.
Get Dr. Diego Vallarino’s stories in your inbox
Join Medium for free to get updates from this writer.
The first is non-linearity: how scrambled the telling is compared to the real order of events. A procedural where investigation, revelation, and resolution arrive in order sits near zero. A series like Bodies, with four timelines pointing at one murder, sits at the top. The second is latency: how long a question stays open before the story answers it — the distance, in episodes, between a clue and its payoff. The third is surprise, and here I borrowed from information theory. A twist is surprising not because it is loud, but because it was unlikely given everything before it. A betrayal you could not have predicted scores high; an expected next step scores low. The fourth is closure: how many of the threads a season opens actually get tied off by the end.
The interesting part is not the measuring. It is that each of these is a knob with a hidden cost. Non-linear brilliance buys you surprise, but it charges you effort — the work of rebuilding the story in your head — and that bill comes due exactly when you are watching in pieces across a busy week, coming back after days away, half-remembering who did what. Long latency sustains suspense, but it is a bet that you will still be there when the payoff finally lands. Too much closure removes your reason to return; too little leaves you frustrated. And surprise only works while it stays clear: a twist with no connection to what came before is not suspense, it is just noise. The framework does not tell storytellers what to do. It makes their trade-offs visible.
To test whether any of this held up, I put together a small, deliberately mixed group of ten European and global crime and mystery series. Then I modeled abandonment as a hazard — episode by episode, the risk of leaving — next to a comparison of which structures keep viewers better than others. One result stayed with me. The British institutional dramas — The Long Shadow, Vigil, The Gold — held their audiences with no time tricks at all. Their coherence, their steady procedure, the weight of real events, did the work that a scrambled timeline does elsewhere. There is more than one way to keep someone in the room.
Here is where it connects back to the rest of my research, and why I could not let the idea go. In my economic work I use the same machinery — event-history and hazard models — to ask when a firm fails, when a shock spreads through a network, when a system that has held for years suddenly gives way. The same “surprise” measure appears when I study information in economic structures. And the deepest question under all of it — when does something stable suddenly abandon its state? — is the sandpile question that runs through everything I do. A season of detective television turned out to be a small, human-sized model of that: tension building grain by grain, episode by episode, until someone walks away. I went looking for why I stopped watching, and found the same shape I had been studying in fragile economies all along.
One caution matters enough to end on. Persistence is not worth. Finishing a series does not make it good; leaving one does not make it a failure. Some of the finest things I watched this winter I left unfinished, and I am not sorry. The framework is built exactly to hold those two ideas apart — to measure continuation without pretending it measures value. A story can be abandoned by everyone and still be the best thing on the platform.
We now live inside fragmented attention — second screens, easy exits, a hundred other things one tap away. Storytellers are quietly negotiating with that condition, whether they name it or not, and I think it is leaving marks on the form of stories themselves. This paper is my attempt to make that negotiation visible. It began, as these things do, with a cold season, too many detective shows, and a walk that lasted a good deal longer than I had planned.
