← Adapt

Trust & choices · A fictional story

The shelf has an opinion.

You want a film that feels different. Last weekend was all crime thrillers. Tonight’s home screen offers three more. It has a very clear idea of you. What do you open?

Where would you look?

One choice changes the story.

Pick a move. See what changes in the fictional example beside it.

Fictional home screen

  1. 1

    Harbor After Dark

    Crime thriller

  2. 2

    Platform Eleven

    Crime thriller

  3. 3

    The Copper Alibi

    Crime thriller

Read every outcome

First pick: First tile. First look.

Harbor After Dark gets opened. The top spot put another thriller in front of you before you searched for anything different. Where an option appears can shape whether you consider it.

Opened from top recommendation: Harbor After Dark (Crime thriller · opened); Platform Eleven · The Copper Alibi (Other recommendations still waiting)

Other genre: A different set of possibilities.

You browse beyond the original shelf. A comedy, a nature film and an animation enter your shortlist. Changing how you browse changes what reaches the decision.

New shelf · previous shelf: thrillers: The Very Long Lunch (Comedy); A Week Below the Tide (Nature); Spare Parts Parade (Animation)

Why these?: Past you is doing the choosing.

This demo puts recently watched genres first. Your wish for something different never entered its simple rule. A recommendation can fit your past and miss your present intention.

The authored rule behind the shelf: Recently watched genres first (Demo rule); Recent genre: crime thriller (Tonight’s wish: not included)

Borrow the suggestion. Keep the intention.

Recommendations can be useful shortcuts. Before letting the first screen become the whole shortlist, ask what you wanted today. What would you search for if no suggestions appeared?

How much of this is real?

This shelf uses an authored rule to make the effect visible. Real services use more complex systems, and users can get useful discoveries from them. The scene is not a claim that every recommendation traps people in familiar choices.

Netflix says it personalizes which rows appear, which titles appear within them and their order. This demo does not reproduce its algorithm. All titles here are fictional story props; no film, preference profile or recommendation service is queried.

All people, products, messages and outcomes in the story are invented. This is a scripted comparison, not a live AI call or a prediction. Your choices stay in page memory and reset on reload.

Netflix: How our recommendations system works. Primary description of personalized rows, titles and ordering. Reviewed October 12, 2026; background only, not the demo’s source of rankings.