**Not measured, and therefore not asserted.** Every other uncertainty constant in this package is a 90th percentile against a stated reference – `nar_nrcan_uncertainty_m()` is 150 m over the sample `data-raw/probe_geolocator.R` draws – and there is no equivalent run for this source yet. `data-raw/probe_osm.R` is the harness that would produce one; until it has been run over a national sample, a surviving answer reports `NA` rather than a number invented to look like the others.
This is also why [osm_geocode()] is not wired into [geocode()] as a tier. `uncertainty_m` is a column callers filter on, and a tier contributing `NA` to it would quietly make that filter mean something different depending on which tier answered.
What can be said without measuring: a surviving answer is a `place_rank` 30 object carrying its own house number, which is an address someone entered or imported rather than a position interpolated along a range. That is a different **kind** of answer from the geolocator's, not necessarily a better one – OSM's Canadian address coverage is uneven and concentrated in municipalities whose open data was imported.