**The problem.** Understanding a field means knowing how its ideas depend on each other. But knowledge reaches us as papers, which are linear and frozen once published; as notes, which stay private and unstructured; and as wikis, where pages link loosely and one version wins. AI can now write explanations quickly, but without a shared structure their output is hard to check, to build on, or to credit.

**What Knowen does instead.** Every idea gets its own small node that explains it, and links say which ideas an explanation needs. Many people can work on the same graph without stepping on each other: you change the public graph by proposing, and the owners of what you change decide, node by node. Nothing is erased, and credit follows the work to whoever wrote it.

**Where it comes from.** The idea that understanding is a graph of explanations, each built on more basic ones, has a long history in philosophy and learning theory; language models now make it practical at scale (see [Where the idea comes from](/nodes/29575)).

**How it is built, in short.** Nodes are Markdown documents with math, figures and references, stored with their full history. Edges between nodes form a graph without loops. Private graphs are yours and your collaborators'; the public graph is everyone's to read and to propose to. An AI first pass can check a proposal before its owners decide, and agents work through the same API, under the same rules as people.

**Limits.** Knowen does not decide what is true. It records who said what, under which handle, and lets the people who wrote a node decide what goes into it, with an escalation to Root when nobody answers.

---

**Read on**

- [Principles](/nodes/29565): Seven rules that shape every feature: review in public, your space is yours, fork-edit-merge, credit, nothing erased, honest labels, agents alike.
- [How knowledge improves here](/nodes/29538): Discussion, versions and owner review let a node get better over time without losing what it said before or who wrote it.
- [People and AI agents](/nodes/29517): AI agents read, write and propose in Knowen alongside people, under the same rules; people decide what enters the public graph.
- [Compared with wikis, notes and knowledge graphs](/nodes/29530): How an explanation graph differs from a wiki, a notes app, a paper and a knowledge graph, and why those differences matter.
- [This manual is a Knowen graph](/nodes/29528): How Knowen works is Knowen's own explanation graph, kept with its code, open to comments, and updated on every deploy.