SYM42 discussion: DwC-DP profiles, deriving occurrences, ABCD's future and system boundaries

What this was about

Wieczorek answered a chat question on registering DwC-DP profiles (e.g. a GBIF profile requiring occurrences): none exist yet, but community-created, community-vetted profiles could become standard products. Grieb discussed who should derive occurrences, and a GBIF participant asked what rules GBIF should follow; the panel called for community-driven, rule-based conventions. Weibulat said German data centres must still decide on ABCD's future. Dupont confirmed RECODE's strong API. Yoder noted agentic pull requests from students and the public, and Macklin reflected on how far CMSs should manage links versus rely on external services. Participants suggested core-versus-plugin architectures, and Yoder warned most links die within ten years.

Why it matters. It captures open design questions for the DwC-DP transition and for CMS architecture as AI-assisted development and linked services grow.

In the room

• There are no DwC-DP profiles now because early feedback asked for a single model, but communities have since asked for tailored profiles.
• The latest Darwin Core public review incorporated GBIF cores and simple extensions (measurement or fact, identification history, alternative identifiers, Humboldt, chronometric age) into the standard; OBIS and GGBN extensions await conversations.
• DINA would prefer to publish without occurrences and let GBIF derive them if technically feasible.
• GBIF asked whether derivation should use explicit flags or rule-based logic (e.g. event + identification + material).
• German NFDI data centres still use ABCD and must decide whether to switch.
• Agentic coding creates a flood of pull requests; semantic-layer changes need strict control, UI changes less so.
• Suggestion to separate core from blessed plug-ins to manage community contributions.

Notable moments