The species is included in the CV model, but the geomodel link 404s, and it shows up as an equally likely suggestion vs. Anthus rubescens on all American observations, e.g. https://www.inaturalist.org/observations/253564517.
1 Like
Is this possibly just because it’s so new? The same thing seems to happen for American Herring Gull (Larus smithsonianus), for example.
(Edit: Both of these are species that used to be subspecies, in case that helps reproduce the issue.)
A taxon swap and a taxon split have been commited a few weeks ago : https://www.inaturalist.org/taxon_changes?taxon_id=1580267
It will take a few weeks/months before the CV learns how to differentiate them.
2 Likes
Yeah, I understand the context, but this isn’t a solution, just an explanation. The fact that the geomodel is broken after such splits is a bug in my opinion.
How else do you think it could work ?
Does this happen in all browsers? Please fill out all the requested information when filling out a bug report. It isn’t arbitrary, it helps us help you.
See similar issue with barn owls here.
The CV was updated 3 days ago and these issues should be addressed now.
1 Like
It shouldn’t work. A geomodel built on a certain species concept should be invalidated if that species is split (or merged); the false confidence is a lie. There also shouldn’t be links on the public website that 404, obviously.
It’s clearly not a browser issue.
Here’s a comment from Scott (from the thread I linked above) addressing this:
The current model in production is 2.17. 2.18 was a bit delayed by the holidays but will be released any day now, it will include Siberian Pipit and some of the other new birds introduced via the2024 taxonomic updates to the eBird/Clements Checklist.
We’ve worked hard to get our model internal between data export and release to an average of 40 days. In 2022, it was closer to a year. Since it takes time to train a new model, we probably won’t be able to reduce the lag much more than that. Fortunately, these bird updates just come once a year usually in the late fall.
Edit: Sorry, I saw that you commented on that thread and have probably seen this already. My assumption is that he’s saying there isn’t any feasible way to fix these issues faster?
I’m curious what happens to the CV if a species is entirely removed from the taxonomy. I guess probably a similar suggestion with a 404 error.
1 Like
It is exactely what happens.
It’s not what happens. After A. japonicus was split off but before the model was retrained:
Are you mixing up geomodel and CV suggestion ?
The geomodel informs the CV suggestions.