01 — The friction
The collection changes. The need to make a selection doesn’t.
A streaming catalog, a shelf of physical media, and a carefully maintained digital library all have the same limitation: they wait for me to know what I want. Owning the media solves access. It does not solve the moment when I want to watch something but have no idea what that something is.
Network television and cable solve that problem by giving everyone the same programming. Streaming and personal collections offer far more control, but hand the work of programming every viewing session to the viewer.
I want the range and personal relevance of my own collection without having to manage it one title at a time. I also don’t want a stream made only of full-length shows and movies. Part of what makes programmed television feel alive is the texture between them: music videos, retro commercials, old game shows, network IDs, and bumpers.
The old Adult Swim and Nickelodeon bumpers can evoke an absurd amount of nostalgia in seconds. They weren’t merely filler; they gave a programming block its rhythm, era, and personality. A list can help me choose a title. It cannot create that continuity, pacing, surprise, or the strange pleasure of finding something I would not have deliberately selected.
Network television and cable choose for everyone. Streaming asks everyone to choose for themselves. What if a system could curate for an audience of one?
02 — The experiment
Build the channel, not the playlist.
At the center of the experiment is my own collection: a Plex server running from a NAS, housing personally owned media alongside archival files sourced from places such as the Internet Archive.
This is not an attempt to rebuild a cable lineup. I’m exploring the concept of a system that can program across physical media, digital files, and streaming sources—and natively intuit the vibe well enough to roll with it. Television episodes, movies, music videos, commercials, bumpers, and other media could become one continuous stream shaped for an individual viewer, with the system autonomously curating content on the fly.
The goal is not randomness. It is programming: making one thing flow naturally into the next without asking the viewer to keep reaching for the remote.
03 — The programming problem
Getting the vibe right takes more than matching genres.
Duration, pacing, and how much attention the next choice asks for.
Genre, era, mood, and the relationships between adjacent pieces.
Novelty and familiarity without repetition becoming predictable.
Short-form material, bumpers, and interruptions that make the stream feel intentional.
04 — The control problem
Should the system learn from me as I watch, treating skips, finishes, rewatches, and the things I simply let play as signals? Or should there be a brief preselection moment built around what I want from the interaction? Am I turning my brain off and relaxing, letting my mind wander as it’s prone to do? Am I seeking intellectual stimulation? Comfort? Nostalgia? The useful answer may live somewhere between asking and observing.
05 — Where it’s leading
As with most things in life, I couldn’t help but start asking questions.
How much does it need to know before it can get started? Could I give it a mood, a little context, and then let it figure out the rest? Could it learn the difference between “not this” and “not now”? And could it know me well enough to get it right without becoming so predictable that nothing surprises me?
The experiment continues…