Currently if CV doesn’t find any possible id from the list of seen nearby species, it resorts to old ways of suggesting everything from anywhere. It happens very often, open a page of a moth species and you will see from one to dozens observations with cv suggestions out of nowhere, European species suggested for US, Mexico, Central America, South Africa, etc., some are even RG (not a registered oocurance, but another new user clicking on it or first to use cv).
It would be much better that if cv couldn’t find a fitting species nearby it would then only check a specific standart place (country or continent), and if there it doesn’t find anything suitable, it would then only propose higher taxonomical levels. That would reduce incorrect ids and RG while also increasing correct ids as tribe, subfamily or family is more likely to be correct in those cases.
Yep, this is one of the things that makes people’s indiscriminate use of the CV feature a real pain. The number of Northern Hemisphere spiders that seem to turn up in New Zealand is remarkable ;) I strongly suspect these IDs are made solely (yes, anecdata) based on bad CV suggestions.
To your second point, might I suggest that at some point the CV just throws its computer hands in the air, says “I have no idea” and leaves well enough alone. Most times (for me at least) no lead is better than a false lead.
It could suggest order or class or higher levels if it’s confused, I think it’s better than suggesting life (more chances to have someone visit the observation).
Yes please!
Also it would be good if certain groups could be manually opted-out of species-level CV suggestions entirely - for example, mosses, which almost always need microscopy to correctly ID. Now every single moss observation gets a suggestion of one of the few moss species observed enough to get in the CV…
If seen nearby doesn’t help, I flip to visual match. (You still have to weigh up what looks possible here) I use the map to limp up the taxonomy until I reach a broader taxon, which is seen nearby.
This has basically been implemented in that seen nearby is the filter used by default. How has it been working?
It’s working differently in situations, these screenshots show, that sometimes if it doesn’t find a fitting thing nearby it starts suggesting low-% probable species from seen nearby, in other cases it turns off seen nearby and shows high-% taxa from other places, in this case those look fine, in others they’re of different classes, so getting those two together, and making a category “seen in the region” built in probably would give better results more consistently.
