Platform: Website
Browser, if a website issue: Chrome
When IDing this observation: https://www.inaturalist.org/observations/368903984 , I noticed that all but 1 of the CV’s suggestions were out of range; 4 of them from North/South America. I didn’t click on ‘Include suggestions not expected nearby.’ I even refreshed the tab and closed and reopened it and the suggestions were still the same. Adding an ID also didn’t change the suggestions.
If you change the place to Rizal PH - then CV offers your sp. That is using Compare.
Only 51 obs so still Pending for CV.
The CV seems to only be suggesting visually similar species, not ones expected nearby. Not sure why that might be, but it’s not necessarily a bug.
Luzon Boobook is not in the CV model yet as there aren’t enough observations. My guess would be there are no visually similar species expected nearby (the geomodel does not expect Northern Boobook at this location), so the model falls back and suggests visually similar species that are not expected nearby instead. I don’t know exactly how it’s programmed, but I’ve seen this happen before with rarely observed species and my understanding is this is expected behavior and not a bug (the alternatives would be to either suggest nothing, or to suggest expected nearby species that are much less visually similar, and I don’t think either of these would be clearly better in general).
This is common in Australia, a lot of NA, EU species get suggested when the CV has no idea
I have been experiencing this over the last few days with moths and fish in Peru. I am getting recommendations from New Zealand, Asia, North America and other locals. I’ve done nothing to change any of my settings so it seems to be a change with iNat and not me.
Has been happening to me, only getting “visually similar” and not “expected nearby” recommendations around half the time.
I thought that by default the CV only suggests species expected nearby and will only show visually similar (and not expected nearby) when ‘Include suggestions not expected nearby’ is selected, especially with the introduction of the Geomodel. (However, I can’t currently find anything that says this, so I might be misremembering.)
I’ve seen the CV only suggest 1 species. For example, moluccan scops owl isn’t in the Geomodel, and for this observation:
it suggests sulawesi scops owl instead (but I do see that the Geomodel predicts it to be there). For the northern boobook, most of the tiles to the left, right, and diagonal are predicted though, although I too am unsure exactly how it’s programmed. If this is the expected behavior though, I’d much rather the default CV suggest something nearby and/or saying it’s not sure than saying ‘We’re pretty sure this is in this genus’ (or similar) and suggest something very out of range.
This case isn’t too bad, at least one suggestion is in the proper genus. I have seen worse.
Having every species included in CV will not happen any time soon. It would really be more appropriate for CV to compare and suggest higher taxonomic levels in case there is no similar species nearby.
i think it’s working as intended, it’d rather suggest something from a different locale than nothing. it’s like that in africa a lot, since the cv is so barebones over there
and yet for Flora of Africa we achieve 59% RG ! (You know - green stuff - chuck it in the compost …
Arthropods in Africa doesn’t have a project, but 2.3 million obs, of which just over 1 million are RG. Also not bad considering the need for identifiers in many regions.
my understanding is to get the “expected nearby” designation, the tile it occurs in has to actually be in the geomodel, so the adjacent ones being in shouldn’t affect anything.
In this case would that even have helped? Like instead of suggesting a bunch of incorrect owl taxa from across the globe it would suggest a bunch of incorrect owl taxa from nearby. Is that better? If anything this is at least easier to detect.
I actually think the current functionality is sometimes useful, e.g. with arthropods I find the CV suggestions are sometimes completely unhelpful if it’s forced to filter to expected nearby (e.g. because there are no species in the genus with enough observations to be in the model at that location) but if I turn it off some congener across the globe that’s a much better match and helps me narrow things down.
Maybe an improvement would be, instead of just labeling taxa expected nearby as so, also labeling taxa that are not expected nearby with “NOT EXPECTED NEARBY”? It’s a lil peculiar to me that iNat will add a label when something is expected nearby but not when it isn’t, as the latter seems much more important to communicate. Of course I agree the best solution would be to
but I imagine this might require a substantial amount of work to implement.
You can “include suggestions not seen nearby” on the Android app (not sure about other apps). That might be the behavior you are thinking of.
https://forum.inaturalist.org/t/testing-updates-to-the-android-taxon-chooser/19742
But according to the help page, what you are describing seems like expected behavior.
By default, iNaturalist only displays suggested taxa that are visually similar and are expected nearby (if visually similar taxa are expected nearby). If no visually similar taxa are expected nearby, it displays visually similar taxa regardless of where they are expected to occur.
It’s possible to do it on the website and in both iOS apps.
