Related: sources · notes · metadata · Published Pieces

Mental Models Are Media Primitives

A person can publish a way of thinking, not just a thought.

A mental model becomes much more powerful when it stops being private advice and becomes a reusable operation.

“Invert, always invert” is not just a phrase. It is a way to process reality. Apply it to a business and you get failure modes. Apply it to a personal habit and you get self-sabotage maps. Apply it to software architecture and you get hidden coupling, corrupted invariants, and routes to collapse. Apply it to politics and you get incentives that produce the opposite of stated goals.

That is what makes it powerful. It travels across domains.

Most mental models work like this. They are functions over thought. They take one representation and produce another. First principles reasoning decomposes a situation into fundamentals. Opportunity cost maps an action into the alternatives it excludes. Chesterton’s fence maps a proposed reform into the history of the existing constraint. Girardian mimetic desire maps preference into imitation and rivalry. Marxian analysis maps social life into class, production, ownership, and exploitation. McLuhan maps media into sensory and institutional form. Systems thinking maps objects into feedback loops.

A mental model is not merely something one believes. It is something one applies.

That makes mental models media primitives.

A media primitive is a basic operation that changes what can be produced, distributed, and perceived. The hyperlink made texts traversable. The hashtag made topics aggregate. The retweet made redistribution frictionless. The quote tweet made commentary attach to circulation. The podcast feed made serialized audio portable. The embedding made external objects appear inside other media.

Cognitive transforms are the next primitive.

A transform says: take this artifact and look at it this way.

Extract the claims. Invert the plan. Find the strongest objection. Identify missing priors. Map the contradictions. Generate the historical analogy. Separate evidence from interpretation. Compare to another tradition. Turn it into radio. Turn it into a decision memo. Turn it into a research agenda.

In ordinary chat, these are prompts. In Choir, they become objects.

That shift matters. A prompt disappears into the session. A transform persists. It can be named, reused, improved, composed, cited, and rewarded. A person can publish a way of thinking, not just a thought.

This changes what it means to contribute to a media platform.

The internet has trained people to think of contribution as content: posts, clips, comments, essays, podcasts, videos. But some of the most valuable contributions are not content at all. They are lenses. They are ways to produce better content, better critique, better research, better argument, better correction.

A good editor has transforms. A good lawyer has transforms. A good anthropologist has transforms. A good security engineer has transforms. A good mother has transforms. A good organizer has transforms. A good artist has transforms. They know how to look.

Choir can make those ways of looking executable.

Once encoded, a transform can be applied by agents across artifacts. It can help generate an automatic radio segment. It can improve search. It can critique a published vtext. It can help users learn. It can be combined with other transforms. It can acquire a track record.

A transform that repeatedly helps future work becomes intellectual property.

Not because it excludes others from using it, but because its use can be traced. If an authored transform helps produce valuable artifacts, the system can cite it. If the transform becomes central to a line of inquiry, its author can receive credit and reward. This is protocol-native IP at the level of cognition itself.

That is a richer vision than creator monetization.

Creator monetization pays people for audience attention. Cognitive-transform IP pays people for reusable mental machinery. The value is not merely that someone watched or listened. The value is that a way of thinking made later thought better.

This also resists monoculture. The default assistant model tends to produce one house style: helpful, polite, moderate, flattened, procedural. A platform built around cognitive transforms can preserve many styles of cognition. Different cultures and disciplines can encode different transforms. Users can apply them, compare them, modify them, and cite them.

That is how pluralism becomes operational.

Not “everyone has a voice” in the abstract.

Everyone can contribute a way of seeing.