Currently, an entity's list of claims is in some cases represented as a flat list, and in other cases grouped by property ID. Similar confusion exists for the snak lists that make up qualifiers and references.
To resolve this:
- There needs to be a serializer option for toggeling grouping (on different levels)
- Unserializers need to support both forms everywhere
- The overall JSON documentation needs to document both forms
- API modules as well as Special:EntityData need to clearly document which form is used when, where and for what.
- API modules may provide options to specify the desired output mode for claim lists and snak lists.