I haven’t had time to investigate this thoroughly (yet). I did some poking around in the places file, and I did find some examples of crudely drawn polygons for places I’m reasonably familiar with (I’ve drawn my own crude polygons for them).
In one case, I found two different versions of one place, both created by the same iNat user. The earlier attempt doesn’t have the correct shape, but at least it roughly overlaps the real correct boundaries. The later attempt is a better approximation of the overall shape of the boundaries, but it is inflated in size by about double. The depressing part is that the user is someone with whom I have passing familiarity, and kind of expected more from.
Clearly, these two place_ids would NOT be useful for assigning location descriptors. I’m wondering - is there any mechanism by which these places could be used by iNat to assign lat/long to an observation at the time of submission? For example, a user (not necessarily the person who created the place) is entering an observation into iNat. For the location, they enter the place name, and iNat assigns the centroid of the polygon as the lat/long. Can this happen?
I tried to submit an observation to test it out. When I type the place name into location, iNat uses the lat/long of the label for this place in google maps. This is the trailhead at the edge of the property, and doesn’t appear to have any relationship to the problematic hand-drawn polygons.
So unless I’m missing something, it doesn’t appear that one can use place_ids to geolocate observations at submission time. That would be a relief. iNat’s assignment of lat/long based on a place name can be problematic in some cases. In the past, I was often frustrated by certain observers who appeared to persistently set their lat/long at inappropriate places, which often coincided with the exact spot where google maps happened to display the park name. I assumed the users were simply clicking on the park name on the map, but now I realize that it may be that they are entering the park name and accepting the default lat/long suggested by iNat. I will probably construct a table of frequently visited locations and their default lat/long. I’ll use this table to flag observations whose lat/long and accuracy should be reviewed.
I suspect that the list of iNat places with accurately drawn boundaries may turn out to be rather small, but that will have to wait until I have time to go through the list and check them all.