Standards or vocabulary: what is the difference?

We publish two things that sound alike and are not: a register of standards and a vocabulary. People reasonably ask which is which. The short answer:

• The register is a list of rulebooks. Most are written by someone else, and we say that we follow them.
• The vocabulary is a list of words. They are ours, and we publish them because no one else has.

The register

At standards.bundu.org/register you will find 78 standards, such as UCUM for units of measure, ISO 3166 for country codes, ISO 8601 for dates, WGS 84 for coordinates and OpenAPI for describing an API. For each one we record:

• what it governs;
• the version or the commitment we make to it;
• who stewards it;
• why and when we decided to adopt it;
• whether it is adopted or still proposed.

The register answers the question which rulebook do we follow for this? When a team needs to store a temperature, a country or a date, the register tells them the answer before they invent one.

The vocabulary

At vocab.bundu.org there are 39 terms, 38 of them stable and one still testing. They are defined in RDF, the same family of formats that schema.org uses, under the namespace https://vocab.bundu.org/. The home page describes them as RDF terms for concepts schema.org has no word for.

Examples:

• Family and Community, which describe how people belong together in the places we work;
• verificationTier, an ordered ladder from 0 (unverified) to 4 (licensed) rather than a yes-or-no flag, because a bare verified cannot tell a claimed place from one whose registration documents have been reviewed;
• ubuntuScore, communityVouches and isInformal, for trust and recognition, including of places that are not formally registered;
• twelve rules of the Bundu AI Guardrails Standard, from BAGS-01 to BAGS-12.

The vocabulary answers a different question: what do we call this, precisely, when nobody else has a word?

Terms are a permanent contract. A release may add a term or promote a testing term to stable. It does not remove a term or change what a stable term means, because other people's data may already point at it.

How they connect

The two are joined by one rule: if a standard does not exist, we build one and publish it. Otherwise we adopt the global standard.

1. We look for an existing standard. If there is one, it goes in the register as adopted, and we convert whatever data we receive into its form.
2. If nothing exists, we write the standard ourselves and publish it in the same repository.
3. The new words that standard needs become terms in the vocabulary.

The AI Guardrails Standard shows all three: the standard itself is published, its twelve rules are vocabulary terms, and the register lists it as one we hold.

The vocabulary is itself a register entry, so the register says we adopt it too.

A quick test

If you are asking...	It belongs in
Which unit, code, format or protocol do we use?	the register
Who maintains it, and which version do we follow?	the register
What do we call this concept in our data?	the vocabulary
Is this word stable enough to build on?	the vocabulary

A useful example is a weather reading. The unit of the number (degrees Celsius, written Cel) comes from UCUM, which is a standard in the register. The sort of place the reading belongs to, and how far we trust whoever reported it, use words from the vocabulary.

Why we keep them apart

If we mixed them, a reader could not tell which rules are the world's and which are ours. Keeping them apart makes that clear: the register is where we say we did not invent this, and the vocabulary is where we say we had to. Both are public, and both can be argued with.

To look at them yourself, see How to see what we have built.