Why map-first?

Existing bat-recording tools are mostly spreadsheet-shaped: a table of files, a spectrogram viewer, batch analysis. That's the right shape for processing a big survey dataset, but it's the wrong shape for the simpler, more common thing this project actually set out to do: make it genuinely pleasant to revisit a night out with a detector — where you were, what you heard, and when.

Location and time are how a night out is naturally remembered, so the map is the primary view rather than a list. Clicking around the map to see what turned up where is meant to feel like browsing a journal, not querying a database.

That said, "pleasant to browse" and "honest data" aren't in tension here. Every identification keeps its source and its own claim (see The classifier model) rather than collapsing straight to a single label, and recordings group into sessions and sites rather than sitting as an unstructured pile of files — so the same data that makes a map fun to click around also holds up if you later want to treat it as real survey documentation.