The hard part after launch wasn't the map — it was provenance

One thing KoreAves made obvious after launch: the hardest part of a bird atlas wasn't rendering regions. It was deciding what a clean interface is allowed to imply.

A record is not abundance. Missing records are not absence. A matched name is not automatically the same taxon concept. And a modern bird photo can be easier to find than one whose rights and provenance are actually auditable.

That last problem became a guest piece on 10,000 Birds:
https://www.10000birds.com/the-b...

I'm now treating source-to-concept reconciliation and provenance as first-class product data rather than metadata added at the end. For evidence-heavy products, that distinction has turned out to matter more than I expected.