Findings · updated live
What bags actually say
Every number below comes from a real bag somebody photographed. Instead of arguing about what a coffee record should contain, this counts what labels print.
Fields people record
Ranked by how many different people used each one — one enthusiast submitting thirty bags shouldn't outrank five people independently seeing the same thing. proposed marks a field the schema has no term for yet.
Distinct values answers a question people usually argue about
instead of measuring: does a field like roaster need to become a real
entity rather than a string? If forty bags produce thirty-eight distinct roaster
spellings, identity is fragmenting and doing real work. If they produce twelve, it
isn't a problem worth modelling yet.
| Field | People | Bags | Distinct values | Most on one bag |
|---|
Values the lists are missing
Read straight off a label by someone who found the dropdown had no room for it.
These are already applied to sh:in in the shape below.
Objections to the current fields
Raised while recording a specific bag, so each one is traceable to a photograph.
| Field | Objection | People | Times |
|---|
The shape this adds up to
Generated, not written by hand — this is what ships as
grindograph-shapes.ttl. It is advisory: a bag is
never rejected for violating it, because a bag that violates it is the evidence
that it's wrong.
Vocabulary: /vocab · /shapes · /context.jsonld