What this is
Show Type describes television instead of ranking it. Every show here is scored 0–10 on the same eight axes — how propulsive it is, how big its world is, how much it asks of you — which turns each show into a point in an eight-dimensional space. Shows that sit near each other feel alike, even when they share no genre.
That’s the useful part: “more like this” usually means more of a certain texture, not more of a category. A crime drama and a hospital drama can be near-neighbours here.
Getting started
- Explore — search a show you already know. You’ll see its profile across the eight axes, why it scored that way, and its nearest neighbours. Start with something you love and see who its neighbours are.
- For You — add a handful of shows and mark each one loved, liked, never interested, or started & stopped. Recommendations are drawn from the centre of what you liked, pushed away from what you didn’t. Five or six shows is enough to get something useful.
- Browse — skip the ratings and dial in a profile directly. Drag the sliders to say what you’re in the mood for — low propulsion, high verisimilitude, whatever — and the catalogue narrows as you go.
The eight axes
Each is scored 0–10. None of them means “good” — they only say where a show sits.
Good to know
- The axes are descriptive, not evaluative. A 2 is not worse than a 9. A show scoring 2 on Density isn’t dumb, it just doesn’t demand much; a 10 isn’t smart, it just won’t hold your hand.
- Quality is tracked separately, as the execution score — the Q on a row, or Execution n/10 on a show. That one is a judgement; the eight axes are not.
- Δ is distance in the eight-axis space. Smaller means a closer match to the show or profile you started from.
- Nothing leaves your browser. Your ratings, reasons, and watchlist are stored on this device only. No account, no upload, no tracking — and clearing your browser data clears them.
- The scores are model-generated. Every show was characterised by Claude against a written rubric, not by critics or viewer votes. Treat them as a consistent reading rather than a verdict — and expect the occasional obscure show to be read poorly.
Open source
The code, the scoring rubric, and the full dataset are on GitHub, MIT-licensed — github.com/roskosevan-code/showtype. If an axis reads wrong to you, the rubric is the thing to argue with.