Principles

The standards, the reference code and the discussion happen in public. There is no private draft that ships before the public one.

A standard exists so a representative, human or AI, can parse it without asking a person what it means. Human-readable documentation comes second, and never replaces the machine-readable form.

Machine-readable does not mean opaque. A person should be able to read a document under one of these standards and understand what it claims, without needing a library to decode it.

Nothing published under Open Data should depend on any single provider staying online. A person, or a product, must be able to take their data to another provider without changing what it says.

A claim without evidence is self-asserted, and the standard says so plainly. Consumers should be able to check a claim's evidence and decide how much to trust it, rather than being told it is "verified".

Where an existing open standard already covers a piece of the problem, Open Data reuses it. New vocabulary is for genuinely new domains, not for reinventing plumbing.

A wider, simpler standard beats a narrower, cleverer one. Every proposed addition is tested against "can this be expressed with what already exists": if yes, it does not get added.

Domains differ, so the vocabulary grows. The primitives underneath (profile, relationship, claim, evidence) do not.

Beautiful writes the first draft because someone has to. The intent is for direction to move to the people and organisations who actually use each standard.