Interface Design / Music Discovery

A Danceability Dial for Spotify

Spotify knows an extraordinary amount about music. Hidden inside their API is a danceability score for every track, a numerical value that captures something real about how a piece of music moves in the body. But that knowledge is locked away from the listener. The interface you get is a recommendation engine that gives you more of what you already know.

This is an experiment in building a different kind of interface. A danceability dial. Not a search for an artist or a track name but a way to use a song you love as a seed and ask: give me something more like this, but more danceable. Or less. Or different in a direction I cannot yet name.

The provocation underneath it is simple: optimization algorithms know a great deal that their interfaces deliberately conceal. What would discovery feel like if that knowledge were put in the hands of the listener rather than used to keep them inside a known world?

Status

Early prototype. Vibecoded a working proof of concept. Developing further.

Metaphor / Systems Thinking

The Internet as Modular Synthesis

A modular synthesizer has no fixed signal path. You patch it yourself. You decide what flows into what, how frequencies modulate each other, what emerges from the combination. The same oscillator becomes something entirely different depending on what you connect it to. The instrument is not a product. It is a set of possibilities waiting for a player.

What would it mean to think about the internet the same way? Not as a set of destinations to visit or services to consume, but as an instrument to be tuned and played. A system where you decide the signal path. Where information from one source modulates another. Where the mashup is not a metaphor but a method.

Most of what we call the internet today is a fixed patch. Someone else decided the signal path. You move through it but you do not play it. This is an exploration of what it would mean to change that.

Status

Conceptual. Looking for collaborators and contexts to develop it further.

AI / Curiosity Systems

Daydreaming Bots

Most AI tools are task machines. You ask, they answer. Picasso said of early computers: they are useless, they can only give you answers. The observation is even more pointed now. We have built extraordinarily capable systems and pointed them almost entirely at efficiency.

This is an experiment in a different proposition. A bot that daydreams on your behalf. You bring it something you are curious about, a project, a question, a fragment of interest. It translates that curiosity into queries and sends them out to servers across the world, tapping into Japanese, Korean, and other non-English language webs to find what the English-speaking internet has not indexed or considered. It collects what it finds, looks for unexpected connections, and returns.

The output is not an answer. It is an RSS feed of links worth exploring. A return from a daydream. Something to follow into the world rather than a conclusion to accept.

The inspiration comes from Johann at Dobbin and Gwern's writing on LLM daydreaming and the idea that the most valuable thing a thinking system can do may not be to optimize but to wander with intention.

Status

Conceptual and in early development. Deeply interested in collaborators who want to build this.