Web of the Future: User Model Specification

WOTF is an open specification for a small, portable user model that describes how someone wants to consume content and what they’re interested in. Readablog is the first implementation.

What WOTF is

The whole user model is a single binary file. The minimum size is 38 bytes. A fully-loaded one with 64 interest strands is still only about 64 KB. Small enough to send with every API request, portable enough to travel between services.

What’s in the format

Three sections: a Fingerprint (privacy controls and metadata), a Style Vector (how you want content delivered), and a list of Embeddings (what you’re interested in). See How It Works for the friendly walkthrough.

Built to be open

The spec defines a data shape and a set of privacy controls. It says nothing about who provides recommendations, who hosts the data, or which app you use. Any reader app and any content service can implement either side, and the parts will still fit together.

Versioning and the model registry

The spec includes a fingerprint version, a user model schema version, and a model registry that identifies which embedding model produced the vector strands. WOTF is designed to evolve without breaking older models.

The full spec is being prepared for public release. Once it lands, we’ll link to it here. In the meantime, send questions to hello@reada.blog or head back to the Home page.