Knowledge Machine

Reading easily ends as consumption. A few days later, I may no longer know where an idea came from or what I actually understood. Knowledge Machine is a personal knowledge system that keeps a way back to the source and a procedure for checking understanding, so what I read does not simply disappear.

The source of truth is knowledge-machine-database. I keep material and notes in an Obsidian vault, while coding agents manage files and document states according to defined rules. This blog reads only what is published and builds static pages from it.

sourceMaterial I can check again: links, videos, books, excerpts, and conversations.the source and way back, never a copied original

referenceOne Source organized into a reusable document of summaries, quotes, questions, and notes.a space to test understanding · not published

knowledgeA short piece that explains a source’s facts and context, with opinion kept light.published · sources linked in endnotes

articleA piece that connects sources to make a complex argument and include experience and opinion.published · also uses the journal category

Why build it

Reading a lot without retaining a thought is not only a problem of missing notes. The path back to the material, what I understood, and the questions I still have often live apart. Knowledge Machine brings these three things into one document structure. Its purpose is not to publish more pieces, but to make it possible to trace what I read, what I understood, and which writing came from it.

What makes knowledge

A summary is only a starting point. Every reference starts unread and turns read only when a question and answer or a note is actually attached. The system favors leaving behind what I understood over counting what I finished.

People and agents

I choose and read the material, then decide what I understood. Agents structure the material and maintain filenames, schemas, read and publication states. They do not fill in what a source does not say or turn unexpressed thoughts into notes.

Publication comes last

References stay as documents for working through understanding. Only knowledge and articles appear on the blog. Both keep their sources in endnotes, while articles can carry more complex claims and experience. Each document’s frontmatter decides whether it is published.

Source and blog

knowledge-machine-database is the source of truth, and this blog is its rendered result. Each build reflects the repository’s latest contents without keeping a separate copy of the writing. The structure keeps recording and publishing in one place.